Firebird Documentation Index → Firebird 3.0.7 Release Notes

1. General Notes

Thank you for choosing Firebird 3.0. We cordially invite you to test it hard against your expectations and engage with us in identifying and fixing any bugs you might encounter.

If you are upgrading from a previous major release version, be sure to study Chapter 12, Compatibility Issues before you attempt to do anything to your existing databases!

Sub-release V.3.0.7

Upgrade notice

All users of Firebird v3.0.6 are strongly encouraged to upgrade to v3.0.7 as soon as possible due to several serious bugs found in v3.0.6 and fixed in this sub-release.

Bugs reported and fixed prior to the version 3.0.7 release are listed HERE.

Caution for Windows users

Please avoid using both Firebird v3.0.6 and v3.0.7 simultaneously on the same host, it may cause troubles due to lack of shared memory synchronization between these releases.

V.3.0.7 Improvements

() — Removed PIDFile/-pidfile option from Super(Server/Classic) systemd unit.

Implemented by A. Peshkov


() — Added better diagnostic for the 'Missing security context' error.

Implemented by A. Peshkov


() — Server was modified to disconnect from the security database when missing plugin data structures cause an error

Implemented by A. Peshkov


Sub-release V.3.0.6

Bugs reported and fixed prior to the version 3.0.6 release are listed HERE.

Note for Pascal developers

VERSION constants for Firebird API interfaces inside the distributed Firebird.pas file were re-generated to match numeration rules used for other languages.

V.3.0.6 Improvements

() — Added missing relocation support to MacOS builds.

Implemented by A. Peshkov


() — Increased parsing speed of long queries.

Implemented by A. dos Santos Fernandes


() — Improved performance when using SRP plugin.

Implemented by A. Peshkov


() — Added better transaction control to ISQL.

Implemented by V. Khorsun


Sub-release V.3.0.5

Bugs reported and fixed prior to the version 3.0.5 release are listed HERE.

V.3.0.5 Improvements

() — Improved the engine providers compatibility across Firebird versions.

Implemented by A. Peshkov


() — Added a configuration switch to disable the "TCP Loopback Fast Path" option (Windows only).

Implemented by KarloX2


() — Improved the WIN_SSPI plugin to produce keys for the WireCrypt plugin.

Implemented by A. Peshkov


() — Made it possible for the AuthClient plugin to access the authentication block from DPB.

Implemented by A. Peshkov


() — Added ability to use “install.sh -path /opt/my_path” without a need to install Firebird first in the default folder.

Implemented by A. Peshkov


() — Implemented option to restore compressed .nbk files without explicitly decompressing them.

Implemented by A. Peshkov, V. Khorsun


Sub-release V.3.0.4

Bugs reported and fixed prior to the version 3.0.4 release are listed HERE.

BLOB Vulnerability

Because of the way BLOBs are implemented in Firebird, it is possible for a knowledgeable user to gain unauthorised access to their contents by a brute force method without having the necessary privileges to access the table containing them. Some work was done to ameliorate this risk in databases accessed by Firebird 3.0.4 or higher.

Security Patch for Srp Client Proof

See V. 3.0.4 SRP Security Patch in the Security chapter of these notes.

Minor ODS Change for Some Platforms

An incompatibility in the structure of the page storing generators was discovered between ODS 12.0 databases from Windows and Linux-x64 and some others, including MacOSX. The fix resulted in a minor ODS change, from 12.0 to 12.2, for some but not all platforms. This has certain implications for compatibility, so please read these notes.

V.3.0.4 Improvements

() — Context variables WIRE_COMPRESSED and WIRE_ENCRYPTED were added to the SYSTEM namespace to report compression and encryption status, respectively, of the current connection. See Context Variables: v.3.0.4 for details.

Implemented by V. Khorsun


() — Enhanced reporting of errors when a dynamic library fails to load. For more information, see Better Reporting on Dynamic Library Loading Errors.

Implemented by A. Peshkov


() — When an external function (UDF) causes an error of the type “Arithmetic exception, numeric overflow, or string truncation”, the error message will now include the name of the function.

Implemented by A. Peshkov


() — The API now supports passing the DPB/spb item ***_auth_plugin_list from an application to the client interface. More details.

Implemented by A. Peshkov


() — Two new context variables LOCALTIME and LOCALTIMESTAMP that are synonyms for CURRENT_TIME and CURRENT_TIMESTAMP, respectively. They can be used in Firebird 3.0.4 and later, for forward-compatibility with Firebird 4. See Context Variables: v.3.0.4 for details.

Implemented by A. dos Santos Fernandes


() — The read-only restriction for system tables was relaxed to permit CREATE, ALTER and DROP operations on their indexes.

Implemented by R. Abzalov, V. Khorsun


Sub-release V.3.0.3

Bugs reported and fixed prior to the version 3.0.3 release are listed HERE.

Security Alert

If you are using the database encryption feature, or plan to do so, it is essential to upgrade to this sub-release. Refer to this report for details.

V.3.0.3 Improvements

() — Engine response has been improved on cancel/shutdown signals when scanning a long list of pointer pages.

Implemented by V. Khorsun


() — The name of the encryption key is not top secret information. It can be read using the gstat utility or service, for example. However, for working with that name from a program it was desirable to access the key name using the API call Attachment::getInfo(). Hence, this facility is now provided via the information item fb_info_crypt_key.

Implemented by A. Peshkov


() — Some clauses of the ALTER DATABASE statement require updating of the single row in RDB$DATABASE: SET DEFAULT CHARACTER SET, SET LINGER, DROP LINGER. Others, such as BEGIN|END BACKUP, ENCRYPT, DECRYPT, et al., do not need to touch that record.

In previous versions, to prevent concurrent instances of ALTER DATABASE running in parallel transactions, the engine would run an update on the RDB$DATABASE record regardless of the nature of clauses specified by the user. Hence, any other transaction that read the RDB$DATABASE record in READ COMMITTED NO RECORD VERSION mode would be blocked briefly, even by a “dummy update” that in fact did not update the record.

In some cases, such as with an ALTER DATABASE END BACKUP the blockage could last 10 minutes or more. A user would seem to be unable to connect to the database with isql, for example, while ALTER DATABASE END BACKUP was running. In fact, isql would connect successfully, but it would read RDB$DATABASE immediately after attaching, using a READ COMMITTED NO RECORD VERSION WAIT transaction and then just wait until the work of ALTER DATABASE END BACKUP was committed.

From this sub-release forward, the update of the RDB$DATABASE record is avoided when possible, and an implicit lock is placed to prevent concurrent runs of the ALTER DATABASE statement.

Implemented by V. Khorsun


() — All queries that are semantically the same should have the same plan. However, until now, the optimizer understood only an explicit reference inside an ORDER BY clause and would ignore sorts derived from equivalent expressions. Now, it will consider equivalence classes for index navigation. Refer to the Tracker ticket for an example.

Implemented by D. Yemanov


() — Common Table Expressions are now allowed to be left unused.

implemented by V. Khorsun


() — Flushing a large number of dirty pages has been made faster.

Implemented by V. Khorsun


() — Measures have been taken to avoid serialization of isc_attach_database calls issued by EXECUTE STATEMENT.

Implemented by V. Khorsun


() — Output from gstat now includes the date and time of analysis.

implemented by A. Peshkov


() — The merge stage of a physical backup stage could run too long, especially with huge page cache. Changes have been made to reduce it.

Implemented by V. Khorsun


() — Message “Error during sweep: connection shutdown” now provides information about the database that was being swept.

Implemented by A. Peshkov


() — Improvement in performance of ALTER DOMAIN when the domain has many dependencies.

Implemented by V. Khorsun


() — Compression details and encryption status of the connection (fb_info_conn_flags) have been added to the getInfo() API call. For more information see notes in the API chapter.

() — Restoring a pre ODS 11.1 database now correctly populates RDB$RELATION_TYPE field in the metadata.

implemented by D. Yemanov


() — Speed of backup with nBackup when directed to NAS over SMB protocol has been improved.

Implemented by J. Hejda & V. Khorsun


() — The optimizer can now estimate the actual record compression ratio.

Implemented by D. Yemanov


Sub-release V.3.0.2

Bugs reported and fixed prior to the version 3.0.2 release are listed HERE.

One important bug fix addresses a serious security vulnerability present in all preceding Firebird releases and sub-releases.

The exploit is available to authenticated users only, somewhat limiting the risks. However, it is strongly recommended that any previous installation be upgraded to this one without delay.

New Feature

() — Support was added for fast/low-latency “TCP Loopback Fast Path” functionality introduced in Windows 8 and Server 2012.

This feature is said to improve the performance of the TCP stack for local loopback connections, by short-circuiting the TCP stack for local calls. The details of the feature can be found in .

Implemented by V. Khorsun

Improvements

The following improvements appear in this sub-release:

() — IMPROVEMENT: It is now possible to filter out info and warnings from the trace log.

implemented by V. Khorsun


() — IMPROVEMENT: Enhanced control capability when sharing the database crypt key between Superserver attachments.

implemented by A. Peshkov


() — IMPROVEMENT: The physical numbers of frequently used data pages are now cached to reduce the number of fetches of pointer pages.

implemented by V. Khorsun


() — IMPROVEMENT: A read-only transaction will no longer force write the Header/TIP page content to disk immediately after a change. This improvement gives a significant performance gain where there are numerous light read-only transactions. At this stage, it affects only servers in SS mode. For CS and SC it is more complex to implement and should appear in Firebird 4.0.

implemented by V. Khorsun


() — IMPROVEMENT: The database name was made available to an encryption plug-in.

implemented by A. Peshkov


() — IMPROVEMENT: libfbclient.so was compiled for Android (x86/x86-64/arm64).

implemented by M. A. Popa


() — IMPROVEMENT: Nesting of keys in a plug-in configuration was enabled.

implemented by V. Khorsun


() — IMPROVEMENT: The Linux code is now built with --enable-binreloc and an option was included in the installer script to install in locations other than /opt/firebird.

implemented by A. Peshkov


() — IMPROVEMENT: For Trace, a filter has been provided to INCLUDE / EXCLUDE errors by their mnemonical names.

implemented by V. Khorsun


() — IMPROVEMENT: Android port (arm32).

implemented by A. Peshkov


() — IMPROVEMENT: A port was done and tested for Linux on the ancient Motorola 680000 CPU platform to satisfy some requirement from Debian.

implemented by A. Peshkov


() — IMPROVEMENT: Support has been added to enable SELECT expressions to be valid operands for the BETWEEN predicate.

implemented by D. Yemanov


Sub-release V.3.0.1

Bugs reported and fixed prior to the version 3.0.1 release are listed HERE.

The following improvements appear in this sub-release:

() — IMPROVEMENT: The statement CREATE OR ALTER USER SYSDBA PASSWORD password can now be used to initialize an empty securityN.fdb security database.

implemented by A. Peshkov


() — IMPROVEMENT: Nesting of keys in a plug-in configuration was enabled.

implemented by A. Peshkov


() — IMPROVEMENT: For URL-like connection strings on Windows, restriction of lookup to IPv4 only was enabled.

implemented by Michal Kubecek


() — IMPROVEMENT: Line and column numbers (location context) are now provided for runtime errors raised inside EXECUTE BLOCK.

implemented by D. Yemanov


() — IMPROVEMENT: A switch was added to build POSIX binaries with a built-in libtommath library.

implemented by A. Peshkov


() — IMPROVEMENT: gbak now returns a non-zero result code when restore fails on creating and activating a deferred user index.

implemented by A. Peshkov


() — IMPROVEMENT: Implicit conversion between Boolean and string is now done automatically when a string for 'true' or 'false' is used as a value in an expression. Case-insensitive. Not valid when used with a Boolean operator — IS, NOT, AND or OR; not available for UNKNOWN.

implemented by A. dos Santos Fernandes


Bug Reporting

Bugs fixed since the release of version 3.0.0 are listed and described in the chapter entitled Bugs Fixed.

  • If you think you have discovered a new bug in this release, please make a point of reading the instructions for bug reporting in the article How to Report Bugs Effectively, at the Firebird Project website.

  • If you think a bug fix hasn’t worked, or has caused a regression, please locate the original bug report in the Tracker, reopen it if necessary, and follow the instructions below.

Follow these guidelines as you attempt to analyse your bug:

  1. Write detailed bug reports, supplying the exact build number of your Firebird kit. Also provide details of the OS platform. Include reproducible test data in your report and post it to our .

  2. You are warmly encouraged to make yourself known as a field-tester of this pre-release by subscribing to the field-testers' list and posting the best possible bug description you can.

  3. If you want to start a discussion thread about a bug or an implementation, please do so by subscribing to the firebird-devel list. In that forum you might also see feedback about any tracker ticket you post regarding this Beta.

Documentation

You will find all of the README documents referred to in these notes — as well as many others not referred to — in the doc sub-directory of your Firebird 3.0 installation.

--The Firebird Project

2. New In Firebird 3.0

The primary goals for Firebird 3 were to unify the server architecture and to improve support for SMP and multiple-core hardware platforms. Parallel objectives were to improve threading of engine processes, and the options for sharing page cache across thread and connection boundaries.

Alongside these aims came new strategies to improve performance, query optimization, monitoring and scalability, and to address the demand for more security options. A number of popular features were introduced into the SQL language, including the long-awaited support for the BOOLEAN data type and the associated logical predications.

Sub-Releases

Details of improvements and links to bug fixes in sub-releases can be found in the General Notes chapter.

Summary of Features

The following list summarises the features and changes, with links to the chapters and topics where more detailed information can be found.

Unification of the Firebird executable is complete

With the completion of true SMP support for Superserver, the Firebird core is now a unified library that supports a single ODS, loadable either as an embedded engine or by the “network listener” executable. Choice of server model is determined by settings for a new configuration parameter ServerMode, defining the locking and cache modes. It is specified at global level in firebird.conf.

By default, ServerMode = Super (alias ThreadedDedicated), i.e., SuperServer.

The previous aliases.conf is replaced by databases.conf, now including not just aliases for databases but also (optionally) configuration parameters to enable configuration of databases and/or alternative security databases individually.

The changes are described in more detail in the chapter Changes in the Firebird Engine.

True SMP support for SuperServer

In Superserver mode, the engine now makes use of multiple CPUs and cores when spawning connections.

Tracker:

Implemented by V. Khorsun

New, object-oriented C++ APIs

Object-oriented C++ APIs enable external code routines to plug in and run safely inside Firebird engine space, including (but not limited to):

  • Encryption schemes for data

  • User authentication schemes, including secure key exchange

  • Eventually, plug-in support for stored procedures, triggers and functions written in Java, C++, ObjectPascal, etc.

“Per-Database” Configuration

Custom configuration at database level can now be achieved with formal entries in databases.conf (formerly aliases.conf).

Increased Limits in Several Areas

Transaction IDs, attachment IDs, statement IDs all have increased maximum values.

Maximum number of page buffers in cache is increased for 64-bit servers.

Maximum database size is increased.

For details, see Chapter 4, Changes to the API and ODS.

Multiple Security Databases

Firebird now supports user access control via more than one security database on the server. Each database can be configured, using the parameter SecurityDatabase in databases.conf, to use a specific database other than the default security3.fdb. The user structures may even be defined within the user database itself.

The flag MON$SEC_DATABASE was added to the monitoring table MON$DATABASE to assist in determining what type of security database is used — Default, Self or Other.

Several New SQL Commands to Manage Users and Access

Changes in architecture, stiffening of rules for security and data integrity, along with feature requests, have given rise to a raft of new SQL commands for managing users and their access to objects.

International Characters in User Credentials

Provided the new authentication provisions are used, the system can accept user names and passwords containing non-ASCII characters. See International Character Sets for User Accounts in the Security chapter.

New Data Type Support
BOOLEAN

A true BOOLEAN type (True/False/Unknown), complete with support for logical predicates, e.g.,

UPDATE ATABLE
  SET MYBOOL = (COLUMN1 IS DISTINCT FROM COLUMN2)

For details, see BOOLEAN Type.

IDENTITY

IDENTITY type, spawning unique identifiers for the defined column from an internal generator. For details, see IDENTITY-Style Column.

Manage NULL/NOT NULL Column/Domain Attribute

The NOT NULL attribute on a column or domain can now be managed using the ALTER TABLE or ALTER DOMAIN syntax, respectively. For details, see Manage Nullability in Domains and Columns.

Support for SQL Packages

For details, refer to Packages.

DDL Triggers

Now, triggers can be written to execute when database objects are modified or deleted. A typical use is to block unauthorised users from performing these tasks.

For details, refer to DDL Triggers.

'Window' functions in DML

A whole new series of analytical functions to work with multiple subsets in DML. See Window (Analytical) Functions.

Statistical functions

A suite of statistical functions returning values for a variety of variance, standard deviation and linear regression formulae. See Statistical Functions.

Scrollable Cursors

The query engine now supports bi-directional (“scrollable”) cursors, enabling both forward and backward navigation in PSQL and in DSQL with support from the API. See Scrollable (Bi-directional) Cursor Support.

SQL:2008-Compliant OFFSET and FETCH Clauses

Support implemented for SQL:2008-compliant OFFSET and FETCH clauses as an alternative for {FIRST and SKIP} or {ROWS and TO} clauses. See SQL:2008-Compliant OFFSET and FETCH Clauses.

IPv6 Support

Firebird 3 can use IPv6 connections on both client and server sides. See the notes for the new configuration parameter IPv6V6Only.

Validation Whilst Database is On-line

On-line validation, first implemented in Firebird 2.5.4, has been ported forward to Firebird 3.0. See Perform Some Validation Services On-line.

Run-time Statistics in gbak Verbose Output

Verbose output from gbak can now include run-time statistics, reporting times elapsed, page reads and page writes. Also supported in the Services API.

Compatibility with Older Versions

A series of notes about compatibility with older Firebird versions is collated in Chapter 12, “Compatibility Issues”. Included there are instructions for initializing the security database, if it was not done by your installer kit, and for configuring the server for the legacy style of authentication.

3. Changes in the Firebird Engine

In Firebird 3, the remodelling of the architecture that was begun in v.2.5 was completed with the implementation of full SMP support for the Superserver model. In the new scheme, it is possible to configure the execution model individually per database.

Remodelled Architecture

Dmitry Yemanov

The remodelled architecture integrates the core engine for Classic/Superclassic, Superserver and embedded models in a common binary. The cache and lock behaviours that distinguish the execution models are now determined externally by the settings in the new configuration parameter ServerMode. The connection method is determined by the order and content of another parameter, Providers and the connection protocol that is deduced at run-time from the connection string supplied when a client requests an attachment.

The parameters for configuring the architecture are specified globally (in firebird.conf). Providers can be overridden specifically for a database (in databases.conf).

databases.conf is the old aliases.conf from previous versions, with a new name. In Firebird 3, the role of this file involves (potentially) much more than being just a lookup for database file paths. For more details about what can be configured at database level, refer to the chapter Configuration Additions and Changes.

Server Modes

Table 1. Matrix of Server Modes
ServerMode Synonym Resource Model Provider[s]

Super

ThreadedDedicated

Database is opened exclusively by a single server process. User attachments are processed by threads launched from the common pool and all share a single database page cache inside the process. This is the installation default.

Remote,
Engine121,
Loopback

Superclassic

ThreadedShared

Databases are opened by a single server process, but access is not exclusive: an embedded process can open the same database concurrently. User attachments are processed by threads launched from the common pool, each having its own database page cache.

Remote,
Engine12,
Loopback

Classic

MultiProcess

A separate process is started for each attachment to server. A database may be opened by multiple Classic processes, including local processes for embedded access. Each process has its own database page cache.

Remote,
Engine12,
Loopback

1 Only if exclusive access is available

Providers

The providers are more or less what we traditionally thought of as the methods used to connect a client to a server, that is to say, across a network, host-locally, via the local loopback (“localhost”) or by a more direct local connection (the old libfbembed.so on POSIX, now implemented as the plug-in library libEngine12.so; on Windows, engine12.dll; on MacOSX, engine12.dylib).

  • In firebird.conf, all are available by default, as follows:

    #Providers = Remote,Engine12,Loopback
  • In databases.conf, one or more providers can be blocked by pasting the line from firebird.conf, uncommenting it, and deleting the unwanted provider[s].

The Providers Architecture

Alex Peshkov

Although a key feature of Firebird 3, the Providers architecture is not new. Providers existed historically in Firebird’s predecessors and, though well hidden, are present in all previous versions of Firebird. They were introduced originally to deal with a task that has been performed since then by “interface layers” such as ODBC, ADO, BDE and the like, to enable access to different database engines using a single external interface.

Subsequently, this Providers architecture (known then as Open Systems Relational Interface, OSRI) also showed itself as very efficient for supporting a mix of old and new database formats — different major on-disk structure versions — on a single server having mixed connections to local and remote databases.

The providers implemented in Firebird 3 make it possible to support all these modes (remote connections, databases with differing ODS, foreign engines) as well as chaining providers. Chaining is a term for a situation where a provider is using a callback to the standard API when performing an operation on a database.

The Components

The main element of the Providers architecture is the y-valve. On the initial attach or create database call, y-valve scans the list of known providers and calls them one by one until one of them completes the requested operation successfully. For a connection that is already established, the appropriate provider is called at once with almost zero overhead.

Let’s take a look at some samples of y-valve operation when it selects the appropriate provider at the attach stage. These use the default configuration, which contains three providers:

  • Remote (establish network connection)

  • Engine12 (main database engine)

  • Loopback (force network connection to the local server for <database name> without an explicit network protocol being supplied).

The typical client configuration works this way: when one attaches to a database called RemoteHost:dbname (TCP/IP syntax) or \\RemoteHost\dbname (NetBios), the Remote provider detects explicit network protocol syntax and, finding it first in the Provider list, redirects the call to RemoteHost.

When <database name> does not contain a network protocol but just the database name, the Remote provider rejects it, and the Engine12 provider comes to the fore and tries to open the named database file. If it succeeds, we get an embedded connection to the database.

A special “embedded library” is no longer required. To make the embedded connection, the standard client loads the appropriate provider and becomes an embedded server.

Failure Response

But what happens if the engine returns an error on an attempt to attach to a database?

  • If the database file to be attached does not exist there is no interest at all.

  • An embedded connection may fail if the user attaching to it does not have enough rights to open the database file. That would be the normal case if the database was not created by that user in embedded mode or if he was not explicitly given OS rights for embedded access to databases on that box.

    Setting access rights in such a manner is a requirement for correct Superserver operation.

  • After a failure of Engine12 to access the database, the Loopback provider is attempted for an attach. It is not very different to Remote except that it tries to access the named database <dbname> on a server running a TCP/IP local loopback.

    On Windows, XNET is tried first, then TCP/IP loopback (with localhost: prepended to <dbname>), then Named Pipes (NetBEUI) loopback (with \\.\ prepended). The server may be started with XNET (or any other protocol) disabled, so we try all the options. On POSIX only TCP/IP protocol is supported, other options are not available

    If the attachment succeeds, a remote-like connection is established with the database even though it is located on the local machine.

Other Providers

Use of providers is not limited to the three standard ones. Firebird 3 does not support pre-ODS 12 databases. Removing support for old formats from the engine helps to simplify its code and gain a little speed. Taking into account that this speed gain sometimes takes place in performance-critical places, like searching a key in an index block, avoiding old code and related branches really does make Firebird fly faster.

Nevertheless, the Providers architecture does make it possible to access old databases when changing to a higher version of Firebird. A suitable provider may be considered for inclusion in a later sub-release.

Custom Providers

A strong feature of the Providers architecture is ability for the deployer to add their own providers to the server, the client, or both.

So what else might be wanted on a client, other than a remote connection? Recall Provider chaining that was mentioned earlier. Imagine a case where a database is accessed via a very slow network connection, say something like 3G or, worse, GPRS. What comes to mind as a way to speed it up is to cache — on the client — some big tables that rarely change. Such systems were actually implemented but, to do it, one had to rename fbclient to something arbitrary and load it into its own library called fbclient, thus making it possible to use standard tools to access the database at the same time as caching required tables. It works but, as a solution, it is clearly not ideal.

With the Providers architecture, instead of renaming libraries, one just adds a local caching provider which can use any method to detect connections to it (something like a cache@ prefix at the beginning of the database name, or whatever else you choose).

In this example, when the database name cache@RemoteHost:dbname is used, the caching provider accepts the connection and invokes the y-valve once more with the traditional database name RemoteHost:dbname. When the user later performs any call to his database, the caching provider gets control of it before Remote does and, for a locally cached table, can forestall calls to the remote server.

Use of chaining allows a lot of other useful things to be implemented. An example might be MySQL-style replication at statement level without the need for triggers: just repeat the same calls for the replication host, perhaps when a transaction is committed. In this case, the chaining provider would be installed on the server, not the client, and no modification of the command line would be needed.

That said, statement-level replication is a very questionable feature.

To avoid cycling when performing a callback to y-valve at attach time, such a provider can modify the list of providers using the isc_dpb_config parameter in the DPB. The same technique may be used at the client, too.

For details, see the Configuration Additions and Changes chapter.

The ability to access foreign database engines using providers should not be overlooked, either. It might seem strange to consider this, given the number of tools available for this sort of task. Think about the ability to access other Firebird databases using EXECUTE STATEMENT, that became available in Firebird 2.5. With a provider to ODBC or other common tool to access various data sources it is within reach to use EXECUTE STATEMENT to get direct access from procedures and triggers, to data from any database having a driver for the chosen access tool. It is even possible to have a provider to access some particular type of foreign database engine if there is some reason to want to avoid the ODBC layer.

Providers Q & A
  1. Interfaces and providers are probably very good, but I have an old task written using plain API functions and for a lot of reasons I can’t rewrite it in the near future. Does it mean I will have problems migrating to Firebird 3?

    Definitely no problems. The old API is supported for backward compatibility in Firebird 3 and will be supported in future versions as long as people need it.

  2. And what about performance when using the old API?

    The functional API is implemented as a very thin layer over interfaces. Code in most cases is trivial: convert passed handles to pointers to interfaces — hitherto referred to as “handle validation” — and invoke the appropriate function from the interface.

    Functions that execute an SQL operation and fetch data from it are one place where coding is a little more complex, involving the SQLDA construct. The data moves related to the SQLDA have always created an overhead. The logic between the new and old APIs does not add significantly to that old overhead.

Connection String and Protocols

Connection string refers to the local or remote path to the database to which a client requests an attachment (connection). The syntax of the connection string determines the transport protocol by which clients and the server communicate. The legacy syntaxes for the available protocols, supported by all Firebird versions, are as follows:

For TCP/IP (aka INET) protocol

<host> [ / <port>] : <database file path or alias>
For named pipes (aka NetBEUI, aka WNET) protocol

\\ <host> [ @ <port>] \ <database file path or alias>
For local connections, simply
<database file path or alias>....

Local connection is implied if <host> is omitted. Depending on settings, platform and Firebird version, it could be performed via either the embedded engine, XNET (shared memory) protocol or TCP/IP localhost loopback.

Examples
Connect via TCP/IP using database name

192.168.0.11:/db/mydb.fdb
192.168.0.11:C:\db\mydb.fdb
myserver:C:\db\mydb.fdb
localhost:/db/mydb.fdb
Connect via TCP/IP using database alias

192.168.0.11:mydb
myserver:mydb
localhost:mydb
Connect via TCP/IP using non-default port 3051

192.168.0.11/3051:C:\db\mydb.fdb
192.168.0.11/3051:mydb
myserver/3051:/db/mydb.fdb
localhost/3051:/db/mydb.fdb
myserver/3051:mydb
localhost/3051:mydb
Connect via TCP/IP using non-default service name

192.168.0.11/fb_db:C:\db\mydb.fdb
192.168.0.11/fb_db:mydb
localhost/fb_db:/db/mydb.fdb
myserver/fb_db:/db/mydb.fdb
myserver/fb_db:mydb
localhost/fb_db:mydb
Connect via named pipes (Windows only)

\\myserver\C:\db\mydb.fdb
\\myserver@fb_db\C:\db\mydb.fdb
Local connection

/db/mydb.fdb
C:\db\mydb.fdb
mydb
URL-Style Connection Strings

Firebird 3.0 introduces an additional, generalized, URL-style syntax for connection strings. The pattern is:

[ <protocol> : // [ <host> [ : <port> ] ] ] / <database file path or alias>

<protocol> ::= INET | WNET | XNET

INET resolves to TCP/IP, WNET to Named Pipes, while XNET surfaces the old “Windows local protocol” (shared memory).

Examples
Connect via TCP/IP using database name

inet://192.168.0.11//db/mydb.fdb
inet://192.168.0.11/C:\db\mydb.fdb
inet://myserver/C:\db\mydb.fdb
inet://localhost//db/mydb.fdb
Connect via TCP/IP using database alias

inet://192.168.0.11/mydb
inet://myserver/mydb
inet://localhost/mydb
Connect via TCP/IP using non-default port 3051

inet://192.168.0.11:3051/C:\db\mydb.fdb
inet://192.168.0.11:3051/mydb
inet://myserver:3051//db/mydb.fdb
inet://localhost:3051//db/mydb.fdb
inet://myserver:3051/mydb
inet://localhost:3051/mydb
Connect via TCP/IP using non-default service name

inet://192.168.0.11:fb_db/C:\db\mydb.fdb
inet://192.168.0.11:fb_db/mydb
inet://localhost:fb_db//db/mydb.fdb
inet://myserver:fb_db//db/mydb.fdb
inet://myserver:fb_db/mydb
inet://localhost:fb_db/mydb
Connect via named pipes

wnet://myserver/C:\db\mydb.fdb
wnet://myserver:fb_db/C:\db\mydb.fdb
Loopback connection via TCP/IP

inet:///db/mydb.fdb
inet://C:\db\mydb.fdb
inet://mydb
Loopback connection via named pipes

wnet://C:\db\mydb.fdb
wnet://mydb
Local connection via shared memory

xnet://C:\db\mydb.fdb
xnet://mydb
Local (embedded by default) connection

/db/mydb.fdb
C:\db\mydb.fdb
mydb

Local connection is implied if <host> is omitted. Depending on settings, platform and Firebird version, it could be performed via either the embedded engine, XNET (shared memory) protocol or TCP/IP localhost loopback.

On the server side, the provider configuration is in the default order Remote, Engine12, Loopback. If the Remote provider fails to match the connection string because the protocol or host parts are missing, then Engine12, the embedded engine, handles it as a hostless connection. To connect locally using a specific transport protocol, it is necessary to specify that protocol:

inet://<database file path or alias>
or
wnet://<database file path or alias>
or
xnet://<database file path or alias>

WNET (named pipes) and XNET (shared memory) protocols are available only on Windows.

Plug-Ins

Alex Peshkov

From version 3 onward, Firebird’s architecture supports plug-ins. For a number of predefined points in the Firebird code, a developer can write his own fragment of code for execution when needed.

A plug-in is not necessarily one written by a third party: Firebird has a number of intrinsic plug-ins. Even some core parts of Firebird are implemented as plug-ins.

What is a Plug-In?

The term “plug-in” is used to name related but different things:

  • a dynamic library, containing code to be loaded as a plug-in (often called a plug-in module) and stored in the $FIREBIRD/plugins directory;

  • code implementing a plug-in. That is slightly different from the library, since a single dynamic library may contain code for more than one plug-in;

  • a plug-in’s factory: an object created by that code (pure virtual C++ class), creating instances of the plug-in at Firebird’s request;

  • an instance of the plug-in, created by its factory.

Plug-In Types

Firebird’s plug-in architecture makes it possible to create plug-ins of predefined types. Each version of Firebird will have a fixed set of supported plug-in types. To add a further type, the first requirement is to modify the Firebird code. Our plug-in architecture facilitates both adding new types of plug-ins and simplifying the coding of the plug-in along generic lines.

To be able to implement a plug-in, say, for encrypting a database on the disk, the Firebird code has to be prepared for it: it must have a point from which the plug-in is called.

The set of plug-in types implemented in Firebird 3 comprises:

user authentication related
  • AuthServer (validates user’s credentials on server when logins are used)

  • AuthClient (prepares credentials to be passed over the wire)

  • AuthUserManagement (maintains a list of users on a server in a format known to AuthServer)

ExternalEngine

Controls the use of various engines, see External Engines.

Trace

The Trace plug-in was introduced in Firebird 2.5, but the way it interacts with the engine was changed in Firebird 3 to accord with the new generic rules.

Encryption

encrypting plug-ins are for

  • network (WireCrypt)

  • disk (DbCrypt)

  • a helper plug-in (KeyHolder), used to help maintain the secret key(s) for DbCrypt

Provider

Firebird 3 supports providers as a plug-in type.

Technical Details

Plug-ins use a set of special Firebird interfaces. All plug-in-specific interfaces are reference counted, thus putting their lifetime under specific control. Interfaces are declared in the include file plug-in.h. A simple example for writing a plug-in module can be found in DbCrypt_example.

The example does not perform any actual encryption, it is just a sample of how to write a plug-in. Complete instructions for writing plug-ins are not in scope for this document.

Features of a Plug-In

A short list of plug-in features:

  • You can write a plug-in in any language that supports pure virtual interfaces. Interface declarations will need to be written for your language if they are missing.

  • As with UDFs, you are free to add any reasonable code to your plug-in — with emphasis on reasonable. For example, prompting for user input at the server’s console from a plug-in is hardly “reasonable”!

  • Calling the Firebird API from your plug-in is OK, if needed. For example, the default authentication server and user manager use a Firebird database to store accounts.

  • Firebird provides a set of interfaces to help with configuring your plug-ins. It is not obligatory to use them, since the plug-in code is generic and can employ any useful method for capturing configuration information. However, using the standard tools provides commonality with the established configuration style and should save the additional effort of rolling your own and documenting it separately.

Configuring Plug-ins

Configuration of plug-ins has two parts:

  1. The engine has to be instructed what plug-ins it should load

  2. The plug-ins themselves sometimes need some configuration.

The plug-ins to be loaded for each plug-in type are defined in the main configuration file, firebird.conf, usually with defaults. The ones defined in Firebird 3 are discussed in the chapter entitled “Configuration Additions and Changes”. In summary, the set that provides normal operation in the server, client and embedded cases consists of:

  • AuthServer = Srp, Win_Sspi

  • AuthClient = Srp, Win_Sspi, Legacy_Auth

  • UserManager = Srp

  • TracePlugin = fbtrace

  • Providers = Remote,Engine12,Loopback

  • WireCryptPlugin = Arc4

If you want to add other plug-ins, they must be cited in firebird.conf. Apart from other considerations, this requirement acts as a security measure to avoid loading unknown code.

Taking the entry TracePlugin = fbtrace as an example, what does the value fbtrace signify? In a trivial case, it can indicate the name of a dynamic library, but the precise answer is more complicated.

As mentioned earlier, a single plug-in module may implement more than one plug-in. In addition, a single plug-in may have more than one configuration at once, with a separate plug-in factory created for each configuration. Each of these three object contexts (module | implementation | factory) has its own name:

  • The name of a module is the file name of a dynamic library

  • The name of a plug-in implementation is the one given to it by the developer of the plug-in. It is hard-coded inside the module.

  • The name of a factory is, by default, the same as the name of the plug-in implementation’s name. It is the factory name which is actually used in firebird.conf.

In a typical trivial case, a module contains one plug-in that works with just one configuration and all three names are equal, and no more configuration is needed. An example would be libEngine12.so / Engine12.dll / Engine12.dylib, that contains the implementation of the embedded provider Engine12. Nothing other than the record Providers = Engine12 is needed to load it.

For something more complex a file will help you to set up the plug-in factories precisely.

plugins.conf

The file $(root)/plugins.conf has two types of records: config and plugin.

the plugin record is a set of rules for loading and activating the plug-in. Its format is:

Plugin = PlugName ## this is the name to be referenced in firebird.conf
  {
    Module = LibName ## name of dynamic library
    RegisterName = RegName ## name given to plug-in by its developer
    Config = ConfName ## name of config record to be used
    ConfigFile = ConfFile ## name of a file that contains plug-in's configuration
  }

When plug-in PlugName is needed, Firebird loads the library LibName and locates the plug-in registered with the name RegName. The configuration values from the config record ConfName or the config file ConfFile are passed to the library.

If both ConfName and ConfFile are given, then the config record will be used.

If both parameters are missing, the default PlugName is used; except that if the ConfigFile is present and its name is the same as the module’s dynamic library but with a .conf extension, it will be used.

The ConfigFile is expected to use the format Key=Value, in line with other Firebird configuration files.

For the plug-in configuration record the same format is used:

Config = ConfName
  {
    Key1 = Value1
    Key2 = Value2
    ...
  }
A Sample Setup

Suppose you have a server for which some clients trust the wire encryption from one vendor and others prefer a different one. They have different licences for the appropriate client components but both vendors use the name “BestCrypt” for their products.

The situation would require renaming the libraries to, say, WC1 and WC2, since there cannot be two files in the same directory with the same name. Now, the modules stop loading automatically because neither is called “BestCrypt” any longer.

To fix the problem, plug-ins.conf should contain something like this:

Plugin = WC1
  {
    RegisterName = BestCrypt
  }
Plugin = WC2
  {
    RegisterName = BestCrypt
  }

The module names will be automatically set to WC1 and WC2 and found. You can add any configuration info that the plug-ins need.

Remember to modify firebird.conf to enable both plug-ins for the WireCryptPlugin parameter:

WireCryptPlugin = WC1, WC2

The server will now select appropriate plug-in automatically to talk to the client.

Another sample is distributed with Firebird, in $(root)/plugins.conf, configuring one of the standard plug-ins, UDR. Because it was written to a use non-default configuration, the module name and one configuration parameter are supplied explicitly.

Plug-Ins Q & A
  1. There are plug-ins named Remote, Loopback, Arc4 in the default configuration, but no libraries with such names. How do they work?

    They are “built-in” plug-ins, built into the fbclient library, and thus always present. Their existence is due to the old ability to distribute the Firebird client for Windows as a single dll. The feature is retained for cases where the standard set of plug-ins is used.

  2. What do the names of Srp and Arc4 plug-ins mean?

    Srp implements the Secure Remote Passwords protocol, the default way of authenticating users in Firebird 3. Its effective password length is 20 bytes, resistant to most attacks (including “man in the middle”) and works without requiring any key exchange between client and server to work.

    Arc4 means Alleged RC4 — an implementation of RC4 cypher. Its advantage is that it can generate a unique, cryptographically strong key on both client and server that is impossible to guess by capturing data transferred over the wire during password validation by SRP.

    The key is used by Arc4 after the SRP handshake, which makes wire encryption secure without need to exchange any keys between client and server explicitly.

  3. What do Win_Sspi and Legacy_Auth mean?

    Windows SSPI has been in use since Firebird 2.1 for Windows trusted user authentication. Legacy_Auth is a compatibility plug-in to enable connection by the Firebird 3 client to older servers. It is enabled by default in the client.

    And yes, it still transfers almost plain passwords over the wire, for compatibility.

    On the server it works with security3.fdb just as with a security database from Firebird 2.5. It should be avoided except in situations where you understand well what you are sacrificing.

    To use Legacy_Auth on the server you will need to avert network traffic encryption in firebird.conf by reducing the default Required setting for the WireCrypt parameter, either

    WireCrypt = Enabled

    or

    WireCrypt = Disabled
  4. How can I find out what the standard Authentication and User Manager plug-ins are?

    They are listed in firebird.conf.

External Engines

Adriano dos Santos Fernandes

The UDR (User Defined Routines) engine adds a layer on top of the FirebirdExternal engine interface with the purpose of

  • establishing a way to hook external modules into the server and make them available for use

  • creating an API so that external modules can register their available routines

  • making instances of routines “per attachment”, rather than dependent on the internal implementation details of the engine

External Names

An external name for the UDR engine is defined as

'<module name>!<routine name>!<misc info>'

The <module name> is used to locate the library, <routine name> is used to locate the routine registered by the given module, and <misc info> is an optional user-defined string that can be passed to the routine to be read by the user.

Module Availability

Modules available to the UDR engine should be in a directory listed by way of the path attribute of the corresponding plugin_config tag. By default, a UDR module should be on <fbroot>/plugins/udr, in accordance with its path attribute in <fbroot>/plugins/udr_engine.conf.

The user library should include FirebirdUdr.h (or FirebirdUdrCpp.h) and link with the udr_engine library. Routines are easily defined and registered, using some macros, but nothing prevents you from doing things manually.

A sample routine library is implemented in examples/udr, showing how to write functions, selectable procedures and triggers. It also shows how to interact with the current attachment through the legacy API.

Scope

The state of a UDR routine (i.e., its member variables) is shared among multiple invocations of the same routine until it is unloaded from the metadata cache. However, it should be noted that the instances are isolated “per session”.

Character Set

By default, UDR routines use the character set that was specified by the client.

In future, routines will be able to modify the character set by overriding the getCharSet method. The chosen character set will be valid for communication with the old Firebird client library as well as the communications passed through the FirebirdExternal API.

Enabling UDRs in the Database

Enabling an external routine in the database involves a DDL command to “create” it. Of course, it was already created externally and (we hope) well tested.

Syntax
{ CREATE [ OR ALTER ] | RECREATE | ALTER } PROCEDURE <name>
  [ ( <parameter list> ) ]
  [ RETURNS ( <parameter list> ) ]
  EXTERNAL NAME '<external name>' ENGINE <engine>

{ CREATE [ OR ALTER ] | RECREATE | ALTER } FUNCTION <name>
  [ <parameter list> ]
  RETURNS <data type>
  EXTERNAL NAME '<external name>' ENGINE <engine>

{ CREATE [ OR ALTER ] | RECREATE | ALTER } TRIGGER <name>
  ...
  EXTERNAL NAME '<external name>' ENGINE <engine>
Examples
create procedure gen_rows (
  start_n integer not null,
  end_n integer not null
) returns (
  n integer not null
) external name 'udrcpp_example!gen_rows'
  engine udr;

create function wait_event (
  event_name varchar(31) character set ascii
) returns integer
  external name 'udrcpp_example!wait_event'
  engine udr;

create trigger persons_replicate
  after insert on persons
  external name 'udrcpp_example!replicate!ds1'
  engine udr;
How it Works

The external names are opaque strings to Firebird. They are recognized by specific external engines. External engines are declared in configuration files, possibly in the same file as a plug-in, as in the sample UDR library that is implemented in $(root)/plugins.

external_engine = UDR {
  plugin_module = UDR_engine
}

plugin_module = UDR_engine {
  filename = $(this)/udr_engine
  plugin_config = UDR_config
}

plugin_config = UDR_config {
  path = $(this)/udr
}

When Firebird wants to load an external routine (function, procedure or trigger) into its metadata cache, it gets the external engine through the plug-in external engine factory and asks it for the routine. The plug-in used is the one referenced by the attribute plugin_module of the external engine.

Depending on the server architecture (Superserver, Classic, etc) and implementation details, Firebird may get external engine instances “per database” or “per connection”. Currently, it always gets instances “per database”.

Optimizer Improvements

Dmitry Yemanov

  • See Tracker item .

    Hash/merge joins for non-field (DBKEY or derived expression) equalities are now allowed.

  • See Tracker item .

    The optimizer now considers the ORDER BY optimization when making its decision about join order.

Other Optimizations

Vlad Khorsun

  • See Tracker item .

    Data pages are now allocated as a group of sequential ordered pages (extents).

  • See Tracker item .

    The main database file extends faster when physical backup state changes from stalled to merge.

  • See Tracker item .

    Linux systems that support “fast file growth” can now use it.

  • See Tracker item .

    Attachments no longer block others when the allocation table is read for the first time.

  • See Tracker item .

    Contention has been reduced for the allocation table lock while database is in stalled physical backup state.

Remote Interface/Network Protocol

Dmitry Yemanov

Tracker item .

Further improvements were made to Firebird’s network protocol, providing a denser data stream and better prefetch logic. The following improvements were implemented:

  1. The full length of a field whose value is NULL is no longer sent over the wire. (Tracker item ). NULL flags (4 bytes per field) are replaced with a bitmap and only these flags are transmitted, in the bitmap.

    This improvement is available for the DSQL API only, so gbak does not benefit from this improvement, as it uses a lower level BLR API.

  2. The prefetch (batch receive) algorithm is now aware of variable-length messages, so that VARCHARs and NULLs may reduce the transmitted message size, allowing more rows to be transmitted in each batch.

Acknowledgement

This work was sponsored by donations collected at the 9th Firebird Developers' Day conference in Brazil.

Miscellaneous Improvements

Miscellaneous engine improvements include:

Connections Limit Raised

Paul Beach

() — Maximum connections (FD_SETSIZE) on Windows Superserver and Superclassic was raised from 1024 to 2048.

Better Error Diagnosis

Dmitry Yemanov

() — The error reported for index/constraint violations has been extended to include the problematic key value.

ICU Version Upgraded

Adriano dos Santos Fernandes

() — The ICU version was upgraded to v.52.1.

Internal Debug Info Made Human-readable

Vlad Khorsun

A new BLOB filter translates internal debug information into text.

A Silly Message is Replaced

Claudio Valderrama C.

A silly message sent by the parser when a reference to an undefined object was encountered was replaced with one that tells it like it really is.

New Pseudocolumn RDB$RECORD_VERSION

Adriano dos Santos Fernandes

A pseudocolumn named RDB$RECORD_VERSION returns the number of the transaction that created the current record version.

It is retrieved the same way as RDB$DB_KEY, i.e., select RDB$RECORD_VERSION from aTable where…​

systemd init Scripts

Alex Peshkov

systemd init scripts are available in Firebird 3 POSIX installers. See Tracker ticket .

Firebird 3.0.4: Better Reporting on Dynamic Library Loading Errors

Alex Peshkov

A mistake in the initial design of the ModuleLoader class meant that errors displayed when dynamic libraries failed to load were lacking any OS-specific information about the reason for the failure (no such file, invalid format, unresolved external reference, etc.). The only report was “Module/library not loaded”. In many cases, that made it extremely hard to find and fix the related bugs.

Firebird 3.0.4 extends the reported errors to show the exact reason a library failed to load.

Exceptions

Because the use of external functions (UDFs) has been deprecated, those modules do not get this extended error treatment. The other exception is the ICU modules, due to the fact more than 50 retries are done when those libraries fail to load. Reporting such an avalanche of error detail was considered impracticable.

4. Changes to the Firebird API and ODS

ODS (On-Disk Structure) Changes

New ODS Number

Firebird 3.0 creates databases with an ODS (On-Disk Structure) version of 12. In the initial release, a database with an older ODS cannot be opened by Firebird 3.0. In order to work with a database with an older ODS it will be necessary to make a backup using gbak under the older server and restore it with gbak on Firebird 3.

A legacy provider for databases with ODS 8 to 11.2 is planned for a future sub-release.

Firebird 3.0.4: ODS 12.2 for POSIX Platforms

In the version 3.0.4 sub-release we have a minor ODS change for databases created or restored on a Linux x86 (32-bit) platform and all other POSIX platforms including MacOSX. The new ODS for these databases is 12.2.

The change is due to a change in the page layout of the system table RDB$GENERATORS affecting only some platforms. ODS 12.0 databases copied across platforms, e.g., from Linux-x86 to Linux-x64, show wrong values for generators, typically very large.

Other Platforms

The most popular platforms — Windows and Linux x64 (64-bit) — were not affected by the change. They were carefully checked to verify that, for them, there is no actual difference between ODS 12.0 and 12.2. To avoid disturbing users of these platforms with a new ODS, it was decided to retain the ODS number 12.0 for them.

For non-Windows and non-Intel-Linux platforms, developers could not perform all the required checks. The only guaranteed safe solution for them was the minor ODS upgrade.

To restate, MacOSX (32-bit and 64-bit) and 32-bit Linux builds will create (and restore) databases as ODS 12.2.

Existing databases (with ODS 12.0) are always opened successfully with Firebird 3.0.4 on the platform where they were initially created. Under these conditions, there will be no problems opening existing databases with the new Firebird sub-release.

Cross-Platform File-Copying

Cross-platform copying of any ODS 12.0 database other than those created on Windows and Linux-x64 is best avoided: Firebird versions < 3.0.3 may produce errors with generators, while version 3.0.4 will report an issue and not attach to such a database.

Our checks did not reveal any cross-platform compatibility issues between ODS 12.2 databases and ODS 12.0 databases from Windows and Linux-x64 with matching Endianness.

Implementation ID is Deprecated

Alex Peshkov

The Implementation ID in the ODS of a database is deprecated in favour of a new field in database headers describing hardware details that need to match in order for the database to be assumed to have been created by a compatible implementation.

The old Implementation ID is replaced with a 4-byte structure consisting of hardware ID, operating system ID, compiler ID and compatibility flags. The three ID fields are just for information: the ODS does not depend upon them directly, and they are not checked when opening the database.

The compatibility flags are checked for a match between the database and the engine opening it. Currently, we have only one flag, for endianness. As previously, Firebird will not open a database on little-endian that was created on big-endian, nor vice versa.

Sample gstat Output
# ./gstat -h employee
Database /usr/home/firebird/trunk/gen/Debug/firebird/examples/empbuild/employee.fdb
Database header page information:
        ..............
        Implementation          HW=AMD/Intel/x64 little-endian OS=Linux CC=gcc
        ..............

The purpose is to make it easier to do ports of Firebird for new platforms.

Maximum Database Size

Maximum database size is increased to 232 pages (previously 231 -1 pages). The new limit is 16TB/32TB/64TB, depending on the page size.

Maximum Page Size

The maximum page size remains 16 KB (16384 bytes).

Maximum Number of Page Buffers in Cache

The maximum number of pages that can be configured for the database cache depends on whether the database is running under 64-bit or 32-bit Firebird:

  • 64-bit : 231 -1 (2,147,483,647) pages

  • 32-bit : 128,000 pages, i.e. unchanged from version 2.5

Extension of Transaction ID Space Limit

Dmitry Yemanov

Historically, transaction ID space was limited to 231 -1 transactions, counted from the time the database was created. After that point, the database becomes unavailable until backup and restore is performed to reset the transaction ID counter back to zero. Initially in Firebird 3.0, the transaction ID space was raised to 232 transactions, doubling the database up-time without backup and restore.

This improvement request is about shifting this limit even further, with the introduction of 48-bit internal transaction IDs that are publicly (via the API and the MON$ tables) represented as 64-bit numbers. This makes the new limit roughly equal to 2.8 * 1014 transactions. Later, it could be extended up to the 263 -1 limit.

The implemented solution has no additional storage overhead until the transaction counters grow beyond the 232 boundary.

Limits Raised for Attachment and Statement IDs

Attachment IDs and statement IDs were changed to 64-bit numbers, both internally and externally via the API and the MON$ tables.

System Tables

New System Tables
RDB$AUTH_MAPPING

Stores authentication and other security mappings

RDB$PACKAGES

Header for SQL packages

RDB$DB_CREATORS

A list of users granted the “CREATE DATABASE” privilege when using the specified security database

SEC$USERS

Virtual table to query the local user list

SEC$USER_ATTRIBUTES

Virtual table storing local user attributes

SEC$DB_CREATORS

SQL interface to access the list in RDB$CREATORS, i.e. select * from SEC$DB_CREATORS

SEC$GLOBAL_AUTH_MAPPING

SQL interface to access the members of RDB$AUTH_MAPPING that have access to all databases using the specified security database, i.e. select * from SEC$GLOBAL_AUTH_MAPPING.

For information about authentication mapping, see Mapping of Users to Objects in the Security chapter.

Changes to System Tables

From Firebird 3 forward, all non-virtual system tables (RDB$*) are read-only.

In version 3.0.4, the read-only restriction was relaxed to permit CREATE, ALTER and DROP operations on the indexes of system tables.

RDB$SYSTEM_FLAG

Claudio Valderrama C.

RDB$SYSTEM_FLAG has been made NOT NULL in all tables.

.

RDB$TYPES

Dmitry Yemanov

Missing entries were added to RDB$TYPES. They describe the numeric values for these columns:

RDB$PARAMETER_TYPE     (table RDB$PROCEDURE_PARAMETERS)
RDB$INDEX_INACTIVE     (table RDB$INDICES)
RDB$UNIQUE_FLAG        (table RDB$INDICES)
RDB$TRIGGER_INACTIVE   (table RDB$TRIGGERS)
RDB$GRANT_OPTION       (table RDB$USER_PRIVILEGES)
RDB$PAGE_TYPE          (table RDB$PAGES)
RDB$PRIVATE_FLAG       (tables RDB$PROCEDURES and RDB$FUNCTIONS)
RDB$LEGACY_FLAG        (table RDB$FUNCTIONS)
RDB$DETERMINISTIC_FLAG (table RDB$FUNCTIONS)
Monitoring Tables

Dmitry Yemanov

Changes to Client Address Reporting

Prior to Firebird 3.0, the network address of remote clients were reported in MON$ATTACHMENTS.MON$REMOTE_ADDRESS and RDB$GET_CONTEXT('SYSTEM', 'CLIENT_ADDRESS'). For TCP/IP protocol (a.k.a. INET), it contained a TCPv4 dot-separated address. For Named Pipes (a.k.a. WNET, NetBeui) protocol, it was always NULL. For shared memory (aka XNET) protocol, it contained the local host name.

Starting with Firebird 3.0, the network address of a remote client contains the TCP/IP port number of the remote client, separated with a slash:

<IP address>/<port>

The port number is also retrieved via the new built-in context variable RDB$GET_CONTEXT('SYSTEM', 'CLIENT_PORT').

The host name is also reported now, in the new column MON$REMOTE_HOST.

Alert

The WNET (Named Pipes/Netbeui) protocol should be considered as deprecated. It is likely to be abandoned in a future version.

Per-table performance counters added to the monitoring tables

Per-table performance counters have been added to all of the monitoring tables. See Tracker .

Monitoring table changes MON$ATTACHMENTS

New information is now available:

  • Operating system user name. See Tracker .

  • Protocol and client library version. See Tracker .

  • Client host name. See Tracker .

  • Authentication method used for connection (MON$AUTH_METHOD). See Tracker .

  • MON$REMOTE_ADDRESS now contains the <IP>/<port> string. See Tracker .

MON$DATABASE
  • Database owner (MON$OWNER) added. See Tracker .

  • Security database type (MON$SEC_DATABASE) flag added. Value will be one of Default/Self/Other. See Tracker .

MON$STATEMENTS

The PLAN is now included. See Tracker .

Application Programming Interfaces

A new public API replaces the legacy one in new applications, especially object-oriented ones. The interface part can be found in the header file Interfaces.h in the directory /include/firebird beneath the installation root directory.

POSIX installations have a symlink pointing to /usr/include/firebird/Interfaces.h

The new public API can be also used inside user-defined routines (UDR, q.v.) for callbacks inside the engine, allowing a UDR to select or modify something in the database, for example.

The main difference between the new API and the legacy one is that UDRs can query and modify data in the same connection or transaction context as the user query that called that UDR. It is now possible to write external triggers and procedures, not just external functions (UDFs).

Interfaces and the New Object-oriented API

Alex Peshkov

Firebird needed a modernised API for a number of compelling reasons.

  • High on the list was the limitation of the 16-bit integer pervading the legacy API, encompassing message size, SQL operator length, BLOB data portions, to name a few examples. While 16-bit was probably adequate when that old API came to life, in today’s environments it is costly to work around.

    A trivial solution might be to add new functions that support 32-bit variables. The big downside is the obvious need to retain support for the old API by having pairs of functions with the same functionality but differing integer sizes. In fact, we did something like this to support 64-bit performance counters, for no better reason than being pressed to provide for it without having a more elegant way to implement it.

  • Another important reason, less obvious, derives from the era when Firebird’s predecessor, InterBase, did not support SQL. It used a non-standard query language, GDML, to manage databases. Data requests were transported between client and server using messages whose formats were defined at request compilation time in BLR (binary language representation). In SQL, the operator does not contain the description of the message format, so the decision was taken to surround each message with a short BLR sequence describing its format.

    The ISC API also has the XSQLDA layer over BLR. The trap with the XSQLDA solution is that it encapsulates both the location of the data and their format, making it possible to change location or format (or both) between fetch calls. Hence, the need for the BLR wrapping in every fetch call — notwithstanding, this potential capability to change the data format between fetches was broken in the network layer before Firebird existed.

    To support the XSQLDA layer that rides on top of the message-based API, that lower level API also has support sending format BLR at every turn.

    This system involving calls processing data through multiple layers is hard to extend and wastes performance; the SQLDA is not simple to use; the desire to fix it was strong.

  • Other reasons — numerous but perhaps less demanding — for changing the API included enhancing the status vector and optimizing dynamic library loading. Interfaces also make it so much easier and more comfortable to use the messages API.

The Non-COM Choice

The new interfaces are not compatible with COM, deliberately, and the reasons have to do with future performance enhancement.

At the centre of the Providers architecture in Firebird 3.0 is the y-valve, which is directed at dispatching API calls to the correct provider. Amongst the potential providers are older ones with potentially older interfaces. If we used COM, we would have to call the method IUnknown for each call (including record fetch), just to ensure that the provider really had some newer API method. Along with that comes the likelihood of future additions to the catalogue of API calls to optimize performance. A COM-based solution does not play well with that.

Firebird interfaces, unlike COM, support multiple versions. The interface version is determined by the total number of virtual functions it encompasses and is stored as a pointer-size integer at the beginning of the virtual functions table. This makes it possible for very fast checking of the interface version, since it requires no virtual call. That is to say, the pointer check has no overhead, unlike COM.

The Hierarchy of Interfaces

A detailed discussion of all the functions presented by all the interfaces is outside the scope of this overview. The general schematic looks like this:

interfaces hierarchy

The base of the structure is IVersioned. It is the interface that enables a version upgrade. A lot of interfaces not requiring additional lifetime control are based directly on IVersioned. IMaster is one example already mentioned. Others include a number of callback interfaces whose lifetimes must match the lifetimes of the objects from which they were to be used for callback.

Two interfaces deal with lifetime control: IDisposable and IRefCounted. The latter is especially active in the creation of other interfaces: IPlugin is reference counted, as are many other interfaces that are used by plug-ins. These include the interfaces that describe database attachment, transaction management and SQL statements.

Not everything needs the extra overhead of a reference-counted interface. For example, IMaster, the main interface that calls functions available to the rest of the API, has unlimited lifetime by definition. For others, the API is defined strictly by the lifetime of a parent interface; the IStatus interface is non-threaded. For interfaces with limited lifetimes, it is of benefit to have a simple way to destroy them, that is, a dispose() function.

Each plug-in has one and only one main interface — IPlugin — which is responsible for basic plug-in functionality. In fact, a lot of plugins have only this interface, although that is not a requirement.

Finally, there is IProvider, a kind of “main” plug-in in the Firebird API. IProvider is derived from IPlugin and must be implemented by every provider. If you want to write your own provider you must implement IProvider. It is implemented also by the y-valve: it is the y-valve implementation that is returned to the user when the getDispatcher() function from the master interface is called.

IProvider contains functions enabling creation of an attachment to a database (attach and create) or to the Services Manager.

Interfaces Q & A
  1. We access new API using IMaster, but how to get access to IMaster itself?

    This is done using just the one new API function fb_get_master_interface(). It is exported by the fbclient library. Also, IMaster is passed as a parameter to each plug-in during its registration in the system.

  2. The non-use of COM-based interfaces was said to be to avoid working with IUnknown methods and that this is done due to performance issues, instead you have to check the interface version. Why is that faster than using IUnknown?

    As was already mentioned, we do not need to execute virtual calls when checking the interface version. Taking into an account that each virtual call means a reset of the CPU cache, it is an important difference, especially for the very small calls like getting specific metadata properties from IMetadata.

Other New APIs

Other new APIs support various plug-ins by declaring the interfaces between the engine and the plug-in. Besides pluggable authentication and pluggable encryption, Firebird 3 supports “external engines”, bridges between the engine and the execution environments that can run UDRs: native code, Java and others. By and large they are intended for use by third-party solution providers, rather than for client application development.

For creating custom plug-ins and bridges, the relevant interface (API) needs to be implemented in the plug-in code.

API Improvements

The following improvements to the API should be noted.

Some SQL Size Limits Removed Using New API

Dmitry Yemanov

If and only if the new API is being used:

  • The size of the body of a stored procedure or a trigger can exceed the traditional limit of 32 KB. The theoretical limit provided by the new API is 4GB. At the moment — as a security measure — a hard-coded limit of 10MB is imposed. The same limit of 10MB also applies to any user-defined DSQL query.

  • The total size of all input or output parameters for a stored procedure or a user-defined DSQL query is no longer limited to the traditional size of (64KB minus overhead). The theoretical limit provided by the new API is 4GB.

Legacy API

Improvements to the legacy API include:

Scrollable Cursor Support

Dmitry Yemanov

In PSQL, a scrollable cursor can be operated on directly to navigate flexibly from the current row to any another row either forwards or backwards. API support is available to make scrollable cursors available to DSQL applications.

Scrollable Cursor Usage

The result set must be opened with the flag IStatement::CURSOR_TYPE_SCROLLABLE explicitly specified.

Fetch Methods

The following fetch methods of the IResultSet interface are available:

int fetchNext(IStatus* status, void* message);
// equivalent to FETCH NEXT FROM <cursor name>

Moves the cursor’s current position to the next row and returns it. If the cursor is empty or already positioned at the last row, the condition NO_DATA is returned.

int fetchPrior(IStatus* status, void* message);
// equivalent to FETCH PRIOR FROM <cursor name>

Moves the cursor’s current position to the prior row and returns it. If the cursor is empty or already positioned at the first row, the condition NO_DATA is returned.

int fetchFirst(IStatus* status, void* message);
// equivalent to FETCH FIRST FROM <cursor name>

Moves the cursor’s current position to the first row and returns it. If the cursor is empty, the condition NO_DATA is returned.

int fetchLast(IStatus* status, void* message);
// equivalent to FETCH LAST FROM <cursor name>

Moves the cursor’s current position to the last row and returns it. If the cursor is empty, the condition NO_DATA is returned.

int fetchAbsolute(IStatus* status, int position, void* message);
// equivalent to FETCH ABSOLUTE <position> FROM <cursor name>

Moves the cursor’s current position to the specified position and returns the located row. If position is beyond the cursor’s boundaries, the condition NO_DATA is returned.

int fetchRelative(IStatus* status, int offset, void* message);
// equivalent to FETCH RELATIVE <offset> FROM <cursor name>

Moves the cursor’s current position backward or forward by the specified offset and returns the located row. If the calculated position is beyond the cursor’s boundaries, the condition NO_DATA is returned.

  1. When a scrolling option is omitted, NO SCROLL is implied (i.e. the cursor is opened as forward-only). This means that only the fetchNext() API call can be used. Other fetch methods will return an error.

  2. Scrollable cursors are internally materialized as a temporary record set, thus consuming memory/disk resources, so this feature should be used only when really necessary.

SPB Support for New Statistics Feature in gbak Output

Vlad Khorsun

A new, much requested feature was added to gbak verbose output: optional run-time statistics. Read about it here. The feature is fully supported in the Services API with a new item in the SPB (Services Parameter Block),

#define isc_spb_bkp_stat 15

along with its synonym

#define isc_spb_res_stat isc_spb_bkp_stat
Usage
isc_spb_bkp_stat, <len>, <string>
isc_spb_res_stat, <len>, <string>

where <len> (2 bytes) indicates the length of the following string parameter, and <string> (1-4 bytes) is a string consisting of one character per statistics item.

The fbsvcmgr utility also supports the new SPB tags.

Better Error Reports for String Overflows

Alex Peshkov

Include expected and actual string length in the error message for string overflows (SQLCODE -802).

More Detail in “Wrong Page Type” Error Reports

Alex Peshkov

More details in the error message "wrong page type", i.e. identifying expected and encountered page types by name instead of numerical type.

New Item for isc_database_info() Call

Vlad Khorsun

An option was added to the API function isc_database_info() to return the number of free pages in a database. See .

Compression and Encryption Status Info

Alex Peshkov

Added in Firebird 3.0.3, see .

Compression details and encryption status of a connection (fb_info_conn_flags) have been added to the getInfo() API call. The data stored in the information block are of type integer in network format, accessible as isc_vax_integer.

Currently only 2 bits are meaningful:

#define isc_dpb_addr_flag_conn_compressed   0x01
#define isc_dpb_addr_flag_conn_encrypted    0x02
New Services Tag for Overriding LINGER

Alex Peshkov

The Services API now includes the tag isc_spb_prp_nolinger, for example (in one line):

fbsvcmgr host:service_mgr user sysdba password xxx
       action_properties dbname employee prp_nolinger

For information regarding LINGER, see the write-up in the DDL chapter.

Enable Use of ***_auth_plugin_list Item from Application

Firebird 3.0.4: To specify the names of authentication plug-ins, an application must create a config string with “AuthClient = <plugin-list>”, along with lines for other options. A specific DPB/SPB item actually exists, item isc_dpb_auth_plugin_list/isc_spb_auth_plugin_list that the client itself uses to communicate the plug-in list to the server. However, if an application populated that same item, it was ignored by the client.

This improvement allows an application to use the item item isc_dpb_auth_plugin_list/isc_spb_auth_plugin_list in lieu of creating and passing a config string for it.

New Connection Formats for Local Superserver Clients on Windows

In previous Firebird versions, a serverless protocol known as “Windows Local” was available to local clients connecting to Superserver on a Windows platform, using the XNET subsystem. A typical connection string looked like this:

c:\Program Files\Firebird_2_5\examples\empbuild\employee.fdb

Under the new unified server, that form of connection attempts to load an embedded server. It is no longer valid for a serverless client connection to Superserver. If you try, you will get a refusal message to the effect “File is in use by another process”. This is not a bug. Since Superserver clients share resources, another server (in this case, an embedded server) cannot attach a client to the same database that Superserver has any clients attached to.

However, all is not lost. The XNET subsystem can still do local client sessions for Superserver. You just need a more elaborate connection string now. You have a few choices:

  • this one is the former “Windows local”, using the XNET subsystem and shared memory for a (nominally) serverless connection:

    xnet://alias-or-path-to-database

    So, for our connection to the employee database:

    xnet://c:\Program Files\Firebird_3_0\examples\empbuild\employee.fdb

    or using an alias:

    xnet://employee
  • Connection to host/port via TCP:

    inet://host:port/alias-or-path-to-database
  • Connection to host/port via TCP, restricting the name search to IPv4:

    inet4://host:port/alias-or-path-to-database

    This option was added in version 3.0.1. See also the configuration parameter IPv6V6Only.

  • Connection to host/port via Named Pipes (aka NetBEUI):

    wnet://host:port/alias-or-path-to-database
  • Connection to localhost via TCP:

    inet://alias-or-path-to-database
  • Connection to localhost via named pipes (aka NetBEUI):

    wnet://alias-or-path-to-database
Perform Some Validation Services On-line

Vlad Khorsun

This feature was ported forward from Firebird 2.5.4.

Database validation enables low-level checks of the consistency of on-disk structures and even to fix some minor corruptions. The recommended procedure for any valuable database is for the DBA to validate a database periodically to ensure it is healthy.

Exclusive access to the database is required: any kind of concurrent access is forbidden during validation. Sometimes, blocking user access could be a major hold-up, especially if the database is large and complex.

Online validation is a new feature that allows some consistency checks to be performed without exclusive access.

What Online Validation Can Do
  • validate some (or all) user tables in a database.

    System tables are not validated.

  • validate some (or all) indices

Other ODS checks, such as Header/PIP/TIP/Generators pages, are not performed.

Protection During Online Validation

While a table (and/or its index) is undergoing validation, user attachments are allowed to read this table. Any attempt to change data (INSERT/UPDATE/DELETE) will wait until validation finishes or, depending on the lock timeout of the user transaction, will return a lock timeout error.

Any kind of garbage collection on the table or its indexes is disabled whilst it is undergoing validation:

  • background and cooperative garbage collection will just skip this table

  • sweep will be terminated with an error

When online validation starts to check a table, it acquires a couple of locks to prevent concurrent modifications of its data:

  • a relation lock in PR (protected read) mode

  • (NEW) a garbage collection lock in PW (protected write) mode

Both locks are acquired using a user-specified lock timeout. An error is reported for any lock request that fails and that table is skipped.

Once the locks are acquired, the table and its indexes are validated in the same way as a full validation does it. The locks are released when it completes and the whole procedure is repeated for the next table.

The New Services API action: isc_action_svc_validate

Online validation is implemented as a Firebird service and is accessed through the Services API. Thus, it cannot be run from the gfix utility.

The call involves the following elements:

Action:
  isc_action_svc_validate

Parameters:
  isc_spb_dbname :
    database file name, string, mandatory

  isc_spb_val_tab_incl, isc_spb_val_tab_excl,
  isc_spb_val_idx_incl, isc_spb_val_idx_excl :
    patterns for tables\indices names, string, optional

  isc_spb_val_lock_timeout :
    lock timeout, integer, optional

Output:
  text messages with progress of online validation process
Using isc_action_svc_validate Interactively

The fbsvcmgr utility has full support for the new service. The syntax is:

fbsvcmgr [host:]service_mgr [user <...>] [password <...>]
  action_validate dbname <filename>
  [val_tab_incl <pattern>]
  [val_tab_excl <pattern>]
  [val_idx_incl <pattern>]
  [val_idx_excl <pattern>]
  [val_lock_timeout <number>]

where

val_tab_incl

pattern for table names to include in validation run

val_tab_excl

pattern for table names to exclude from validation run

val_idx_incl

pattern for index names to include in validation run, by default ‘%’, i.e. all indexes

val_idx_excl

pattern for index names to exclude from validation run

val_lock_timeout

lock timeout, used to acquire locks for table to validate, in seconds, default is 10 secs. ‘0’ is no-wait, ‘-1’ is infinite wait

Usage Notes
  • Patterns are regular expressions, processed by the same rules as SIMILAR TO expressions.

  • All patterns are case-sensitive, regardless of database dialect.

  • If the pattern for tables is omitted then all user tables will be validated.

  • If the pattern for indexes is omitted then all indexes of the appointed tables will be validated.

  • System tables are not validated.

  • To specify a list of tables or indexes:

    1. Separate names with the pipe character ‘|

    2. Do not add spaces: “TAB1 | TAB2” is wrong, use “TAB1|TAB2

    3. Enclose the whole list in double quotes to avoid confusing the command interpreter

Examples
  1. Validate all tables in database c:\db.fdb with names starting with A. Indexes are not validated. Lock wait is not performed.

    fbsvcmgr.exe service_mgr user SYSDBA password masterkey
      action_validate dbname c:\db.fdb
      val_tab_incl A%
      val_idx_excl %
      val_lock_timeout 0
  2. Validate tables TAB1 and TAB2 and all their indexes. Lock wait timeout is 10 seconds (the default):

    fbsvcmgr.exe service_mgr user SYSDBA password masterkey
      action_validate dbname c:\db.fdb
      val_tab_incl "TAB1|TAB2"
  3. Default behavior of val_XXX options: validate all user tables and their indexes in database c:\db.fdb, lock wait is the default 10 seconds:

    fbsvcmgr.exe service_mgr user SYSDBA password masterkey
      action_validate dbname c:\db.fdb
Code Improvement

Alex Peshkov

() — The functions IStatement::execute() and IAttachment::execute() now return an error pointer to the old transaction interface.

5. Reserved Words and Changes

New Keywords in Firebird 3.0

Reserved

Items marked with asterisks (*) were previously non-reserved.

BOOLEAN

CORR

COVAR_POP

COVAR_SAMP

DELETING *

DETERMINISTIC

FALSE

INSERTING *

LOCALTIME (since 3.0.4)

LOCALTIMESTAMP (since 3.0.4)

OFFSET

OVER

RDB$RECORD_VERSION

REGR_AVGX

REGR_AVGY

REGR_COUNT

REGR_INTERCEPT

REGR_R2

REGR_SLOPE

REGR_SXX

REGR_SXY

REGR_SYY

RETURN

ROW

SCROLL

SQLSTATE

STDDEV_POP

STDDEV_SAMP

TRUE

UNKNOWN

UPDATING *

VAR_POP

VAR_SAMP

Non-reserved

ABSOLUTE

ACOSH

ASINH

ATANH

BODY

CONTINUE

DDL

DECRYPT

DENSE_RANK

ENCRYPT

ENGINE

FIRST_VALUE

IDENTITY

INCREMENT

LAST_VALUE

LAG

LEAD

LINGER

NAME

NTH_VALUE

PACKAGE

PARTITION

PLUGIN

PRIOR

RANK

RELATIVE

ROW_NUMBER

SERVERWIDE

TAGS

TRUSTED

USAGE

 

 

6. Configuration Additions and Changes

The file aliases.conf has been renamed to databases.conf. An old aliases.conf from a previous version can simply be renamed, and the new engine will just continue to use it as before. However, databases.conf can now include some configuration information for individual databases.

Scope of Parameters

Some parameters are marked as configurable per-database or per-connection.

  • Per-database configuration is done in databases.conf.

  • Per-connection configuration is primarily for client tool use and is done using the DPB parameter isc_dpb_config or, for Services, the SPB parameter isc_spb_config.

  • In the case of Embedded, the DPB can be used to tune per-database entries on first attaching to a database.

Macro Substitution

A number of predefined macros (syntax $(name)) is available for use in the configuration files to substitute for a directory name:

$(root)

Root directory of Firebird instance

$(install)

Directory where Firebird is installed. $(root) and $(install) are initially the same. $(root) can be overridden by setting or altering the environment variable FIREBIRD, in which case it becomes different from $(install).

$(this)

Directory where current configuration file is located

$(dir_conf)

Directory where firebird.conf and databases.conf are located

$(dir_secdb)

Directory where the default security database is located

$(dir_plugins)

Directory where plugins are located

$(dir_udf)

Directory where UDFs are located by default

$(dir_sample)

Directory where samples are located

$(dir_sampledb)

Directory where sample DB (employee.fdb) is located

$(dir_intl)

Directory where international modules are located

$(dir_msg)

Directory where the messages file (firebird.msg) is located. $(dir_msg) usually should be the same as $(root) but can be overridden by the environment variable FIREBIRD_MSG.

You can observe the usage of some of these macros in databases.conf.

In our pre-built binaries, $(dir_conf) and $(dir_secdb) would normally be the same as $(root) and $(install).

$(dir_plugins), $(dir_udf), $(dir_sample), $(dir_sampledb) and $(dir_intl) are predefined sub-directories inside $(root).

The build conventions are not “rules” that could be expected to apply in every distribution of Firebird. Distro-specific Linux packages, for example, each prefer to fit the Firebird components into standard layouts that comply with their own conventions. As an illustration, user binaries, such as isql might be located in /usr/bin, server binaries in /usr/sbin, configuration files in /etc/firebird.d and so on. Obviously, $(root) would then make no sense, even if the $(dir_something) macros still pointed to actual directories.

Includes

One configuration file can be included in another by using an “include” directive, e.g.,

include some_file.conf

A relative path is treated as relative to the enclosing configuration file. So, if our example above is inside /opt/config/master.conf then our include refers to the file /opt/config/some_file.conf.

Wildcards

The standard wildcards ‘*’ and ‘?’ may be used in an include directive, to include all matching files in undefined order. For example,

include $(dir_plugins)/config/*.conf

Expression of Parameter Values

Previously, byte values were specified by default as integer, representing the number of bytes. However, now you can optionally specify them in Kilobytes, Megabytes or Gigabytes, as appropriate, by adding ‘K’, ‘M’ or ‘G’ (case-insensitive). For example, “24M” is read as 25165824 (24 * 1024 * 1024).

Boolean values are expressed as non-zero (true)|zero (false) by default, but you may now use the quoted strings 'y', 'yes' or 'true' instead of a non-zero digit.

“Per-database” Configuration

Custom configuration at database level is achieved with formal entries in databases.conf.

Format of Configuration Entries

In aliases.conf the format for specifying a database alias was

aliasname = /absolute/path/to/database_file

If you are not adding any database-specific configuration directives for an alias, the format is just as it was before, e.g.,

emp = c:\Program Files\examples\empbuild\employee.fdb
 or
emp = /opt/firebird/examples/empbuild/employee.fdb
 or
emp = $(dir_sampleDb)/employee.fdb

A slightly more complex format is used for cases where certain non-global parameters are to be targeted at an individual database. The entry for the database is defined by the alias declaration, as previously. The database-specific directives are listed below it, within curly brackets.

#
# Directives for MYBIGDB
MYBIGDB = opt/databases/mybigdb.fdb
   {
      LockMemSize = 32M           # We know that MYBIGDB needs a lot of locks
      LockHashSlots = 19927       # and a hash table large enough for them
   }

Parameters Available

The following parameters can be copy/pasted to databases.conf and used as overrides for specific databases.

Table 2. Parameters available in databases.conf
Engine-related

DatabaseGrowthIncrement

DeadlockTimeout

DefaultDbCachePages

EventMemSize

FileSystemCacheThreshold

ExternalFileAccess

GCPolicy

LockAcquireSpins

LockHashSlots

LockMemSize

MaxUnflushedWrites

MaxUnflushedWriteTime

SecurityDatabase

UserManager

 

WireCompression

WireCrypt

WireCryptPlugin

Client-related

Some parameters can be configured at the client connection via the DPB/SPB, as an alternative to configuring them in databases.conf. to Scope of Parameters at the beginning of this chapter to understand these differences.

AuthClient

Providers

 

The following parameters can be configured only via the DPB/SPB

ConnectionTimeout

DummyPacketInterval

IpcName

RemoteAuxPort

RemotePipeName

RemoteServiceName

RemoteServicePort

TCPNoNagle

 

New Parameters

New parameters added to firebird.conf are:

SecurityDatabase

Defines the name and location of the security database that stores login user names and passwords used by the server to validate remote connections. By default, in firebird.conf, it is $(root)/security3.fdb. It can be overridden for a specific database by a configuration in databases.conf.

AuthServer and AuthClient

Two parameters that determine what authentication methods can be used by the network server and the client redirector. The enabled methods are listed as string symbols separated by commas, semicolons or spaces.

  • Secure remote passwords (Srp), using the plug-in is the default, using the OS-appropriate plug-in (libSrp.s0/Srp.dll/Srp.dylib)

  • On Windows, the Security Support Provider Interface (Win_Sspi) is used when no login credentials are supplied

  • Client applications can use legacy authentication (Legacy_Auth) to talk to old servers.

For AuthServer, Srp and Win_Sspi are listed; for AuthClient, Srp, Win_Sspi and Legacy_Auth.

To disable a method, erase the comment marker (‘#’) and remove the unwanted method from the list.

Both parameters can be used in databases.conf. They can both be used in the DPB or the SPB for a connection-specific configuration.

WireCrypt

Sets whether the network connection should be encrypted. It has three possible values: Required | Enabled | Disabled. The default is set such that encryption is Required for connections coming in to the server and Enabled for connections outgoing to a server.

To access a server using an older client library and, thus, no encryption, WireCrypt in the server configuration file should be set to Enabled or Disabled to avert the default Required.

The rules are simple: if one side has WireCrypt = Required and the other sets the parameter to Disabled, the side with WireCrypt=Required rejects the connection and it is not established.

A missing WireCrypt plug-in or encryption key in cases where the channel must be encrypted also thwarts a connection.

In all other cases, connection is established without encryption if at least one side has WireCrypt = Disabled. In other cases, the encrypted connection is established.

UserManager

Sets the plug-in that will operate on the security database. It can be a list with blanks, commas or semicolons as separators: the first plug-in from the list is used as the default.

The default plug-in is Srp (libSrp.s0/Srp.dll/Srp.dylib).

The UserManager parameter can be used in databases.conf for a database-specific override.

TracePlugin

Specifies the plug-in used by Firebird’s Trace facility to send trace data to the client app or audit data to the log file.

The default plug-in is fbtrace (libfbtrace.s0/fbtrace.dll/fbtrace.dylib).

WireCryptPlugin

A wire-crypt plug-in is used to encrypt and decrypt data transferred over the network.

The installation default Arc4 implies use of an Alleged RC4 plug-in. The configured plug-in, which requires a key generated by the configured authentication plug-in, can be overridden in the API for a specific connection via the DPB or the SPB.

For information about configuring plug-ins, see Configuring Plug-ins in the Engine chapter.

KeyHolderPlugin

This parameter would represent some form of temporary storage for database encryption keys. Nothing is implemented as a default plug-in, but a sample Linux plug-in named libCryptKeyHolder_example.so can be found in /plugins/.

Providers

List of allowed transports for accessing databases, discussed in the Engine chapter.

ServerMode

Determines the execution mode of the server (“server model”). Discussed in the Engine chapter.

RemoteAccess

Parameter in firebird.conf and databases.conf provides an efficient, configurable replacement for hard-coded rules limiting access to security3.fdb. It can also be used to configure limited remote access to any other database, including non-default security databases.

By default RemoteAccess is enabled for all databases except the security database. If you intend using more than one dedicated security database, then disabling remote access to it (or them) via databases.conf is recommended.

For stricter security, server-wide, you can set RemoteAccess to false in firebird.conf and use entries in database.conf to re-enable it for specific databases.

RemoteAccess is a Boolean. It can be expressed with either true/false, 1/0 or Yes/No.

WireCompression

Alex Peshkov

Parameter in firebird.conf or databases.conf, enabling or disabling compression of data over the wire at global or individual database level.

The default setting is disabled (= False). Settings and environment must be correct at both server and client for WireCompression to take effect:

  • To enable it at the server side, in firebird.conf and/or databases.conf, change the setting to True

  • To activate Wirecompression from the client side, pass the appropriate tag in the config item of the DPB or SPB call:

    isc_dbp_config/isc_sbp_config <string-length> "WireCompression=true"
  • Both server and client versions must be Firebird 3 or greater (protocol >=13)

See Tracker item .

IPv6V6Only

Michael Kubecek

Parameter in firebird.conf only. (TCP ports are created before any connection is established.)

Firebird 3 supports IPv6 connections, on both client and server sides.

Server

By default, the Firebird server listens on the zero IPv6 address (::) and accepts all incoming connections, whether IPv4 or IPv6, and IPv6V6Only is set to false (=0). If it is set to true, the server, still listening implicitly or explicitly on the zero IPv6 address, will accept only IPv6 connections.

A different listening address, either IPv4 or IPv6, can be set using the RemoteBindAddress parameter. If an IPv4 address or a non-zero IPv6 address is used, the IPv6V6Only directive has no effect.

On POSIX platforms, in Classic mode, the parameters RemoteBindAddress, RemoteServicePort and RemoteServiceName are ignored by fbserver, since the listening socket is set up by (x)inetd. The listening address and/or port need to be set in the (x)inetd.

IPv6V6Only is a Boolean. It can be expressed with either true/false, 1/0 or Yes/No.

Client

The standard text form of an IPv6 address uses the colon character to separate groups of digits (upto 8 groups of digits). In the connection string, the IPv6 address must be enclosed in square brackets, to resolve the ambiguity with the use of the colon as the separator between the host IP address and the database path. For example:

connect '[2014:1234::5]:test';
connect '[2014:1234::5]/3049:/srv/firebird/test.fdb';

For consistency, square brackets can be optionally used around an IPv4 address or a domain name.

If a domain name is used in connection string, all addresses (IPv4 and IPv6) are tried in the order returned by resolver until a connection is established. If all attempts fail, the client fails to connect.

Parameters Changed or Enhanced

The following parameters have been changed or enhanced:

ExternalFileAccess

Entries in the “Restrict” list of the ExternalFileAccess parameter can be used to mangle file names with relative paths.

Entries in the “Restrict” list were already used to mangle file names with no path component. For example, with

ExternalFileAccess = /opt/extern

and the following sequence of commands:

SQL> create table qq external file 'zz' (x int);
SQL> insert into qq values(1);
SQL> commit;

the file /opt/extern/zz will be created.

However, if something like this is submitted,

create table qq external file 'dir/zz' (x int);

the result is an error about denied access to file /opt/firebird/bin/dir/zz.

The improvement avoids this gap by mangling the file name in accord with the value of the parameter and, if necessary, creating the missing path components, such as 'dir' in the example above.

Parameters Removed or Deprecated

The following parameters have been removed or deprecated:

RootDirectory

In older version, this parameter provided a superfluous option for recording the file system path to Firebird’s “root” files (firebird.conf, the security database and so on).

LegacyHash

This parameter used to make it possible to use the old security.fdb from Firebird 1.X installations after it had been subjected to an upgrade script and thence to enable or disable use of the obsolete DES hash encrypting algorithm. It is no longer supported.

OldSetClauseSemantics

This parameter enabled temporary support for an implementation fault in certain sequences of SET clauses in versions of Firebird prior to version 2.5. It is no longer available.

OldColumnNaming

This parameter temporarily enabled legacy code support for an old InterBase/Firebird 1.0 bug that generated unnamed columns for computed output which was not explicitly aliased in the SELECT specification. It is no longer available.

LockGrantOrder

This parameter used to allow the option to have Firebird’s Lock Manager emulate InterBase v3.3 lock allocation behaviour, whereby locks would be granted in no particular order, as soon as they were available, rather than by the normal order (first-come, first-served). The legacy option is no longer supported.

Obsolete Windows priority settings

UsePriorityScheduler, PrioritySwitchDelay and PriorityBoost, which were marginally relevant to obsolete processors on obsolete Windows versions, are no longer supported.

7. Security

Security improvements in Firebird 3 include:

Location of User Lists

Alex Peshkov

Firebird now supports an unlimited number of security databases. Any database may act as a security database and can be a security database for itself.

Use databases.conf to configure a non-default security database. This example configures /mnt/storage/private.security.fdb as the security database for the first and second databases:

first = /mnt/storage/first.fdb
{
   SecurityDatabase = /mnt/storage/private.security.fdb
}

second = /mnt/storage/second.fdb
{
   SecurityDatabase = /mnt/storage/private.security.fdb
}

Here we use a third database as its own security database:

third = /mnt/storage/third.fdb
{
   SecurityDatabase = third
}

The value of the SecurityDatabase parameter can be a database alias or the actual database path.

Creating an Alternative Security Database

To start using a separate, non-default security database, the first step is to create it, unless it already exists. An embedded isql connection is used:

> isql -user sysdba
SQL> create database '/mnt/storage/private.security.fdb';

Now connect to any database which will be served by the security database you are currently preparing, in order to create its SYSDBA user:

SQL> connect first;
SQL> create user sysdba password 'sysdba-in-private-security-password';
SQL> commit;
SQL> exit;

Database Encryption

Alex Peshkov

With Firebird 3 comes the ability to encrypt data stored in database. Not all of the database file is encrypted: just data, index and blob pages.

To make it possible to encrypt a database you need to obtain or write a database crypt plug-in.

The sample crypt plug-in in examples/dbcrypt does not perform real encryption, it is merely a sample of how to go about it.

Secret Key

The main problem with database encryption is how to store the secret key. Firebird provides a helper to transfer that key from the client but that does not imply that storing the key on a client is the best way: it is no more than a possible alternative. A very bad option is to keep the key on the same disk as the database.

Encryption Tasks

To separate encryption and key access efficiently, a database crypt plug-in is split into two parts: encryption itself and the secret key holder. This may be an efficient approach for third-party plug-ins when you want to use some good encryption algorithm, but you have your own secret way to store a key.

Encrypting a Database

Once you have decided on a crypt plug-in and a key, you can encrypt the database with:

ALTER DATABASE ENCRYPT WITH <PLUGIN_NAME> [ KEY <key-name> ]

The optional KEY argument allows the key name to be passed in the command, if the plug-in requires it.

Encryption will start right after this statement commits and will be performed in background. Normal database activity is not disturbed during encryption.

Monitoring Encryption

During encryption, progress can be monitored using the field MON$CRYPT_PAGE in the pseudo-table MON$DATABASE. For example:

select MON$CRYPT_PAGE * 100 / MON$PAGES
  from MON$DATABASE

The example query will output the percentage of encryption completed so far.

You can also watch the database header page using repeated calls to gstat -e. If the database has been encrypted, gstat -h can also provide limited information about encryption state.

Decrypting a Database

To decrypt the database do:

ALTER DATABASE DECRYPT

For Linux, an example plug-in named libDbCrypt_example.so can be found in the /plugins/ sub-directory.

New Authentication Method in Firebird 3

Alex Peshkov

All of the code related to authentication is plug-in-enabled. Though Firebird performs the generic work, like extracting authentication data from a network message or putting it into such messages as appropriate, all the activity related to calculating hashes, storing data in databases or elsewhere, using specific prime numbers and so on is done by plug-ins.

Firebird 3 has new method of user authentication implemented as a default plugin: secure remote password (SRP) protocol. Quoting from :

The SRP protocol creates a large private key shared between the two parties in a manner similar to Diffie-Hellman key exchange, then verifies to both parties that the two keys are identical and that both sides have the user’s password. In cases where encrypted communications as well as authentication are required, the SRP protocol is more secure than the alternative SSH protocol and faster than using Diffie-Hellman key exchange with signed messages. It is also independent of third parties, unlike Kerberos.

SSH needs key pre-exchange between server and client when placing a public key on the server to make it work. SRP does not need that. All a client needs are login and password. All exchange happens when the connection is established.

Moreover, SRP is resistant to “man-in-the-middle” attacks.

Use of the new authentication method is not compatible with old security databases and passwords from them. However, an upgrade procedure is available to migrate users from a Firebird 2.x security2.fdb database. For instructions, see Upgrading a v.2.x Security Database in the Compatibility chapter.

Use of an old security database can be supported with the Legacy_Auth authentication plug-in, but this kills the security benefits of Firebird 3.

The Firebird 3 client is built to make it possible to talk to old servers with the default configuration.

SSL/TLS Support

So, the answer to the question “Does Firebird use SSL/TLS for password validation?” is “yes and no”. The “No” answer comes because, by default, SSL is not used. That is due to a minor licensing incompatibility between Firebird and OpenSSL, the most popular SSL implementation.

The “Yes” applies because anyone is free to write an authentication plug-in that uses SSL and TLS.

Increased Password Length

.

Implementation of SRP in our plugin has increased the password length from 8 bytes to 20 or more. Because of the use of SHA1 for hashes, maximum security is provided for passwords up to 20 characters in length. Longer passwords can be used without restriction but there is a remote possibility of hash collisions between passwords that differ beyond the 20th byte. Just be aware of the possibility that any password longer than 20 characters password could share the same hash with some shorter password so, theoretically, they could be attacked using brute force.

A custom SRP plug-in could be built quite easily, using a hash algorithm that would guarantee unique hashes for longer passwords.

The increased length limit means the default SYSDBA password in Windows and MacOS installations is the full 'masterkey' string (9 chars), no longer 'masterke' (8 chars) as in older versions!

Support for the LegacyHash and Authentication parameters in firebird.conf has been dropped. Authentication is overtaken by an AuthServer parameter in firebird.conf or elsewhere.

The Authentication Plug-in

The Authentication plug-in comprises three parts:

Client

prepares data at the client to be sent to server on client

Server

validates password for correctness

User Manager

adds, modifies and deletes users on the server. It is not needed if some external authentication method, such as Windows trusted authentication, is used.

All three parts are actually separate plug-ins which should be configured separately in firebird.conf. Let’s look at an example of configuring a server to accept connections from old clients. The default setting are:

AuthServer = Srp, Win_Sspi
UserManager = Srp

To enable access from old clients, AuthServer needs to be changed:

AuthServer = Srp, Win_Sspi, Legacy_Auth

If we also want to manage the list of users in the old format we must add:

UserManager = Legacy_UserManager

Setting UserManager to only Legacy_UserManager will disable to option to manage Srp users. See the next section for details on allowing both to be managed.

Multiple User Managers

Multiple user managers can be enumerated in firebird.conf. The first member of the list of user managers is the default. Selecting from SEC$USERS would produce something like the following:

SQL> select SEC$USER_NAME, SEC$PLUGIN from sec$users;

SEC$USER_NAME                   SEC$PLUGIN
=============================== ===============================
SYSDBA                          Legacy_UserManager
SYSDBA                          Srp
QA_USER1                        Srp
QA_USER2                        Srp
QA_USER3                        Srp
QA_USER4                        Srp
QA_USER5                        Srp
GUEST                           Srp
SHUT1                           Srp
SHUT2                           Srp
QATEST                          Srp

There might well be two users named SYSDBA in such a list, because each user manager has its own SYSDBA.

  • All user management commands can have the USING PLUGIN clause, whose purpose is to enable selection of a particular UserManager plug-in from the list in firebird.conf.

  • The default user manager in firebird.conf is Srp. If you need to manage legacy logins, add Legacy_UserManager to UserManager, e.g. UserManager = Legacy_UserManager,Srp.

  • UserManager can be configured at database level, in databases.conf.

V. 3.0.4 SRP Security Patch

Alex Peshkov; Tony Whyman

Review of the Firebird SRP implementation in terms of the NIST guidance on the use of SHA-1 appears to indicate that most uses of SHA-1 continue to be permitted except for its use in generating the client proof.

The SHA-256 message digest may be used instead of SHA-1 for generating the client proof. For compatibility, SHA-1 remains the default for Firebird 3 but SHA-256 is available as an alternative from version 3.0.4 onward, until the decision is taken to make it the default.

Plug-Ins

Separate AuthServer and AuthClient plugins are available for each supported message digest, with the following names:

Srp

SHA-1 Client Proof

Srp256

SHA-256 Client Proof

Both client and server must have an SRP authentication plug-in in common for authentication of a user’s password to succeed.

Configuration Details

The SRP UserManager is still called “Srp” and it is unchanged. Neither the user manager nor the security database is affected by the choice of message digest used to compute the client proof.

The default configuration file entries in firebird.conf for AuthServer and AuthClient are now:

AuthServer = Srp

All platforms

AuthClient = Srp, Srp256, Legacy_Auth

POSIX clients

AuthClient = Srp, Srp256, Win_Sspi, Legacy_Auth

Windows clients

With these settings, a Firebird client can use SRP to authenticate using either SHA-256 or SHA-1 to compute the client proof and is thus backwards-compatible with Firebird 3 servers.

Beware!

A deployment where both clients and server support both the legacy Srp (using SHA-1) and the Srp256 SHA-256 authentication plug-ins should be avoided, as it might enable an attacker to disrupt the Srp256 authentication, thereby forcing Firebird to use the weaker Srp SHA-1 client proof without the user being aware of it.

For a more detailed description, please read the document /doc/README.SecureRemotePassword.html in your Firebird 3.0.4 installation.

NIST Compliance

If you need NIST compliance and/or higher security you can set

AuthServer = Srp256

This will break compatibility with pre-3.0.4 clients.

"Over the wire" Connection Encryption

Alex Peshkov

…​

All network traffic in Firebird 3 may be optionally encrypted. As with authentication, plug-ins are used for encrypting and decrypting network traffic.

The default plug-in is arc4 (Alleged RC4). It is eminently possible to write your own crypt plug-in to encrypt data travelling over the wire. Whatever you use for your plug-in, it is necessary to use the Firebird 3 version of the fbclient library.

The Secret Session Key

The challenge with use of a symmetric cypher is where to get a key for it. Firebird assumes that such a key, also called a secret session key, is produced by the authentication plug-in at the connection establishment phase. SRP meets this requirement just fine by producing a cryptographically strong session key.

If you want to use encryption with an authentication plug-in that does not provide the session key and agree to use some pre-defined key, say, one stored at the client side as a file and on the server in the security database for that specific client, then make that plug-in inform Firebird that it does have a session key.

Specifications for the Key

Specifications for the key’s size, its format, how it is calculated and verified, etc., are not generalised. The key’s format and other details are specific to the wire encryption/decryption plug-in.

In particular, RC4 uses a symmetric key which can have any length, while the key produced by SRP has a length of 20 bytes. That key is a SHA-1 hash on SRP’s session key and some other SRP-related things, such as user name.

Exporting a Key from an Authentication Plug-in

To export a key from your authentication plug-in, use the ServerBlock or the ClientBlock interface. One of these is always passed to the server/client part of an authentication plug-in. Both have a “newKey” method that returns a pointer to the CryptKey interface. That interface in turn has the methods setSymmetric and setAsymmetric for storing the symmetric or asymmetric key in the interface, i.e., exporting that key.

Mapping of Users to Objects

Alex Peshkov

Firebird 3 introduces new SQL privileges to map access between users and groups and security objects and between databases. See Tracker item .

With Firebird now supporting multiple security databases, some new problems arise that could not occur with a single, global security database. Clusters of databases using the same security database were efficiently separated. Mappings provide the means to achieve the same efficiency when multiple databases are using their own security databases. Some cases require control for limited interaction between such clusters. For example:

  • when EXECUTE STATEMENT ON EXTERNAL DATA SOURCE requires some data exchange between clusters

  • when server-wide SYSDBA access to databases is needed from other clusters, using services.

  • comparable problems that have existed on Firebird 2.1 and 2.5 for Windows, due to support for Trusted User authentication: two separate lists of users — one in the security database and another in Windows, with cases where it was necessary to relate them. An example is the demand for a ROLE granted to a Windows group to be assigned automatically to members of that group.

The single solution for all such cases is mapping the login information assigned to a user when it connects to a Firebird server to internal security objects in a database — CURRENT_USER and CURRENT_ROLE.

The Mapping Rule

The mapping rule consists of four pieces of information:

  1. mapping scope — whether the mapping is local to the current database or whether its effect is to be global, affecting all databases in the cluster, including security databases

  2. mapping name — an SQL identifier, since mappings are objects in a database, like any other

  3. the object FROM which the mapping maps. It consists of four items:

    • The authentication source

      • plug-in name or

      • the product of a mapping in another database or

      • use of server-wide authentication or

      • any method

    • The name of the database where authentication succeeded

    • The name of the object from which mapping is performed

    • The type of that name — user name | role | OS group — depending upon the plug-in that added that name during authentication.

      Any item is accepted but only type is required.

  4. the object TO which the mapping maps. It consists of two items:

    • The name of the object TO which mapping is performed

    • The type, for which only USER or ROLE is valid

Syntax for MAPPING Objects

Mappings are defined using the following set of DDL statements:

{CREATE | ALTER | CREATE OR ALTER} [GLOBAL] MAPPING name
  USING {
    PLUGIN name [IN database] | ANY PLUGIN [IN database | SERVERWIDE] |
      MAPPING [IN database] | '*' [IN database]}
  FROM {ANY type | type name}
  TO {USER | ROLE} [name]
  --
DROP [GLOBAL] MAPPING name
Description
  • Any mapping may be tagged as GLOBAL.

    Global mapping works best if a Firebird 3 or higher version database is used as the security database. If you plan to use another database for this purpose — using your own provider, for example — then you should create a table in it named RDB$MAP, with the same structure as RDB$MAP in a Firebird 3 database and with SYSDBA-only write access.

    Beware!

    If global and local mappings of the same name exist then know and make it known that they are different objects!

  • The CREATE, ALTER and CREATE OR ALTER statements use the same set of options. The name (identifier) of a mapping is used to identify it, as in other DDL command sets.

  • The USING clause has a highly complicated set of options:

    • an explicit plug-in name means it will work only for that plug-in

    • it can use any available plug-in; although not if the source is the product of a previous mapping

    • it can be made to work only with server-wide plug-ins

    • it can be made to work only with previous mapping results

    • it can be left to use any method, using the asterisk (*) argument

    • it can be provided with the name of the database that originated the mapping for the FROM object

      This argument is not valid for mapping server-wide authentication.

  • The FROM clause takes a mandatory argument, the type of the object named.

    → When mapping names from plug-ins, type is defined by the plug-in.
    → When mapping the product of a previous mapping, type can be only USER or ROLE.
    → If an explicit name is provided, it will be taken into account by this mapping
    → Use the ANY keyword to work with any name of the given type.

  • In the TO clause, the USER or ROLE to which the mapping is made must be specified. name is optional: if it is not supplied, the name from the originating mapping is used.

Examples

The examples use the CREATE syntax. Usage of ALTER is exactly the same and the usage of DROP should be obvious.

  1. Enable use of Windows trusted authentication in all databases that use the current security database:

    CREATE GLOBAL MAPPING TRUSTED_AUTH
      USING PLUGIN WIN_SSPI
      FROM ANY USER
      TO USER;
  2. Enable SYSDBA-like access for windows admins in current database:

    CREATE MAPPING WIN_ADMINS
      USING PLUGIN WIN_SSPI
      FROM Predefined_Group
      DOMAIN_ANY_RID_ADMINS
      TO ROLE RDB$ADMIN;

    The group DOMAIN_ANY_RID_ADMINS does not exist in Windows, but such a name would be added by the Win_Sspi plug-in to provide exact backwards compatibility.

  3. Enable a particular user from another database to access the current database with another name:

    CREATE MAPPING FROM_RT
      USING PLUGIN SRP IN "rt"
      FROM USER U1 TO USER U2;

    Database names or aliases will need to be enclosed in double quotes on operating systems that have case-sensitive file names.

  4. Enable the server’s SYSDBA (from the main security database) to access the current database. (Assume that the database is using a non-default security database):

    CREATE MAPPING DEF_SYSDBA
      USING PLUGIN SRP IN "security.db"
      FROM USER SYSDBA
      TO USER;
  5. Ensure users who logged in using the legacy authentication plug-in do not have too many privileges:

    CREATE MAPPING LEGACY_2_GUEST
      USING PLUGIN legacy_auth
      FROM ANY USER
      TO USER GUEST;

Legacy Mapping Rule

Previous versions of Firebird have one hard-coded global default rule: users authenticated in the security database are always mapped into any database one-to-one. It is a safe rule: it makes no sense for a security database not to trust itself!

For backward compatibility this rule is retained in Firebird 3.

Generic mapping is used to set the rule defining the user name under which the user accesses a database when performing a request from a database using one security database to a database using a different one, or when server-wide authentication, such as Win_Sspi, is used. The rule comes into action whenever the Firebird engine is processing the authentication block associated with a request to connect to a database.

Mapping Trusted Users to CURRENT_USER and CURRENT_ROLE

Each record in an authentication block contains the name of the plug-in that added it, the type of record (user name, OS group, role, etc.), the name of an object of the specified type (user SYSDBA, role PUBLIC, group DOMAIN_ADMINS) and the name of the security database in which authentication took place. Under server-wide authentication, the security database name could be NULL. The mapping rule assesses all these parameters to define the value to assign to CURRENT_USER and CURRENT_ROLE in the resulting attachment.

In Firebird 3, an explicit mapping must exist in systems with server-wide “trusted user” authentication enabled, including Win_Sspi authentication on Windows, in order for the system user’s user name to be assigned to the context variables CURRENT_USER and CURRENT_ROLE.

International Character Sets for User Accounts

Alex Peshkov

For creating user names and passwords, the new authentication plug-ins bring a degree of flexibility with regard to character sets. To make use of international credentials support, it is necessary to use SRP or any other set of authentication plug-ins that supports the new model.

Advice

Use of the legacy authentication plug-in is not recommended, except where it is necessary for connecting a client to a server version older than Firebird 3. Legacy authentication has never supported international characters in user names and passwords. This restriction is unchanged for Firebird 3 legacy authentication.

When writing authentication plug-ins it is unnecessary to be concerned about character sets, since all data exchange with the new plug-ins is done using UTF8. Just be prepared to handle user names and passwords that contain characters beyond the range of the 7-bit ASCII character subset.

Conditions for Non-ASCII Logins

The simple rule to follow for using non-ASCII user names and passwords is that the character set and code page selected in the client for SQL input must match those in which the credentials are stored on the server and, obviously, the credentials must be supplied in the specified character set. On POSIX terminals and most GUI clients this condition is satisfied. However, there are three fairly common situations where problems could arise:

  1. connecting via a Windows text console with the default OEM code page selected

  2. omitting the SET NAMES command in isql

  3. running scripts

Examining these situations in more detail:

  1. All operating systems except Windows use same the same character set (code page, locale), by default, for all programs. In Windows, for historical reasons, GUI applications use an ANSI code page, but applications started on the command-line console emulator use the corresponding OEM code page. As an example, the Russian version of Windows is uses code page 1251 (ANSI Russian) for GUI applications and code page 866 (OEM Russian) for command-line applications.

    The Firebird client always uses the ANSI code page. In order to provide the fully functional behaviour of a console application, the console must be switched to the ANSI codepage (chcp 1251 for our Russian example).

    The Windows weirdness does not stop there. Parameters passed to a Windows application, but not to 16-bit DOS, are always passed in the ANSI encoding, even if it was started from a command-line console running in the default OEM mode. This might give the illusion that Firebird utilities run correctly out-of-the-box on an OEM console. However, they run correctly only as far as the international information supplied in a command line, e.g.,

    C:> isql server:database -user <intl-login> -pas <intl-password>

    This works. However, as soon as you try to use a CONNECT or CREATE DATABASE command inside isql, or to enter the password from the OEM terminal, you will encounter problems. This is inconvenient and we apologise for that. The OEM console issues are in the plan to be fixed in a future version.

  2. Currently, command-line parameters and the CONNECT and CREATE DATABASE commands in isql are not affected by SET NAMES or the -CHarset parameter. All the other commands, particularly CREATE/ALTER/DROP USER, are affected by them and thus, the character set must be specified. This is very important because, in future versions, the plan is to make all usage of international credentials depend on a character set that is explicitly specified.

    Take careful note of one very confusing use case, an attempt to set a non-ASCII password for the current user:

    SQL> alter current user set password '<intl-password>';

    This command will succeed, even if the character set has not been set correctly. However, a subsequent attempt to log in with the modified password will fail.

  3. In scripts, the first requirement is to set the correct character set for the SQL server that is to run the script, using a SET NAMES statement. For Firebird 3, it is essential to pay attention to the locale and code page settings in the environment in which the script is to execute. They must match the character set that is set in the script.

    Setting the locale correctly affects particularly the credentials passed to the server when attaching to it. The rest of the script should proceed successfully even without configuring the environment properly.

    In summary, if you plan to use international character credentials in the script, you DO need to attend to the international settings in your environment.

Reminder

User names are SQL identifiers and thus conform to the same rules, i.e. enclosed in double quotes when containing international characters or when case-sensitivity is required.

SQL Features for Managing Access

Changes in architecture, stiffening of rules for security and data integrity, along with a bucket list of feature requests, have given rise in this release to a number of new SQL commands for managing users and access to objects.

SQL-driven User Management

Alex Peshkov

The SQL set of DDL commands for managing user accounts has been enhanced in Firebird 3, thus improving the capability to manage (add, modify or delete) users in a security database from a regular database attachment.

Syntax
CREATE USER username [ <options_list> ]
  TAGS ( <tag> [, <tag> ...] )
ALTER USER username [ SET ] [ <options_list> ]
  TAGS ( <tag> [, <tag> ...] )
ALTER CURRENT USER [ SET ] [ <options_list> ]
  TAGS ( <tag> [, <tag> ...] )
CREATE OR ALTER USER username [ SET ] [ <options_list> ]
  TAGS ( <tag> [, <tag> ...] )
DROP USER username [ USING PLUGIN plugin_name ]

<options_list> is a (possibly empty) list with the following options:

PASSWORD 'password'
FIRSTNAME 'string value'
MIDDLENAME 'string value'
LASTNAME 'string value'
ACTIVE
INACTIVE
USING PLUGIN plugin_name

Each <tag> may have one of two forms:

tagname = 'string value'

or the DROP tagname tag form to remove a user-defined attribute entirely:

DROP tagname

The tagname side of the name/value pair can be any valid SQL identifier.

Older Methods Deprecated

From Firebird 3.0, multiple security databases are supported. This capability is not supported by either the gsec utility or the Services API. Use of both of these methods is deprecated.

Usage Details

The CREATE USER, CREATE OR ALTER USER and DROP USER clauses are available only for SYSDBA or another user granted the RDB$ADMIN role in the security database (and logged in under that role, of course.)

The PASSWORD clause is required when creating a new user.

An ordinary user can ALTER their own password, real name attributes and tags. Any attempt to modify another user will fail, as will an attempt to make “self” inactive or active.

If you want to modify “self”, you can use the simplified form ALTER CURRENT USER.

At least one of PASSWORD, FIRSTNAME, MIDDLENAME, LASTNAME, ACTIVE, INACTIVE or TAGS must be present in an ALTER USER or CREATE OR ALTER USER statement.

It is not a requirement to use any of the clauses FIRSTNAME, MIDDLENAME and LASTNAME. Any of them may be left empty or used to store short information about the user.

The INACTIVE clause is used to disable the user’s login capability without dropping it. The ACTIVE clause restores that login ability.

Quick Tip

From version 3.0.1, the statement CREATE OR ALTER USER SYSDBA PASSWORD password can be used to initialize an empty securityN.fdb security database.

TAGS is a list of end-user defined attributes. The length of the string value should not exceed 255 bytes.

Setting a list of tags for the user retains previously set tags if they are not mentioned in the current list.

A UID or GID that was entered by the deprecated gsec utility is treated as a tag in the SQL interface.

Examples

Generic:

CREATE USER superhero PASSWORD 'test';
ALTER USER superhero SET FIRSTNAME 'Clark' LASTNAME 'Kent';
CREATE OR ALTER USER superhero SET PASSWORD 'IdQfA';
DROP USER superhero;
ALTER CURRENT USER SET PASSWORD 'SomethingLongEnough';

Working with tags:

ALTER USER superhero SET TAGS (a='a', b='b');
      NAME             VALUE
      ================ ==============================
      A                a
      B                b

ALTER USER superhero SET TAGS (b='x', c='d');
      NAME             VALUE
      ================ ==============================
      A                a
      B                x
      C                d

ALTER USER superhero SET TAGS (drop a, c='sample');
      NAME             VALUE
      ================ ==============================
      B                x
      C                sample

Displaying the list of users:

SELECT CAST(U.SEC$USER_NAME AS CHAR(20)) LOGIN,
  CAST(A.SEC$KEY AS CHAR(10)) TAG,
  CAST(A.SEC$VALUE AS CHAR(20)) "VALUE",
  SEC$PLUGIN "PLUGIN"
FROM SEC$USERS U LEFT JOIN SEC$USER_ATTRIBUTES A
  ON U.SEC$USER_NAME = A.SEC$USER_NAME;

 LOGIN                TAG        VALUE                PLUGIN
 ==================== ========== ==================== ===============================
 SYSDBA               <null>     <null>               Srp
 SUPERHERO            B          x                    Srp
 SUPERHERO            C          sample               Srp
 SYSDBA               <null>     <null>               Legacy_UserManager

Output depends upon the user management plug-in. If the legacy plug-in is used, bear in mind that some options are not supported and will simply be ignored.

SET ROLE

Alex Peshkov

See Tracker item .

The SQL:2008-compliant operator SET ROLE allows the CURRENT_ROLE context variable to be set to one that has been granted to the CURRENT_USER or to a user assigned to the database attachment as trusted (SET TRUSTED ROLE).

Syntax for SET ROLE

Enable CURRENT_USER access to a role that has been previously granted:

SET ROLE rolename
Example of SET ROLE Usage
SET ROLE manager;
select current_role from rdb$database;

Displays:

ROLE
===============================
MANAGER
SET TRUSTED ROLE

The idea of a separate SET TRUSTED ROLE command is that, when the trusted user attaches to a database without providing any role info, SET TRUSTED ROLE makes a trusted role (if one exists) the CURRENT_ROLE without any additional activity, such as setting it in the DPB.

A trusted role is not a specific type of role but may be any role that was created using CREATE ROLE, or a predefined system role such as RDB$ADMIN. It becomes a trusted role for an attachment when the security objects mapping subsystem finds a match between the authentication result passed from the plug-in and a local or global mapping for the current database. The role may be one that is not even granted explicitly to that trusted user.

  • A trusted role is not assigned to the attachment by default. It is possible to change this behaviour using an appropriate authentication plug-in and a CREATE/ALTER MAPPING command.

  • Whilst the CURRENT_ROLE can be changed using SET ROLE, it is not always possible to revert using the same command, because it performs an access rights check.

Syntax Pattern

Enable access to a trusted role, if the CURRENT_USER is logged in under Trusted User authentication and the role is available:

SET TRUSTED ROLE

An example of the use of a trusted role is assigning the system role RDB$ADMIN to a Windows administrator when Windows trusted authentication is in use.

GRANT/REVOKE Rights GRANTED BY Specified User

Alex Peshkov

Previously, the grantor or revoker of SQL privileges was always the current user. This change makes it so that a different grantor or revoker can be specified in GRANT and REVOKE commands.

Syntax
GRANT <right> TO <object> [ { GRANTED BY | AS } [ USER ] username ]
REVOKE <right> FROM <object> [ { GRANTED BY | AS } [ USER ] username ]

The GRANTED BY clause form is recommended by the SQL standard. The alternative form using AS is supported by Informix and possibly some other servers and is included for better compatibility.

Example (working as SYSDBA)
create role r1;
grant r1 to user1 with admin option;
grant r1 to public granted by user1;

-- (in isql)
show grant;
/* Grant permissions for this database */
GRANT R1 TO PUBLIC GRANTED BY USER1
GRANT R1 TO USER1 WITH ADMIN OPTION

REVOKE ALL ON ALL

When a user is removed from the security database or another authentication source, this new command is useful for revoking its access to all objects in the database.

Syntax
REVOKE ALL ON ALL FROM [USER] username
REVOKE ALL ON ALL FROM [ROLE] rolename
Example
# gsec -del guest
# isql employee
fbs bin # ./isql employee
Database:  employee
SQL> REVOKE ALL ON ALL FROM USER guest;
SQL>

User Privileges for Metadata Changes

Dmitry Yemanov; Roman Simakov

In Firebird 3, the system tables are read-only. This SQL syntax provides the means to assign metadata write privileges to specified users or roles for specified objects. See Tracker item .

Some people have been applying the nickname “DDL privileges” to this feature. Don’t confuse it with “DDL triggers”! A more useful nickname would be “Metadata privileges”.

Syntax Patterns

Granting metadata privileges:

GRANT CREATE <object-type>
  TO [USER | ROLE] { user-name | role-name } [WITH GRANT OPTION];
GRANT ALTER ANY <object-type>
  TO [USER | ROLE] { user-name | role-name } [WITH GRANT OPTION];
GRANT DROP ANY <object-type>
  TO [USER | ROLE] { user-name | role-name } [WITH GRANT OPTION];

Revoking metadata privileges:

REVOKE [GRANT OPTION FOR] CREATE <object-type>
  FROM [USER | ROLE] { user-name | role-name };
REVOKE [GRANT OPTION FOR] ALTER ANY <object-type>
  FROM [USER | ROLE] { user-name | role-name };
REVOKE [GRANT OPTION FOR] DROP ANY <object-type>
  FROM [USER | ROLE] { user-name | role-name };

Special form for database access:

GRANT CREATE DATABASE TO [USER | ROLE] { user-name | role-name };
GRANT ALTER DATABASE
  TO [USER | ROLE] { user-name | role-name } [WITH GRANT OPTION];
GRANT DROP DATABASE
  TO [USER | ROLE] { user-name | role-name } [WITH GRANT OPTION];

REVOKE CREATE DATABASE FROM [USER | ROLE] { user-name | role-name };
REVOKE [GRANT OPTION FOR] ALTER DATABASE
  FROM [USER | ROLE] { user-name | role-name };
REVOKE [GRANT OPTION FOR] DROP DATABASE
  FROM [USER | ROLE] { user-name | role-name };
Notes on Usage
  • <object-type> can be any of the following:

    CHARACTER SET

    COLLATION

    DOMAIN

    EXCEPTION

    FILTER

    FUNCTION

    GENERATOR

    PACKAGE

    PROCEDURE

    ROLE

    SEQUENCE

    TABLE

    VIEW

     

     

     

    The metadata for triggers and indices are accessed through the privileges for the table that owns them.

  • If the ANY option is used, the user will be able to perform any operation on any object

  • If the ANY option is absent, the user will be able to perform operations on the object only if he owns it

  • If the ANY option was acquired via a GRANT operation then, to revoke it, the REVOKE operation must accord with that GRANT operation

Example
GRANT CREATE TABLE TO Joe;
GRANT ALTER ANY TABLE TO Joe;
REVOKE CREATE TABLE FROM Joe;

GRANT EXECUTE Privileges for UDFs

Dmitry Yemanov

: EXECUTE permission is now supported for UDFs (both legacy and PSQL based ones).

Syntax
GRANT EXECUTE ON FUNCTION name TO <grantee list>
  [<grant option> <granted by clause>]
--
REVOKE EXECUTE ON FUNCTION name FROM <grantee list>
  [<granted by clause>]

The initial EXECUTE permission is granted to the function owner (user who created or declared the function).

Improvement for Recursive Stored Procedures

Alex Peshkov

A recursive stored procedure no longer requires the EXECUTE privilege to call itself. See Tracker item .

Privileges to Protect Other Metadata Objects

New SQL:2008 compliant USAGE permission is introduced to protect metadata objects other than tables, views, procedures and functions.

Syntax
GRANT USAGE ON <object type> name TO <grantee list>
  [<grant option> <granted by clause>]
--
REVOKE USAGE ON <object type> name FROM <grantee list>
  [<granted by clause>]
--
<object type> ::=
  {DOMAIN | EXCEPTION | GENERATOR | SEQUENCE | CHARACTER SET | COLLATION}

The initial USAGE permission is granted to the object owner (user who created the object).

In Firebird 3.0.0, only USAGE permissions for exceptions () and generators/sequences (gen_id, next value for: ) are enforced. Granting privileges for character sets, collations and domains is disabled, making these object types unavailable for any type of GRANT or REVOKE commands. Access to them is not subject to any form of enforcement, although this could change in future releases if it is deemed appropriate.

Pseudo-Tables with List of Users

.

To access lists of users and attributes, query the virtual tables SEC$USERS and SEC$USER_ATTRIBUTES.

This feature depends highly on the user management plug-in. Take into an account that some options are ignored when using the legacy user management plug-in.

The pseudo-tables are much like the MON$-family tables used for monitoring the server. The table is created on demand when you run the statement

SELECT * FROM SEC$USERS

or

SELECT * FROM SEC$USER_ATTRIBUTES

The output lists the users (or their attributes) in the security database that is configured for the current database and available for management to the current user. SEC$USERS includes a field indicating whether a user has the RDB$ADMIN role in the security database.

Restriction on DROP ROLE (v.3.0.1)

.

Prior to Firebird 3.0.1, any user could drop a role. This is a bug which has been fixed in sub-release 3.0.1. Now, only the user who created the role and one who has been granted the role WITH ADMIN OPTION can drop it.

To obtain the proper behavior in old databases, roles should be recreated!

8. Data Definition Language (DDL)

DDL Enhancements

The following enhancements have been added to the SQL data definition language lexicon:

New Data Types

A fully-fledged BOOLEAN type is introduced in this release, along with a surfaced emulation of the SQL standard “identity” column.

BOOLEAN Data Type

Adriano dos Santos Fernandes

The SQL:2008 compliant BOOLEAN data type (8 bits) comprises the distinct truth values TRUE and FALSE. Unless prohibited by a NOT NULL constraint, the BOOLEAN data type also supports the truth value UNKNOWN as the null value. The specification does not make a distinction between the NULL value of this data type and the truth value UNKNOWN that is the result of an SQL predicate, search condition, or Boolean value expression: they may be used interchangeably to mean exactly the same thing.

As with many programming languages, the SQL BOOLEAN values can be tested with implicit truth values. For example, field1 OR field2 and NOT field1 are valid expressions.

The IS Operator

Predicates can use the operator IS [NOT] for matching. For example, field1 IS FALSE, or field1 IS NOT TRUE.

Equivalence operators (“=”, “!=”, “<>” and so on) are valid in all comparisons.

BOOLEAN Examples
CREATE TABLE TBOOL (ID INT, BVAL BOOLEAN);
COMMIT;

INSERT INTO TBOOL VALUES (1, TRUE);
INSERT INTO TBOOL VALUES (2, 2 = 4);
INSERT INTO TBOOL VALUES (3, NULL = 1);
COMMIT;

SELECT * FROM TBOOL
          ID    BVAL
============ =======
           1 <true>
           2 <false>
           3 <null>

-- Test for TRUE value
SELECT * FROM TBOOL WHERE BVAL
          ID    BVAL
============ =======
           1 <true>

-- Test for FALSE value
SELECT * FROM TBOOL WHERE BVAL IS FALSE
          ID    BVAL
============ =======
           2 <false>

-- Test for UNKNOWN value
SELECT * FROM TBOOL WHERE BVAL IS UNKNOWN
          ID    BVAL
============ =======
           3 <null>

-- Boolean values in SELECT list
SELECT ID, BVAL, BVAL AND ID < 2
  FROM TBOOL
          ID    BVAL
============ ======= =======
           1 <true>  <true>
           2 <false> <false>
           3 <null>  <false>

-- PSQL Declaration with start value
DECLARE VARIABLE VAR1 BOOLEAN = TRUE;

-- Valid syntax, but as with a comparison
-- with NULL, will never return any record
SELECT * FROM TBOOL WHERE BVAL = UNKNOWN
SELECT * FROM TBOOL WHERE BVAL <> UNKNOWN
Use of Boolean against other data types

Although BOOLEAN is not inherently convertible to any other datatype, from version 3.0.1 the strings 'true' and 'false' (case-insensitive) will be implicitly cast to BOOLEAN in value expressions, e.g.

if (true > 'false') then ...

'false' is converted to BOOLEAN. An attempt use the Boolean operators AND, NOT, OR and IS will fail. NOT 'False', for example, is invalid.

A BOOLEAN can be explicitly converted to and from string with CAST. UNKNOWN is not available for any form of casting.

Other Notes
  • Represented in the API with the FB_BOOLEAN type and FB_TRUE and FB_FALSE constants.

  • The value TRUE is greater than the value FALSE.

Keywords INSERTING, UPDATING and DELETING

To avoid ambiguities when used in Boolean expressions, the previously non-reserved keywords INSERTING, UPDATING and DELETING, which return True or False when tested in PSQL, have been made reserved words in all contexts. If you have used any of these words as identifiers for database objects, columns, variables or parameters, it will be necessary to redefine them, either with new names or by enclosing these identifiers in double quotes.

Identity Column Type

Adriano dos Santos Fernandes

An identity column is a column associated with an internal sequence generator. Its value is set automatically when the column is omitted in an INSERT statement.

Syntax
<column definition> ::=
  name <type> GENERATED BY DEFAULT AS IDENTITY [ (START WITH value) ] <constraints>

When defining a column, the optional START WITH clause allows the generator to be initialised to a value other than zero. See Tracker ticket .

<alter column definition> ::=
  name RESTART [ WITH value ]

A column definition can be altered to modify the starting value of the generator. RESTART alone resets the generator to zero; the optional WITH value clause allows the restarted generator to start at a value other than zero. See Tracker ticket .

Rules
  • The data type of an identity column must be an exact number type with zero scale. Allowed types are thus SMALLINT, INTEGER, BIGINT, NUMERIC(x,0) and DECIMAL(x,0).

  • An identity column cannot have DEFAULT or COMPUTED value.

  • An identity column cannot be altered to become a regular column. The reverse is also true.

  • Identity columns are implicitly NOT NULL (non-nullable).

  • Uniqueness is not enforced automatically. A UNIQUE or PRIMARY KEY constraint is required to guarantee uniqueness.

  • The use of other methods of generating key values for IDENTITY columns, e.g. by trigger-generator code or by allowing users to change or add them, is discouraged to avoid unexpected key violations.

Example
create table objects (
  id integer generated by default as identity primary key,
  name varchar(15)
);

insert into objects (name) values ('Table');
insert into objects (name) values ('Book');
insert into objects (id, name) values (10, 'Computer');

select * from objects;

          ID NAME
============ ===============
           1 Table
           2 Book
          10 Computer
Implementation Details

Two new columns have been inserted in RDB$RELATION_FIELDS to support identity columns: RDB$GENERATOR_NAME and RDB$IDENTITY_TYPE.

  • RDB$GENERATOR_NAME stores the automatically created generator for the column. In RDB$GENERATORS, the value of RDB$SYSTEM_FLAG of that generator will be 6.

  • Currently, RDB$IDENTITY_TYPE will currently always store the value 1 (for BY DEFAULT) for identity columns and NULL for non-identity columns. In the future this column will store the value 0, too (for ALWAYS) when Firebird implements support for this type of identity column.

Manage Nullability in Domains and Columns

Adriano dos Santos Fernandes

ALTER syntax is now available to change the nullability of a table column or a domain

Syntax
ALTER TABLE table-name ALTER field-name { DROP | SET } [NOT] NULL

ALTER DOMAIN domain-name { DROP | SET } [NOT] NULL

The success of a change in a table column from NULL to NOT NULL is subject to a full data validation on the table, so ensure that the column has no nulls before attempting the change.

A change in a domain subjects all the tables using the domain to validation.

An explicit NOT NULL on a column that depends on a domain prevails over the domain. In this situation, the changing of the domain to make it nullable does not propagate to the column.

Modify Generators (Sequences)

Adriano dos Santos Fernandes; Dmitry Yemanov

More statement options have been added for modifying generators (sequences). Where previously in SQL the only option was ALTER SEQUENCE <sequence name> RESTART WITH <value>, now a full lexicon is provided and GENERATOR and SEQUENCE are synonyms for the full range of commands.

RESTART can now be used on its own to restart the sequence at its previous start or restart value. A new column RDB$INITIAL_VALUE is added to the system table RDB$GENERATORS to store that value.

A generator (sequence) can also be [re]created or altered to include an optional “step” clause to make the generator increment the series by two or more steps instead of the default 1. The clause is implemented as INCREMENT BY number and is stored in RDB$GENERATORS in RDB$GENERATOR_INCREMENT.

Syntax
{ CREATE | RECREATE } { SEQUENCE | GENERATOR } sequence-name [ START WITH value ]
   [ INCREMENT BY number ]

CREATE OR ALTER { SEQUENCE | GENERATOR } sequence-name { RESTART | START WITH value }
   [ INCREMENT BY number ]

ALTER { SEQUENCE | GENERATOR } <sequence name> RESTART [ WITH value ]
   [ INCREMENT BY number ]
Function GEN_ID()

The legacy function GEN_ID(generator_name, step_value)) still works to set a one-time step value in its second argument. If it is used, the stored RDB$GENERATOR_INCREMENT value is overridden.

Alter the Default Character Set

Adriano dos Santos Fernandes

ALTER DATABASE
  ...
  SET DEFAULT CHARACTER SET <new_charset>

The alteration does not change any existing data. The new default character set is used only in subsequent DDL commands and will assume the default collation of the new character set.

BLOB Expressions in Computed Columns

Adriano dos Santos Fernandes

A substring from a BLOB column can now be used to define a computed column.

Example
ALTER TABLE ATABLE
  ADD ABLOB
  COMPUTED BY (SUBSTRING(BLOB_FIELD FROM 1 FOR 20))

“Linger” Database Closure for Superserver

Alex Peshkov

Sometimes it is desirable to have the Superserver engine keep the database open for a period after the last attachment is closed, i.e. to have it “linger” a while. It can help to improve performance at low cost, under conditions where the database is opened and closed frequently, by keeping resources “warm” for next time it is reopened.

Firebird 3.0 introduces an enhancement to ALTER DATABASE to manage this optional LINGER capability for databases running under Superserver.

Syntax
ALTER DATABASE SET LINGER TO seconds;
ALTER DATABASE DROP LINGER;
Usage

To set linger for the database do:

ALTER DATABASE SET LINGER TO 30; -- sets linger interval to 30 seconds

Either of the following forms will clear the linger setting and return the database to the normal condition (no linger):

ALTER DATABASE DROP LINGER;
ALTER DATABASE SET LINGER TO 0;

Dropping LINGER is not an ideal solution for the occasional need to turn it off for some once-only condition where the server needs a forced shutdown. The gfix utility now has the -NoLinger switch, which will close the specified database immediately after the last attachment is gone, regardless of the LINGER setting in the database. The LINGER setting is retained and works normally the next time.

The same one-off override is also available through the Services API, using the tag isc_spb_prp_nolinger, e.g. (in one line):

fbsvcmgr host:service_mgr user sysdba password xxx
       action_properties dbname employee prp_nolinger

See also Tracker ticket for some discussion of the development of this feature.

Option to Preserve Shadow File

Alex Peshkov

The DROP SHADOW command has a new option to preserve the shadow file in the filesystem:

DROP SHADOW shadow_num
  [{DELETE | PRESERVE} FILE]

See Tracker ticket .

New SQL for Managing Users and Access Privileges

A number of new features and enhancements have been added to the DDL lexicon for managing users and their access to objects in databases. They are described in detail in Chapter 7, Security.

SQL-driven User Management

The SQL set of DDL commands for managing user accounts has been enhanced in Firebird 3, thus improving the capability to manage (add, modify or delete) users in a security database from a regular database attachment.

gsec is deprecated!

The command-line and shell utility gsec is deprecated from this release forward. It will continue to work with security3.fdb, but it will not work with alternative security databases.

SET ROLE and SET TRUSTED ROLE

The SQL:2008-compliant operator SET ROLE allows the CURRENT_ROLE context variable to be set to one that has been granted to the CURRENT_USER or to a user assigned to the database attachment as trusted (SET TRUSTED ROLE).

GRANTED BY Clause for Privileges

Previously, the grantor or revoker of SQL privileges was always the current user. The GRANTED BY clause makes it so that a different grantor or revoker can be specified in GRANT and REVOKE commands.

REVOKE ALL ON ALL

When a user is removed from the security database or another authentication source, this new command is useful for revoking its access to all objects in the database.

GRANT/REVOKE Metadata Privileges

In Firebird 3, the system tables are read-only. This SQL syntax provides the means to assign metadata write privileges to specified users or roles for specified objects.

EXECUTE Privileges for UDFs

EXECUTE permission is now supported for UDFs (both legacy and PSQL based ones).

GRANT/REVOKE USAGE

New SQL:2008-compliant USAGE permission is introduced to protect metadata objects other than tables, views, procedures and functions.

9. Data Manipulation Language (DML)

In this chapter are the additions and improvements that have been added to the SQL data manipulation language subset in Firebird 3.0.

Supplemental SQL:2008 Features for MERGE

Adriano dos Santos Fernandes

In summary, support for MERGE was supplemented with the introduction of these features:

  • Addition of the DELETE extension ()

  • Enabling the use of multiple WHEN MATCHED | WHEN NOT MATCHED clauses () and ability to apply conditions to WHEN MATCHED | WHEN NOT MATCHED

  • Addition of the RETURNING …​ INTO …​ clause ()

The purpose of MERGE is to read data from the source and INSERT or UPDATE in the target table according to a condition. It is available in DSQL and PSQL.

Syntax
<merge statement> ::=
  MERGE
    INTO table-or-view [ [AS] correlation-name ]
    USING <table or view or derived-table> [ [AS] correlation-name ]
    ON <condition>
    <merge when>...
    <returning clause>

<merge when> ::=
  <merge when matched> |
  <merge when not matched>

<merge when matched> ::=
  WHEN MATCHED [ AND <condition> ] THEN
    { UPDATE SET <assignment list> | DELETE }

<merge when not matched> ::=
  WHEN NOT MATCHED [ AND <condition> ] THEN
    INSERT [ <left paren> <column list> <right paren> ]
      VALUES <left paren> <value list> <right paren>

Rules

At least one of <merge when matched> or <merge when not matched> should be specified.

Example

MERGE INTO customers c
  USING
    (SELECT * FROM customers_delta WHERE id > 10) cd
     ON (c.id = cd.id)
  WHEN MATCHED THEN
    UPDATE SET name = cd.name
  WHEN NOT MATCHED THEN
    INSERT (id, name)
    VALUES (cd.id, cd.name)

A right join is made between the INTO (left-side) and USING tables using the condition. UPDATE is called when a record exists in the left table (INTO), otherwise INSERT is called.

As soon as it is determined whether or not the source matches a record in the target, the set formed from the corresponding (WHEN MATCHED / WHEN NOT MATCHED) clauses is evaluated in the order specified, to check their optional conditions. The first clause whose condition evaluates to true is the one which will be executed, and the subsequent ones will be ignored.

If no record is returned in the join, INSERT is not called.

Window (Analytical) Functions

Adriano dos Santos Fernandes

According to the SQL specification, window functions (also known as analytical functions) are a kind of aggregation, but one that does not “filter” the result set of a query. The rows of aggregated data are mixed with the query result set.

The window functions are used with the OVER clause. They may appear only in the SELECT list or the ORDER BY clause of a query.

Besides the OVER clause, Firebird window functions may be partitioned and ordered.

Syntax
<window function> ::= <window function name> ([<expr> [, <expr> ...]]) OVER (
  [PARTITION BY <expr> [, <expr> ...]]
  [ORDER BY <expr>
    [<direction>]
    [<nulls placement>]
    [, <expr> [<direction>] [<nulls placement>] ...]
)

<direction> ::= {ASC | DESC}

<nulls placement> ::= NULLS {FIRST | LAST}

Aggregate Functions Used as Window Functions

All aggregate functions may be used as window functions, adding the OVER clause.

Imagine a table EMPLOYEE with columns ID, NAME and SALARY, and the need to show each employee with his respective salary and the percentage of his salary over the payroll.

A normal query could achieve this, as follows:

select
    id,
    department,
    salary,
    salary / (select sum(salary) from employee) portion
  from employee
  order by id;
Results
id  department  salary  portion
--  ----------  ------  ----------
1   R & D        10.00      0.2040
2   SALES        12.00      0.2448
3   SALES         8.00      0.1632
4   R & D         9.00      0.1836
5   R & D        10.00      0.2040

The query is repetitive and lengthy to run, especially if EMPLOYEE happens to be a complex view.

The same query could be specified in a much faster and more elegant way using a window function:

select
    id,
    department,
    salary,
    salary / sum(salary) OVER () portion
  from employee
  order by id;

Here, sum(salary) over () is computed with the sum of all SALARY from the query (the EMPLOYEE table).

Partitioning

Like aggregate functions, that may operate alone or in relation to a group, window functions may also operate on a group, which is called a “partition”.

Syntax
<window function>(...) OVER (PARTITION BY <expr> [, <expr> ...])

Aggregation over a group could produce more than one row, so the result set generated by a partition is joined with the main query using the same expression list as the partition.

Continuing the EMPLOYEE example, instead of getting the portion of each employee’s salary over the all-employees total, we would like to get the portion based on just the employees in the same department:

select
    id,
    department,
    salary,
    salary / sum(salary) OVER (PARTITION BY department) portion
  from employee
  order by id;
Results
id  department  salary  portion
--  ----------  ------  ----------
1   R & D        10.00      0.3448
2   SALES        12.00      0.6000
3   SALES         8.00      0.4000
4   R & D         9.00      0.3103
5   R & D        10.00      0.3448

Ordering

The ORDER BY sub-clause can be used with or without partitions and, with the standard aggregate functions, make them return the partial aggregations as the records are being processed.

Example
select
    id,
    salary,
    sum(salary) over (order by salary) cumul_salary
  from employee
  order by salary;
Results
id  salary  cumul_salary
--  ------  ------------
3     8.00          8.00
4     9.00         17.00
1    10.00         37.00
5    10.00         37.00
2    12.00         49.00

Then cumul_salary returns the partial/accumulated (or running) aggregation (of the SUM function). It may appear strange that 37.00 is repeated for the ids 1 and 5, but that is how it should work. The ORDER BY keys are grouped together and the aggregation is computed once (but summing the two 10.00). To avoid this, you can add the ID field to the end of the ORDER BY clause.

It’s possible to use multiple windows with different orders, and ORDER BY parts like ASC/DESC and NULLS FIRST/LAST.

With a partition, ORDER BY works the same way, but at each partition boundary the aggregation is reset.

All aggregation functions, other than LIST(), are usable with ORDER BY.

Exclusive window functions

Beyond aggregate functions are the exclusive window functions, currently divided into ranking and navigational categories. Both sets can be used with or without partition and ordering, although the usage does not make much sense without ordering.

Ranking Functions

The ranking functions compute the ordinal rank of a row within the window partition. In this category are the functions DENSE_RANK, RANK and ROW_NUMBER.

Syntax
<ranking window function> ::=
  DENSE_RANK() |
  RANK() |
  ROW_NUMBER()

The ranking functions can be used to create different type of incremental counters. Consider SUM(1) OVER (ORDER BY SALARY) as an example of what they can do, each of them in a different way. Following is an example query, also comparing with the SUM behavior.

select
    id,
    salary,
    dense_rank() over (order by salary),
    rank() over (order by salary),
    row_number() over (order by salary),
    sum(1) over (order by salary)
  from employee
  order by salary;
Results
id  salary  dense_rank  rank  row_number  sum
--  ------  ----------  ----  ----------  ---
3     8.00           1     1           1    1
4     9.00           2     2           2    2
1    10.00           3     3           3    4
5    10.00           3     3           4    4
2    12.00           4     5           5    5

The difference between DENSE_RANK and RANK is that there is a gap related to duplicate rows (relative to the window ordering) only in RANK. DENSE_RANK continues assigning sequential numbers after the duplicate salary. On the other hand, ROW_NUMBER always assigns sequential numbers, even when there are duplicate values.

Navigational Functions

The navigational functions get the simple (non-aggregated) value of an expression from another row of the query, within the same partition.

Syntax
<navigational window function> ::=
  FIRST_VALUE(<expr>) |
  LAST_VALUE(<expr>) |
  NTH_VALUE(<expr>, <offset>) [FROM FIRST | FROM LAST] |
  LAG(<expr> [ [, <offset> [, <default> ] ] ) |
  LEAD(<expr> [ [, <offset> [, <default> ] ] )
Important to Note

FIRST_VALUE, LAST_VALUE and NTH_VALUE also operate on a window frame. Currently, Firebird always frames from the first to the current row of the partition, not to the last. This is likely to produce strange results for NTH_VALUE and especially LAST_VALUE.

Example
select
    id,
    salary,
    first_value(salary) over (order by salary),
    last_value(salary) over (order by salary),
    nth_value(salary, 2) over (order by salary),
    lag(salary) over (order by salary),
    lead(salary) over (order by salary)
  from employee
  order by salary;
Results
id  salary  first_value  last_value  nth_value     lag    lead
--  ------  -----------  ----------  ---------  ------  ------
3     8.00         8.00        8.00     <null>  <null>    9.00
4     9.00         8.00        9.00       9.00    8.00   10.00
1    10.00         8.00       10.00       9.00    9.00   10.00
5    10.00         8.00       10.00       9.00   10.00   12.00
2    12.00         8.00       12.00       9.00   10.00  <null>

FIRST_VALUE and LAST_VALUE get, respectively, the first and last value of the ordered partition.

NTH_VALUE gets the n-th value, starting from the first (default) or the last record, from the ordered partition. An offset of 1 from first would be equivalent to FIRST_VALUE; an offset of 1 from last is equivalent to LAST_VALUE.

LAG looks for a preceding row, and LEAD for a following row. LAG and LEAD get their values within a distance respective to the current row and the offset (which defaults to 1) passed.

In a case where the offset points outside the partition, the default parameter (which defaults to NULL) is returned.

Advanced Plan Output

Dmitry Yemanov

PLAN output can now be output in a more structured and comprehensible form, e.g.

SELECT statement
   -> First [10]
     -> Sort [SUM, O_ORDERDATE]
       -> Aggregate
         -> Sort [L_ORDERKEY, O_ORDERDATE, O_SHIPPRIORITY]
           -> Inner Loop Join
             -> Filter
               -> Table ≪ORDERS≫ Access By ID
                 -> Bitmap
                   -> Index ≪ORDERS_ORDERDATE≫ Range Scan
             -> Filter
               -> Table ≪CUSTOMER≫ Access By ID
                 -> Bitmap
                   -> Index ≪CUSTOMER_PK≫ Unique Scan
             -> Filter
               -> Table ≪LINEITEM≫ Access By ID
                 -> Bitmap
                   -> Index ≪LINEITEM_PK≫ Unique Scan

Advanced PLAN Output in isql

New syntax SET EXPLAIN [ON | OFF ] has been added to the isql utility to surface this option. For details, refer to SET EXPLAIN Extensions for Viewing Detailed Plans in the Utilities chapter.

Internal Functions

Additions and enhancements to the internal functions set are:

SUBSTRING with Regular Expressions

Adriano dos Santos Fernandes

A substring search can now use a regular expression.

Search Pattern
SUBSTRING(<string> [NOT] SIMILAR TO <pattern> ESCAPE <char>)

Discussion: Tracker

For more information about the use of SIMILAR TO expressions, refer to README.similar_to.txt in the /doc/ subdirectory of your Firebird installation.

The regex used is the SQL one  — the same one that is used for Firebird’s SIMILAR TO searches. A guide is available in the Common Language Elements chapter of the Firebird 2.5 Language Reference, and in the DML chapter of the version 2.5 release notes available at the Firebird web site.

Inverse Hyperbolic Trigonometric Functions

Claudio Valderrama C.

The six inverse hyperbolic trigonometric functions have been implemented internally. They are:

ACOSH

Returns the hyperbolic arc cosine of a number (expressed in radians). Format: ACOSH( <number> )

ASINH

Returns the hyperbolic arc sine of a number (expressed in radians). Format: ASINH( <number> )

ATANH

Returns the hyperbolic arc tangent of a number (expressed in radians). Format: ATANH( <number> )

COSH

Returns the hyperbolic cosine of an angle (expressed in radians). Format: COSH( <number> )

SINH

Returns the hyperbolic sine of an angle (expressed in radians). Format: SINH( <number> )

TANH

Returns the hyperbolic tangent of an angle (expressed in radians). Format: TANH( <number> )

Statistical Functions

Hajime Nakagami; Adriano dos Santos Fernandes

A suite of SQL-standard-compliant statistical functions has been implemented.

Aggregate Statistical Functions

Functions comprise Sample Variance, Population Variance, Sample Standard Deviation, Population Standard Deviation, Sample Population, Population Covariance and Coefficient of Correlation. See Tracker ticket .

Syntax
<single param statistical function> ::= <single param statistical function name>(<expr>)
<single param statistical function name> := { VAR_POP | VAR_SAMP | STDDEV_POP | STDDEV_SAMP }

<dual param statistical function> ::= <dual param statistical function name>(<expr1>, <expr>>)
<dual param statistical function name> := { COVAR_POP | COVAR_SAMP | CORR }
Semantics
  • NULL is returned from VAR_SAMP, STDDEV_SAMP or COVAR_SAMP if the result count is 0 or 1

  • NULL is returned from VAR_POP, STDDEV_POP, COVAR_POP or CORR if the result count is 0

Example
SELECT STDDEV_SAMP(salary) FROM employees;
Table 3. SD/Variance Function Descriptions
Function Format Description

VAR_SAMP

VAR_SAMP( <expr> )

Returns the Sample Variance, equivalent to

(SUM(<expr> ^ 2) - SUM(<expr>) ^ 2 / COUNT(<expr>)) / (COUNT(<expr>) - 1)

VAR_POP

VAR_POP( <expr> )

Returns the the Population Variance, equivalent to

(SUM(<expr> ^ 2) - SUM(<expr>) ^ 2 / COUNT(<expr>)) / COUNT(<expr>)

STDDEV_SAMP

STDDEV_SAMP( <expr> )

Returns the Sample Standard Deviation, equivalent to

SQRT(VAR_SAMP(<expr>))

STDDEV_POP

STDDEV_POP( <expr> )

Returns the Population Standard Deviation, equivalent to

SQRT(VAR_POP(<expr>))

COVAR_SAMP

COVAR_SAMP( <expr1>, <expr2> )

Returns the Sample Population, equivalent to

(SUM(<expr1> * <expr2>) - SUM(<expr1>) * SUM(<expr2>) / COUNT(*)) / (COUNT(*) - 1)

COVAR_POP

COVAR_POP( <expr1>, <expr2> )

Returns the Population Covariance, equivalent to

(SUM(<expr1> * <expr2>) - SUM(<expr1>) * SUM(<expr2>) / COUNT(*)) / COUNT(*)

CORR

CORR( <expr1>, <expr2> )

Returns the Coefficient of Correlation, equivalent to

COVAR_POP(<expr1>, <expr2>) / (STDDEV_POP(<expr2>) * STDDEV_POP(<expr1>))

Linear Regression Functions

The suite of REGR_* functions analyses the relationships between two sets of numeric data, considering only sets that are not NULL in either expression. See Tracker ticket .

Syntax
<regr function> ::= <function name>(<expr1>, <expr2>)
<function name> := { REGR_AVGX | REGR_AVGY | REGR_COUNT | REGR_INTERCEPT |
  REGR_R2 | REGR_SLOPE | REGR_SXX | REGR_SXY | REGR_SYY }
Formulae

The formulae use the following variables:

Y: <expr1> (<expr1> IS NOT NULL AND <expr2> IS NOT NULL)
X: <expr2> (<expr1> IS NOT NULL AND <expr2> IS NOT NULL)
N: COUNT of recordset unless <expr1> IS NULL OR <expr2> IS NULL

Y and X are DOUBLE PRECISION. N is SMALLINT, INTEGER or BIGINT.

All functions eliminate expression pairs where either expression in the pair is NULL. If no rows remain, the functions (except REGR_COUNT()) return NULL.

Table 4. Linear Regression Function Descriptions
Function Format Description

REGR_AVGX

REGR_AVGX(Y, X)

Returns the average of the independent expression (Y) in the expression pair. The return value is of type DOUBLE PRECISION.

REGR_AVGX(Y, X) = SUM(X) / N

REGR_AVGY

REGR_AVGY(Y, X)

Returns the average of the dependent expression (X) in the expression pair. The return value is of type DOUBLE PRECISION.

REGR_AVGY(Y, X) = SUM(Y) / N

REGR_COUNT

REGR_COUNT(Y, X)

Returns the number of expression pairs (Y and X). The return value is of type SMALLINT, INTEGER or BIGINT. If no rows remain after elimination of pairs where either expression is NULL, the function returns 0.

REGR_COUNT(Y, X) = N

REGR_INTERCEPT

REGR_INTERCEPT(Y, X)

Returns the y-intercept of the regression line determined by a set of expression pairs (Y and X).

REGR_INTERCEPT(Y, X) = REGR_AVGY(Y, X) - REGR_SLOPE(Y, X) * REGR_AVGX(Y, X)

REGR_R2

REGR_R2(Y, X)

Returns the square of the correlation coefficient of a set of expression pairs (Y and X).

REGR_R2(Y, X) = POWER(CORR(Y, X),2)

REGR_SLOPE

REGR_SLOPE(Y, X)

Returns the slope of the regression line, determined by a set of expression pairs (Y and X).

REGR_SLOPE(Y, X) = COVAR_POP(Y, X) / VAR_POP(X)

REGR_SXX

REGR_SXX(Y, X)

Returns the sum of squares of the independent expression (Y) in an expression pair (Y and X).

REGR_SXX(Y, X) = N * VAR_POP(X)

REGR_SXY

REGR_SXY(Y, X)

Returns the sum of products of the independent expression multiplied by the dependent expression in an expression pair (Y and X).

REGR_SXY(Y, X) = N * COVAR_POP(Y, X)

REGR_SYY

REGR_SYY(Y, X)

Returns the sum of squares of the dependent expression in an expression pair (Y and X).

REGR_SYY(Y, X) = N * VAR_POP(Y)

TRIM() BLOB Arguments Lose 32 KB limit

Adriano dos Santos Fernandes

In prior versions, TRIM(substring FROM string) allowed BLOBs for both arguments, but the first argument had to be smaller than 32 KB. Now both arguments can take BLOBs of up to 4 GB.

String Literal Limit Adjustments

Adriano dos Santos Fernandes

The internal length of a string can, at some levels, be almost 64 KB. Tests demonstrated that it is safe to accept a string literal of up to that size for writing to a text BLOB. Accordingly,

  1. The (32KB - 3) “safety limit” on literal string length for writing to text BLOBs has been raised to 65,533 bytes (64KB - 3);

  2. A limit, in characters, is calculated in run-time for strings that are in multi-byte character sets, to avoid overrunning the bytes limit. For example, for a UTF8 string (max. 4 bytes/character) the run-time limit is likely to be about (floor (65533/4)) = 16383 characters.

See Tracker ticket .

Enhancements to DATEADD() Internal Function

Adriano dos Santos Fernandes

For the internal function DATEADD()

  • The function now supports a fractional value for MILLISECOND. See Tracker item .

  • the data type of input <amount> arguments has changed from INTEGER to BIGINT. See Tracker item .

DML Improvements

A collection of useful DML improvements is released with Firebird 3.

Alternatives for Embedding Quotes in String Literals

Adriano dos Santos Fernandes

It is now possible to use a character, or character pair, other than the doubled (escaped) apostrophe, to embed a quoted string inside another string. The keyword q or Q preceding a quoted string informs the parser that certain left-right pairs or pairs of identical characters within the string are the delimiters of the embedded string literal.

Syntax
<alternate string literal> ::=
    { q | Q } <quote> <alternate start char> [ { <char> }... ] <alternate end char> <quote>
Rules

When <alternate start char> is ‘(’, ‘{’, ‘[’ or ‘<’, <alternate end char> is paired up with its respective “partner”, viz. ‘)’, ‘}’, ‘]’ and ‘>’. In other cases, <alternate end char> is the same as <alternate start char>.

Inside the string, i.e. <char> items, single (not escaped) quotes can be used. Each quote will be part of the result string.

Examples
select q'{abc{def}ghi}' from rdb$database;        -- result: abc{def}ghi
select q'!That's a string!' from rdb$database;    -- result: That's a string

SQL:2008-Compliant OFFSET and FETCH Clauses

Mark Rotteveel

New SQL:2008 compliant OFFSET and FETCH clauses provide a standard equivalent for FIRST and SKIP, and an alternative for ROWS…​TO, when fetching sets from ordered output.

  • The OFFSET clause specifies the number of rows to skip

  • The FETCH clause specifies the number of rows to fetch.

As with SKIP and FIRST, OFFSET and FETCH clauses can be applied independently, in both top-level and nested query expressions. They are available in PSQL and DSQL.

Syntax
SELECT ... [ ORDER BY <expr_list> ]
  [ OFFSET <simple_value_expr> { ROW | ROWS } ]
  [ FETCH { FIRST | NEXT } [ <simple_value_expr> ] { ROW | ROWS } ONLY ]

<simple_value_expr> is a (numeric) literal, a DSQL parameter (‘?’) or a PSQL named parameter (“:namedparameter”) that resolves to an integer data type.

Examples
-- 1:
SELECT * FROM T1 ORDER BY COL1
  OFFSET 10 ROWS;
-- 2:
SELECT * FROM T1 ORDER BY COL1
  FETCH FIRST 10 ROWS ONLY;
-- 3:
SELECT * FROM (
    SELECT * FROM T1 ORDER BY COL1 DESC
    OFFSET 1 ROW
    FETCH NEXT 10 ROWS ONLY
  ) a
  ORDER BY a.COL1
  FETCH FIRST ROW ONLY;
  1. The FIRST/SKIP and ROWS clauses are non-standard alternatives.

  2. The OFFSET and/or FETCH clauses cannot be mixed with clauses from the FIRST/SKIP or ROWS alternatives in the same query expression.

  3. Expressions and column references are not allowed within either the OFFSET or the FETCH clause.

  4. Unlike the ROWS clause, OFFSET and FETCH are available only in SELECT statements.

  5. The “percentage FETCH” defined in the SQL standard is not supported.

  6. FETCH …​ WITH TIES” defined in the SQL standard is not supported.

Prohibit Edgy Mixing of Implicit/Explicit Joins

Dmitry Yemanov

While mixing of implicit and explict join syntax is not recommended at all, the parser would allows them, nevertheless. Certain “mixes” actually cause the optimizer to produce unexpected results, including “No record for fetch” errors. The same edgy styles are prohibited by other SQL engines and now they are prohibited in Firebird.

To visit some discussion on the subject, see the Tracker ticket .

Support for Left-side Parameters in WHERE Clause

Adriano dos Santos Fernandes

The following style of subquery, with the parameter in the left side of a WHERE…​IN (SELECT…​) condition, would fail with the error “The data type of the parameter is unknown”.

This style is now accepted. For example:

SELECT <columns> FROM table_1 t1
  WHERE <conditions on table_1>
  AND (? IN (SELECT some_col FROM table_2 t2 WHERE t1.id = t2.ref_id))

Better SQL coding practice would be to use EXISTS in these cases; however, developers were stumbling over this problem when using generated SQL from Hibernate, which used this undesirable style.

Enhancements to the RETURNING Clause

Adriano dos Santos Fernandes

Two enhancements were added to the RETURNING clause:

RETURNING Clause Value Can be Aliased

When using the RETURNING clause to return a value to the client, the value can now be passed under an alias.

Example Without and With Aliases
UPDATE T1 SET F2 = F2 * 10
  RETURNING OLD.F2, NEW.F2; -- without aliases

UPDATE T1 SET F2 = F2 * 10
  RETURNING OLD.F2 OLD_F2, NEW.F2 AS NEW_F2; -- with aliases

The keyword AS is optional.

RETURNING Clause from Positioned Updates and Deletes

Support has been added for a RETURNING clause in positioned (WHERE CURRENT OF) UPDATE and DELETE statements.

Example
UPDATE T1 SET F2 = F2 * 10 WHERE CURRENT OF C
  RETURNING NEW.F2;

Cursor Stability

Vlad Khorsun

Until this release, Firebird suffered from an infamous bug whereby a data modification operation could loop infinitely and, depending on the operation, delete all the rows in a table, continue updating the same rows ad infinitum or insert rows until the host machine ran out of resources. All DML statements were affected (INSERT, UPDATE, DELETE, MERGE). It occurred because the engine used an implicit cursor for the operations.

To ensure stability, rows to be inserted, updated or deleted had to be marked in some way in order to avoid multiple visits. Another workaround was to force the query to have a SORT in its plan, in order to materialize the cursor.

From Firebird 3, engine uses the Undo log to check whether a row was already inserted or modified by the current cursor.

This stabilisation does not work with SUSPEND loops in PSQL.

An Improvement for Global Temporary Tables

Vlad Khorsun

Global temporary tables (GTTs) are now writable even in read-only transactions. The effect is as follows:

Read-only transaction in read-write database

Writable in both ON COMMIT PRESERVE ROWS and ON COMMIT DELETE ROWS

Read-only transaction in read-only database

Writable in ON COMMIT DELETE ROWS only

Also

  • Rollback for GTT ON COMMIT DELETE ROWS is faster

  • Rows do not need to be backed out on rollback

  • Garbage collection in GTT is not delayed by active transactions of other connections

The same refinements were also backported to Firebird 2.5.1.

An Improvement for DML Strings

Adriano dos Santos Fernandes

Strings in DML queries are now transformed or validated to avoid passing malformed strings to the engine internals, for example, to the MON$STATEMENTS.MON$SQL_TEXT column.

The solution adopted depends on the character set of the attachment:

  • NONE — non-ASCII characters are transformed to question marks

  • Others — the string is checked for malformed characters

COUNT() Now Returns BIGINT

The COUNT() aggregate function now returns its result as BIGINT instead of INTEGER.

Optimizations

Optimizations made for this release included:

SIMILAR TO

Adriano dos Santos Fernandes

The performance of SIMILAR TO was improved.

OR’ed Parameter in WHERE Clause

Dmitry Yemanov

Performance for (table.field = :param or :param = -1) in the WHERE clause was enhanced.

Better Choices for Navigation

Dmitry Yemanov

Previously, when an ORDER plan was in a SELECT structure, the optimizer would choose the first index candidate that matched the ORDER BY or GROUP BY clause. This “first come” approach is not the best when multiple index choices are available. The Firebird 3 engine surveys all of the available choices and picks the most suitable index.

See Tracker ticket .

Plainer Execution Path for UNION Queries

Dmitry Yemanov

Previously, the execution path for UNION queries was hierarchical, often causing redundant reads. This optimization replaces the hierarchical execution path with a plainer one that improves performance.

See Tracker ticket .

Index Walk for Compound Index

Dmitry Yemanov

The optimizer now allows an index walk (ORDER plan) when a suitable compound index (A, B) is available for a query condition of the style WHERE A = ? ORDER BY B.

See Tracker ticket .

Performance Improvement for SET STATISTICS INDEX

Vlad Khorsun

BTR_selectivity() would walk the whole leaf level of given index b-tree to calculate index selectivity. Throughout the process, the only rescheduling would happen at a disk I/O operation. The effect was to impose long waits for AST requests from concurrent attachments, such as page lock requests, monitoring, cancellation, etc. An improvement in Firebird 3 seems to solve that problem.

See Tracker ticket .

Dialect 1 Interface

Adriano dos Santos Fernandes

Selection of SQL_INT64, SQL_DATE and SQL_TIME in dialect 1 was enabled.

See Tracker

Embedded SQL (ESQL) Enhancements

Dmitry Yemanov

Two enhancements were included in the Embedded SQL subset in this release:

Support for UPDATE OR INSERT statement

See Tracker ticket

Support for the RETURNING clause

See Tracker ticket

Context Variables

Context variables added in version 3.0.4:

WIRE_COMPRESSED and WIRE_ENCRYPTED

Vlad Khorsun

Firebird 3.0.4: Context variables WIRE_COMPRESSED and WIRE_ENCRYPTED were added to the SYSTEM namespace to report compression and encryption status, respectively, of the current connection:

WIRE_COMPRESSED

Compression status of the current connection. If the connection is compressed, returns TRUE; if it is not compressed, returns FALSE. Returns NULL if the connection is embedded.

WIRE_ENCRYPTED

Encryption status of the current connection. If the connection is encrypted, returns TRUE; if it is not encrypted, returns FALSE. Returns NULL if the connection is embedded.

LOCALTIME and LOCALTIMESTAMP

Adriano dos Santos Fernandes

Firebird 3.0.4: Context variables LOCALTIME and LOCALTIMESTAMP were added as synonyms for CURRENT_TIME and CURRENT_TIMESTAMP, respectively, in anticipation of CURRENT_TIME and CURRENT_TIMESTAMP being redefined in Firebird 4 to represent the time and timestamp in UTC time, to comply with the SQL standards.

LOCALTIME

Returns a timestamp carrying no date part. The time part registers the time of day, to thousandths of a second, when execution of the current statement began. If the statement results in multiple operations, the timestamp written to all records created or updated will be the same.

LOCALTIMESTAMP

Returns a timestamp registering the date and the time of day, to thousandths of a second, when execution of the current statement began. If the statement results in multiple operations, all records created or updated will have the same timestamp.

10. Procedural SQL (PSQL)

Advancements in procedural SQL (PSQL) include:

PSQL Stored Functions

Dmitry Yemanov

It is now possible to write a scalar function in PSQL and call it just like an internal function.

Syntax for the DDL
{CREATE [OR ALTER] | ALTER | RECREATE} FUNCTION name
  [(<param1> [, ...])]
  RETURNS <type>
AS
BEGIN
  ...
END

The CREATE statement is the declaration syntax for PSQL functions, parallel to DECLARE for legacy UDFs.

Example
CREATE FUNCTION F(X INT) RETURNS INT
AS
BEGIN
  RETURN X+1;
END;
SELECT F(5) FROM RDB$DATABASE;

PSQL Sub-routines

Adriano dos Santos Fernandes

The header of a PSQL module (stored procedure, stored function, trigger, executable block) can now accept sub-procedure and sub-function blocks in the header declarations for use within the body of the module.

Syntax for Declaring a Sub-procedure
DECLARE PROCEDURE name [(<param1> [, ...])]
[RETURNS (<param1> [, ...])]
AS
...
Syntax for declaring a Sub-function
DECLARE FUNCTION name [(<param1> [, ...])]
RETURNS <type>
AS
...
Examples
SET TERM ^;
--
-- Sub-function in EXECUTE BLOCK
--
EXECUTE BLOCK RETURNS (N INT)
AS
  DECLARE FUNCTION F(X INT) RETURNS INT
  AS
  BEGIN
    RETURN X+1;
  END
BEGIN
  N = F(5);
  SUSPEND;
END ^
--
-- Sub-function inside a stored function
--
CREATE OR ALTER FUNCTION FUNC1 (n1 INTEGER, n2 INTEGER)
  RETURNS INTEGER
  AS
  DECLARE FUNCTION SUBFUNC (n1 INTEGER, n2 INTEGER)
    RETURNS INTEGER
    AS
    BEGIN
      RETURN n1 + n2;
    END
    BEGIN
      RETURN SUBFUNC(n1, n2);
    END ^
--
select func1(5, 6) from rdb$database ^

Packages

Adriano dos Santos Fernandes

Acknowledgement

This feature was sponsored with donations gathered at the fifth Brazilian Firebird Developers' Day, 2008

A package is a group of procedures and functions managed as one entity. The notion of “packaging” the code components of a database operation addresses several objectives:

Modularisation

The idea is to separate blocks of interdependent code into logical modules, as programming languages do.

In programming, it is well recognised that grouping code in various ways, in namespaces, units or classes, for example, is a good thing. With standard procedures and functions in the database this is not possible. Although they can be grouped in different script files, two problems remain:

  1. The grouping is not represented in the database metadata.

  2. Scripted routines all participate in a flat namespace and are callable by everyone (we are not referring to security permissions here).

To facilitate dependency tracking

We want a mechanism to facilitate dependency tracking between a collection of related internal routines, as well as between this collection and other routines, both packaged and unpackaged.

Firebird packages come in two parts: a header (keyword PACKAGE), and a body (keyword PACKAGE BODY). This division is very similar to a Delphi unit, the header corresponding to the interface part, and the body corresponding to the implementation part.

The header is created first (CREATE PACKAGE), and the body (CREATE PACKAGE BODY) follows.

Whenever a packaged routine determines that it uses a certain database object, a dependency on that object is registered in Firebird system tables. Thereafter, to drop, or maybe alter that object, you first need to remove what depends on it. As it is a package body that depends on it, that package body can just be dropped, even if some other database object depends on this package. When the body is dropped, the header remains, allowing you to recreate its body once the changes related to the removed object are done.

To facilitate permission management

It is good practice in general to create routines to require privileged use and to use roles or users to enable the privileged use. As Firebird runs routines with the caller privileges, it is necessary also to grant resource usage to each routine when these resources would not be directly accessible to the caller. Usage of each routine needs to be granted to users and/or roles.

Packaged routines do not have individual privileges. The privileges act on the package. Privileges granted to packages are valid for all package body routines, including private ones, but are stored for the package header.

For example
GRANT SELECT ON TABLE secret TO PACKAGE pk_secret;
GRANT EXECUTE ON PACKAGE pk_secret TO ROLE role_secret;
To enable “private scope”

This objective was to introduce private scope to routines; that is, to make them available only for internal usage within the defining package.

All programming languages have the notion of routine scope, which is not possible without some form of grouping. Firebird packages also work like Delphi units in this regard. If a routine is not declared in the package header (interface) and is implemented in the body (implementation), it becomes a private routine. A private routine can only be called from inside its package.

Signatures

For each routine that is assigned to a package, elements of a digital signature (the set of [routine name, parameters and return type]) are stored in the system tables.

The signature of a procedure or routine can be queried, as follows:

SELECT...
-- sample query to come

Packaging Syntax

<package_header> ::=
  { CREATE [OR ALTER] | ALTER | RECREATE } PACKAGE name
  AS
  BEGIN
      [ <package_item> ... ]
  END

<package_item> ::=
  <function_decl> ; |
  <procedure_decl> ;

<function_decl> ::=
  FUNCTION name [( <parameters> )] RETURNS <type>

<procedure_decl> ::=
  PROCEDURE name [( <parameters> ) [RETURNS ( <parameters> )]]

<package_body> ::=
  { CREATE | RECREATE } PACKAGE BODY name
  AS
  BEGIN
    [ <package_item> ... ]
    [ <package_body_item> ... ]
  END

<package_body_item> ::=
  <function_impl> |
  <procedure_impl>

<function_impl> ::=
  FUNCTION name [( <parameters> )] RETURNS <type>
  AS
  BEGIN
     ...
  END
  |
  FUNCTION name  [( <parameters> )] RETURNS <type>
    EXTERNAL NAME 'name' ENGINE engine

<procedure_impl> ::=
  PROCEDURE name [( <parameters> ) [RETURNS ( <parameters> )]]
  AS
  BEGIN
     ...
  END
  |
  PROCEDURE name [( <parameters> ) [RETURNS ( <parameters> )]]
    EXTERNAL NAME 'name' ENGINE engine

<drop_package_header> ::=
  DROP PACKAGE name

<drop_package_body> ::=
  DROP PACKAGE BODY name
Syntax rules
  • All routines declared in the header and at the start of the body should be implemented in the body with the same signature, i.e. you cannot declare the routine in different ways in the header and in the body.

  • Default values for procedure parameters cannot be redefined in <package_item> and <package_body_item>. They can be in <package_body_item> only for private procedures that are not declared.

  • DROP PACKAGE drops the package body before dropping its header.

  • The source of package bodies is retained after ALTER/RECREATE PACKAGE. The column RDB$PACKAGES.RDB$VALID_BODY_FLAG indicates the state of the package body. See Tracker item .

  • UDF declarations (DECLARE EXTERNAL FUNCTION) are currently not supported inside packages.

  • Syntax is available for a description (COMMENT ON) for package procedures and functions and their parameters. See Tracker item .

Simple Packaging Example

SET TERM ^;
-- package header, declarations only
CREATE OR ALTER PACKAGE TEST
AS
BEGIN
  PROCEDURE P1(I INT) RETURNS (O INT); -- public procedure
END

-- package body, implementation
RECREATE PACKAGE BODY TEST
AS
BEGIN
  FUNCTION F1(I INT) RETURNS INT; -- private function
  PROCEDURE P1(I INT) RETURNS (O INT)
  AS
  BEGIN
  END
  FUNCTION F1(I INT) RETURNS INT
  AS
  BEGIN
    RETURN 0;
  END
END ^

More examples can be found in the Firebird installation, in examples/package/.

DDL triggers

Adriano dos Santos Fernandes

Acknowledgement

This feature was sponsored with donations gathered at the fifth Brazilian Firebird Developers' Day, 2008

The purpose of a “DDL trigger” is to enable restrictions to be placed on users who attempt to create, alter or drop a DDL object.

Syntax
<database-trigger> ::=
  {CREATE | RECREATE | CREATE OR ALTER}
    TRIGGER name
    [ACTIVE | INACTIVE]
    {BEFORE | AFTER} <ddl event>
    [POSITION number]
  AS
  BEGIN
  	...
  END

<ddl event> ::=
	ANY DDL STATEMENT
  | <ddl event item> [{OR <ddl event item>}...]

<ddl event item> ::=
    CREATE TABLE
  | ALTER TABLE
  | DROP TABLE
  | CREATE PROCEDURE
  | ALTER PROCEDURE
  | DROP PROCEDURE
  | CREATE FUNCTION
  | ALTER FUNCTION
  | DROP FUNCTION
  | CREATE TRIGGER
  | ALTER TRIGGER
  | DROP TRIGGER
  | CREATE EXCEPTION
  | ALTER EXCEPTION
  | DROP EXCEPTION
  | CREATE VIEW
  | ALTER VIEW
  | DROP VIEW
  | CREATE DOMAIN
  | ALTER DOMAIN
  | DROP DOMAIN
  | CREATE ROLE
  | ALTER ROLE
  | DROP ROLE
  | CREATE SEQUENCE
  | ALTER SEQUENCE
  | DROP SEQUENCE
  | CREATE USER
  | ALTER USER
  | DROP USER
  | CREATE INDEX
  | ALTER INDEX
  | DROP INDEX
  | CREATE COLLATION
  | DROP COLLATION
  | ALTER CHARACTER SET
  | CREATE PACKAGE
  | ALTER PACKAGE
  | DROP PACKAGE
  | CREATE PACKAGE BODY
  | DROP PACKAGE BODY
Semantics
  1. BEFORE triggers are fired before changes to the system tables. AFTER triggers are fired after system table changes.

    Important Rule

    The event type [BEFORE | AFTER] of a DDL trigger cannot be changed.

  2. When a DDL statement fires a trigger that raises an exception (BEFORE or AFTER, intentionally or unintentionally) the statement will not be committed. That is, exceptions can be used to ensure that a DDL operation will fail if the conditions are not precisely as intended.

  3. DDL trigger actions are executed only when committing the transaction in which the affected DDL command runs. Never overlook the fact that what is possible to do in an AFTER trigger is exactly what is possible to do after a DDL command without autocommit. You cannot, for example, create a table in the trigger and use it there.

  4. With “CREATE OR ALTER” statements, a trigger is fired one time at the CREATE event or the ALTER event, according to the previous existence of the object. With RECREATE statements, a trigger is fired for the DROP event if the object exists, and for the CREATE event.

  5. ALTER and DROP events are generally not fired when the object name does not exist. For the exception, see point 6.

  6. The exception to rule 5 is that BEFORE ALTER/DROP USER triggers fire even when the user name does not exist. This is because, underneath, these commands perform DML on the security database, and the verification is not done before the command on it is run. This is likely to be different with embedded users, so do not write code that depends on this.

  7. If some exception is raised after the DDL command starts its execution and before AFTER triggers are fired, AFTER triggers will not be fired.

  8. Packaged procedures and triggers do not fire individual {CREATE | ALTER | DROP} {PROCEDURE | FUNCTION} triggers.

Permissions

The following users can create, alter or drop DDL triggers and access the trigger-related switches in the Firebird utilities:

  • the database owner

  • SYSDBA

  • a user logged in under the RDB$ADMIN role

  • a user having the ALTER DATABASE metadata privilege

Support in Utilities

A DDL trigger is a type of database trigger, so the parameters -nodbtriggers (gbak and isql) and -T (nbackup) apply to them. Remember that only users with the appropriate metadata privileges can use these switches.

DDL_TRIGGER Context Namespace

The introduction of DDL triggers brings with it the new DDL_TRIGGER namespace for use with RDB$GET_CONTEXT. Its usage is valid only when a DDL trigger is running. Its use is valid in stored procedures and functions called by DDL triggers.

The DDL_TRIGGER context works like a stack. Before a DDL trigger is fired, the values relative to the executed command are pushed onto this stack. After the trigger finishes, the values are popped. So in the case of cascade DDL statements, when a user DDL command fires a DDL trigger and this trigger executes another DDL command with EXECUTE STATEMENT, the values of the DDL_TRIGGER namespace are the ones relative to the command that fired the last DDL trigger on the call stack.

Elements of DDL_TRIGGER Context
EVENT_TYPE

event type (CREATE, ALTER, DROP)

OBJECT_TYPE

object type (TABLE, VIEW, etc)

DDL_EVENT

event name (<ddl event item>), where <ddl_event_item> is EVENT_TYPE || ' ' || OBJECT_TYPE

OBJECT_NAME

metadata object name

OLD_OBJECT_NAME

for tracking the renaming of a domain (see note)

NEW_OBJECT_NAME

for tracking the renaming of a domain (see note)

SQL_TEXT

sql statement text

ALTER DOMAIN old-name TO new-name sets OLD_OBJECT_NAME and NEW_OBJECT_NAME in both BEFORE and AFTER triggers. For this command, OBJECT_NAME will have the old object name in BEFORE triggers, and the new object name in AFTER triggers.

Examples Using DDL Triggers

Here is how you might use a DDL trigger to enforce a consistent naming scheme, in this case, stored procedure names should begin with the prefix “SP_”:

set auto on;
create exception e_invalid_sp_name 'Invalid SP name (should start with SP_)';

set term !;

create trigger trig_ddl_sp before CREATE PROCEDURE
as
begin
  if (rdb$get_context('DDL_TRIGGER', 'OBJECT_NAME') not starting 'SP_') then
    exception e_invalid_sp_name;
end!

-- Test

create procedure sp_test
as
begin
end!

create procedure test
as
begin
end!

-- The last command raises this exception and procedure TEST is not created
-- Statement failed, SQLSTATE = 42000
-- exception 1
-- -E_INVALID_SP_NAME
-- -Invalid SP name (should start with SP_)
-- -At trigger 'TRIG_DDL_SP' line: 4, col: 5

set term ;!

Implement custom DDL security, in this case restricting the running of DDL commands to certain users:

create exception e_access_denied 'Access denied';

set term !;

create trigger trig_ddl before any ddl statement
as
begin
  if (current_user <> 'SUPER_USER') then
    exception e_access_denied;
end!

-- Test

create procedure sp_test
as
begin
end!

-- The last command raises this exception and procedure SP_TEST is not created
-- Statement failed, SQLSTATE = 42000
-- exception 1
-- -E_ACCESS_DENIED
-- -Access denied
-- -At trigger 'TRIG_DDL' line: 4, col: 5

set term ;!

Use a trigger to log DDL actions and attempts:

create sequence ddl_seq;

create table ddl_log (
  id bigint not null primary key,
  moment timestamp not null,
  user_name varchar(31) not null,
  event_type varchar(25) not null,
  object_type varchar(25) not null,
  ddl_event varchar(25) not null,
  object_name varchar(31) not null,
  sql_text blob sub_type text not null,
  ok char(1) not null
);

set term !;

create trigger trig_ddl_log_before before any ddl statement
as
  declare id type of column ddl_log.id;
begin
  -- We do the changes in an AUTONOMOUS TRANSACTION, so if an exception happens
  -- and the command didn't run, the log will survive.
  in autonomous transaction do
  begin
    insert into ddl_log (id, moment, user_name, event_type, object_type,
                         ddl_event, object_name, sql_text, ok)
      values (next value for ddl_seq, current_timestamp, current_user,
              rdb$get_context('DDL_TRIGGER', 'EVENT_TYPE'),
              rdb$get_context('DDL_TRIGGER', 'OBJECT_TYPE'),
              rdb$get_context('DDL_TRIGGER', 'DDL_EVENT'),
              rdb$get_context('DDL_TRIGGER', 'OBJECT_NAME'),
              rdb$get_context('DDL_TRIGGER', 'SQL_TEXT'),
              'N')
      returning id into id;
    rdb$set_context('USER_SESSION', 'trig_ddl_log_id', id);
  end
end!

-- Note: the above trigger will fire for this DDL command. It's good idea to
-- use -nodbtriggers when working with them!
create trigger trig_ddl_log_after after any ddl statement
as
begin
  -- Here we need an AUTONOMOUS TRANSACTION because the original transaction
  -- will not see the record inserted on the BEFORE trigger autonomous
  -- transaction if user transaction is not READ COMMITTED.
  in autonomous transaction do
     update ddl_log set ok = 'Y'
     where id = rdb$get_context('USER_SESSION', 'trig_ddl_log_id');
end!

commit!

set term ;!

-- Delete the record about trig_ddl_log_after creation.
delete from ddl_log;
commit;

-- Test

-- This will be logged one time
-- (as T1 did not exist, RECREATE acts as CREATE) with OK = Y.
recreate table t1 (
  n1 integer,
  n2 integer
);

-- This will fail as T1 already exists, so OK will be N.
create table t1 (
  n1 integer,
  n2 integer
);

-- T2 does not exist. There will be no log.
drop table t2;

-- This will be logged twice
-- (as T1 exists, RECREATE acts as DROP and CREATE) with OK = Y.
recreate table t1 (
  n integer
);

commit;
select id, ddl_event, object_name, sql_text, ok
  from ddl_log order by id;

 ID DDL_EVENT                 OBJECT_NAME                      SQL_TEXT OK
=== ========================= ======================= ================= ======
  2 CREATE TABLE              T1                                   80:3 Y
====================================================
SQL_TEXT:
recreate table t1 (
    n1 integer,
    n2 integer
)
====================================================
  3 CREATE TABLE              T1                                   80:2 N
====================================================
SQL_TEXT:
create table t1 (
    n1 integer,
    n2 integer
)
====================================================
  4 DROP TABLE                T1                                   80:6 Y
====================================================
SQL_TEXT:
recreate table t1 (
    n integer
)
====================================================
  5 CREATE TABLE              T1                                   80:9 Y
====================================================
SQL_TEXT:
recreate table t1 (
    n integer
)
====================================================

Scrollable (Bi-directional) Cursor Support

Dmitry Yemanov

Instead of just fetching rows sequentially in a forward direction, “scrollability” allows flexible navigation through an open cursor set both backwards and forwards. Rows next to, prior to and relative to the current cursor row can be targetted. In PSQL, a scrollable cursor can be operated on directly. API support is available to enable DSQL applications to fetch rows in a similar manner.

Cursor Syntax for PSQL

To declare a cursor:

DECLARE cursor-name SCROLL CURSOR FOR ( <select expression> )

To fetch forward:

FETCH cursor-name [INTO var-name [, var-name ...]];

To fetch in any direction:

FETCH {NEXT | PRIOR | FIRST | LAST | ABSOLUTE n | RELATIVE n}
  FROM cursor-name [INTO var-name [, var-name ...]];

See also Scrollable Cursor Support for DSQL in the chapter entitled Changes to the Firebird API and ODS. The section entitled Scrollable Cursor Usage explains a little more about the usage of the various FETCH options.

  1. When a scrolling option is omitted, NO SCROLL is implied (i.e. the cursor is opened as forward-only). This means that only FETCH [NEXT FROM] commands can be used. Other commands will return an error.

  2. Scrollable cursors are internally materialized as a temporary record set, thus consuming memory/disk resources, so this feature should be used only when really necessary.

Exceptions with parameters

Adriano dos Santos Fernandes

An exception can now be defined with a message containing slots for parameters which are filled and passed when raising the exception, using the syntax pattern

EXCEPTION name USING ( <value list> )

Examples

create exception e_invalid_val 'Invalid value @1 for the field @2';
...
if (val < 1000) then
   thing = val;
else
   exception e_invalid_val using (val, 'thing');
end
CREATE EXCEPTION EX_BAD_SP_NAME
  'Name of procedures must start with ''@1'' : ''@2''';
CREATE TRIGGER TRG_SP_CREATE BEFORE CREATE PROCEDURE
AS
DECLARE SP_NAME VARCHAR(255);
BEGIN
  SP_NAME = RDB$GET_CONTEXT('DDL_TRIGGER', 'OBJECT_NAME');

  IF (SP_NAME NOT STARTING 'SP_')
  THEN EXCEPTION EX_BAD_SP_NAME USING ('SP_', SP_NAME);
END;

The status vector is generated using this code combination: isc_except, <exception number>, isc_formatted_exception, <formatted exception message>, <exception parameters>

Since a new error code (isc_formatted_exception) is used, the client must be version 3.0, or at least use the firebird.msg file from version 3.0, in order to translate the status vector to a string.

Considering, in left-to-right order, each parameter passed in the exception-raising statement as “the Nth”, with N starting at 1:

  • If an Nth parameter is not passed, the text is not substituted.

  • If NULL is passed, it is replaced by the string '*** null ***'.

  • If more parameters are passed than are defined in the exception message, the surplus ones are ignored.

  • The total length of the message, including the values of the parameters, is still limited to 1053 bytes.

CONTINUE in Looping Logic

Adriano dos Santos Fernandes

CONTINUE is a complementary command to BREAK/LEAVE, allowing flow of control to break (leave) and start of the next iteration of a FOR/WHILE loop.

Syntax
CONTINUE [label];
Example
FOR SELECT A, D FROM ATABLE INTO :achar, :ddate
  DO BEGIN
    IF (ddate < current_data - 30) THEN
      CONTINUE;
    ELSE
      /* do stuff */
    ...
  END

PSQL Cursor Stabilization

Vlad Khorsun

PSQL cursors without SUSPEND inside are now stable:

FOR SELECT ID FROM T WHERE VAL IS NULL INTO :ID
  DO BEGIN
    UPDATE T SET VAL = 1
     WHERE ID = :ID;
  END

Previously, this block would loop interminably. Now, the loop will not select the value if it was set within the loop.

This could change the behaviour of legacy code.

If there is a SUSPEND inside the block, the old instability remains: this query, for example, still produces an infinite loop:

FOR SELECT ID FROM T INTO :ID
  DO BEGIN
    INSERT INTO T (ID) VALUES (:ID);
    SUSPEND;
  END

Extension of Colon Prefix Usage

Adriano dos Santos Fernandes

Hitherto, the colon (‘:’) prefix has been used in PSQL to mark a reference to a variable in DML statements. Its use has been extended in Firebird 3 for two unrelated purposes:

  1. to allow OLD/NEW fields in cursors to be read or assigned to and to assign them to variables.

  2. to make variable assignment in both DML and PSQL statements in modules and blocks more flexible and, where needed, to resolve ambiguity between field names and variable names

PSQL Cursors as Variables

Referencing cursors as record variables is now supported in PSQL. Either explicit (DECLARE AS CURSOR) or implicit (FOR SELECT) PSQL cursors make their current record available via the cursor name, thus making the INTO clause optional.

In FOR SELECT loops, it requires the AS CURSOR clause to be specified. For example:

execute block as
begin
  for
    select id, x from t1 as cursor c1
  do begin
    for select id, x from t2 where x = :c1.x as cursor c2 do
    begin
          /* . . . */
    end
  end
end

Notice the extension of the use of a colon (‘:’) as a prefix to the referenced cursor field.

Another example
for
  select rdb$relation_id as id, rdb$relation_name as name
  from rdb$relations
  where rdb$view_blr is null
  as cursor tables
do begin
  out_id = tables.id;
  out_name = tables.name;
  suspend;
end

To avoid ambiguity, the colon prefix could be used:

out_id = :tables.id;
out_name = :tables.name;
/* or */
:out_id = :tables.id;
:out_name = :tables.name;

“tables” here is a cursor name and acts similarly to OLD/NEW in triggers.

Colon Prefix as a Variable Marker

It is now valid to apply the colon prefix to a variable on either side of an assignment statement in PSQL-only constructs.

Previously, these were valid assignments:

var1 = :var2;
/* or */
new.fld = :var;

whereas, these were invalid:

:var1 = :var2;
/* or */
:new.fld = :var;

The extension () fixes this inconsistency.

This is now valid syntax:

create trigger t1 before insert on t1
as
  declare v integer;
begin
  :v = :old.n;
  :new.n = :v;
end

In fact, using this example, there is no difference between

:v = :old.n;
/* and */
v = :old.n;

Here, it is just “syntactic sugar” but, in other cases, it provides the means to resolve ambiguity between field name references and variable names.

Example of possible ambiguity
for
  select rdb$relation_id as id, rdb$relation_name as name
  from rdb$relations
  where rdb$view_blr is null
  as cursor tables
do begin
  out_id = :table.id;
  select tables.name from tables where tables.id = :tables.id into :out_name;
  suspend;
end

Inside the nested SELECT, “tables” is both a table name and a cursor name here, so the colon is used to resolve the ambiguity.

SQLSTATE in Exception Handlers

Dmitry Yemanov

An SQLSTATE code becomes a valid condition for trapping an exception with a WHEN statement. In alignment with SQLCODE and GDSCODE, the SQLSTATE code can be used as in the following snippet:

...
WHEN SQLSTATE '22006' DO
BEGIN
  -- do something
END
...

SQLSTATE codes are string literals.

EXECUTE BLOCK Run-time Errors

From version 3.0.1, line and column numbers (location context) are now provided for run-time errors raised inside EXECUTE BLOCK.

PSQL Body Size Limit Raised

If and only if the new API is being used, the size of the body of a stored procedure or a trigger can exceed the traditional limit of 32 KB. Currently, as a security measure, a hard-coded limit of 10MB is imposed, although the theoretical limit is 4GB. The traditional size of all input or output parameters no longer limited to the traditional size of (64KB minus overhead), either.

For details, see Some SQL Size Limits Removed Using New API in the API chapter.

11. Monitoring & Command-line Utilities

No new monitoring features or other utilities are released with Firebird 3.0. Existing features have undergone a few improvements.

Monitoring

Dmitry Yemanov

Several changes have been made to the set of virtual tables storing the monitoring information. These are listed in the System Tables section of Chapter 4, Changes to the Firebird API and ODS.

Remote Client Recognition

From this version forward, the port number of the remote client address for TCPv4 and TCPv6 protocols is included in MON$ATTACHMENTS.MON$REMOTE_ADDRESS. A new column in that table also reports the host name of the remote client. For details, see Changes to Client Address Reporting in the same chapter.

Tracing

Latest improvements to the Trace functions include:

  • Trace output now supports showing the explained plan. See Tracker (V. Khorsun).

  • Tracing execution of stored functions. See Tracker (V. Khorsun).

  • Trace no longer ignores the fact that users from different security databases are actually different users. It now “knows” in which security database a particular SYSDBA was authenticated or where a privileged user’s elevated privileges are established. It is no longer possible, for example, for a user named SYSDBA to access security3.fdb if it is not currently authenticated for that access. See See Tracker (A. Peshkov).

gbak

New “Skip Data” Backup Option

Alex Peshkov

gbak has a new option switch — -skip_d(ata) to ignore the data from specific tables during a backup.

The switch -skip_d(ata) accepts a regular expression as its argument. For example, to skip two tables in the employee database (aliased here as 'employee'):

gbak -skip_d '(sales|customer)' employee e1.fbk

The regex used is the SQL one — the same one that is used for Firebird’s SIMILAR TO searches. A guide is available in the Common Language Elements chapter of the Firebird 2.5 Language Reference, and in the DML chapter of the version 2.5 release notes available at the Firebird web site.

Long Names for Log Files

Alex Peshkov

This improvement allows the gbak log to take an extra-long name without encountering the message “Attempt to store 256 bytes in a clumplet”.

Run-time Statistics in Verbose Output

Vlad Khorsun

gbak can now show some runtime statistics in its verbose output. A new command-line switch -STATISTICS has been added to specify which statistics items should be produced. Currently, four items are implemented:

Table 5. Arguments for gbak STATISTICS Output
Item Argument Data Reported

Total time

T

Time elapsed since the gbak process started, in seconds and milliseconds

Time delta

D

Time elapsed since the previous line of output, in seconds and milliseconds

Page reads

R

Number of page reads since the previous line of output, integer

Page writes

W

Number of page writes since the previous line of output, integer

At least one item is mandatory for the STATISTICS switch. The arguments are case-insensitive and they can be in any order. For example, “TDRW” and “WdrT” are equivalent.

The STATISTICS switch will have no effect if the -v[er] switch is not specified.

Verbose output with STATISTICS includes two special lines:

  • a line with headers for the specified statistics, printed before the other statistics lines:

    gbak: time delta reads writes
  • a line with total statistics summaries for the specified items, printed after the end of the main process:

    gbak: 46.684 0.002 171 82442 total statistics

The feature is fully supported in the Services API with a new item in the SPB (Services Parameter Block). The fbsvcmgr utility also supports the SPB implementation.

Examples
  1. Show total time since gbak start:

    gbak -v -STATISTICS T -b employee emp_bkp.fbk -y log21.log
  2. Show delta time and page reads:

    ... -STAT DR ...
  3. Show all statistics items:

    ... -sta TDRW ...
  4. Sample of gbak verbose output:

    firebird>gbak -v -stat tdrw -r a.fbk a.fdb
    gbak:opened file a.fbk
    gbak: time delta reads writes
    gbak: 0.173 0.173 0 0 transportable backup -- data in XDR format
    gbak: 0.175 0.002 0 0 backup file is compressed
    gbak: 0.177 0.001 0 0 backup version is 10
    gbak: 0.270 0.092 0 650 created database a.fdb, page_size 8192 bytes
    gbak: 0.273 0.002 0 2 started transaction
    ...
    gbak: 18.661 0.002 0 0 restoring data for table TEST1
    gbak: 18.698 0.036 0 0 10000 records restored
    ...
    gbak: 25.177 0.036 0 0 1770000 records restored
    gbak: 25.220 0.042 0 1633 1780000 records restored
    ...
    gbak: 38.702 0.002 0 0 restoring privilege for user SYSDBA
    gbak: 38.707 0.004 22 0 creating indexes
    gbak: 45.015 6.308 82 38394 activating and creating deferred index T2_VAL
    ...
    gbak: 46.682 0.008 4 13 finishing, closing, and going home
    gbak: 46.684 0.002 171 82442 total statistics
    gbak:adjusting the ONLINE and FORCED WRITES flags
See also

Tracker ticket

New Return Code

Alex Peshkov

From version 3.0.1, gbak returns a non-zero result code when restore fails on creating and activating a deferred user index.

gsec

The gsec utility is deprecated from Firebird 3 forward. This means you are encouraged to use the new SQL features for managing access described in Chapter 7, Security, in preference to existing equivalents provided by gsec.

gsec will continue to work with security3.fdb. However, it does not work with alternative security databases.

isql

SET EXPLAIN Extensions for Viewing Detailed Plans

Dmitry Yemanov

A new SET option is added: SET EXPLAIN [ON | OFF]. It extends the SET PLAN option to report the explained plan instead of the standard one.

If SET PLAN is omitted, then SET EXPLAIN turns the plan output on. SET PLANONLY works as in previous versions.

Usage options
SET PLAN

simple plan + query execution

SET PLANONLY

simple plan, no query execution

SET PLAN + SET EXPLAIN

explained plan + query execution

SET PLAN + SET EXPLAIN + SET PLANONLY

explained plan, no query execution

SET EXPLAIN

explained plan + query execution

SET EXPLAIN + SET PLANONLY

explained plan, no query execution

Metadata Extract

Claudio Valderrama C.

The metadata extract tool (-[e]x[tract] switch) was improved to create a script that takes the dependency order of objects properly into account.

A label has also been added, reflecting the deterministic flag for stored functions. (A. dos Santos Fernandes)

Path to INPUT Files

Adriano dos Santos Fernandes

The INPUT command will now use a relative path based on the directory of the last-opened, unclosed file in the chain to locate the next file.

Command Buffer Size Increase

Adriano dos Santos Fernandes

The size of the isql command buffer has increased from 64 KB to 10 MB to match the new engine limits. See Tracker ticket .

Deterministic Label for Stored Functions

Adriano dos Santos Fernandes

A label was added in the SHOW FUNCTION command reflecting the deterministic flag for stored functions.

See Tracker item .

fb_lock_print

Input Arguments

Dmitry Yemanov

fb_lock_print now accepts 32-bit integers as the input arguments for seconds and intervals. Previously they were limited to SMALLINT.

Usability Improvements

Vlad Khorsun

A few other small improvements:

  1. More detailed usage help is available from the command line (-help).

  2. Events history and list of owners are no longer output by default: they may be requested explicitly if required. Header-only is the new default.

  3. New -o[wners] switch to print only owners (locks) with pending requests

gfix

-NoLinger Switch

Alex Peshkov

gfix has a new switch -NoLinger to provide a one-off override to the LINGER setting of a database.

For information regarding LINGER, see the write-up in the DDL chapter.

Improvements to Validation Messages

Vlad Khorsun

  • Critical validation messages are now split from minor ones

  • The table name is now returned in the text of validation contraint error messages, to help identify the error context

Other Tweaks

Some implementation annoyances were cleared up in several utilities.

All Command-line Utilities

Resolution of Database Path

Alex Peshkov

All utilities resolve database paths in databases.conf when they need to access a database file directly. But not all of them would follow the same rules when expanding a database name. Now, they do.

Help and Version Information

Claudio Valderrama C.

All command-line utilities except gpre and qli now present help and version information in a unified and coherent way.

No info yet at .

Hard-coded Messages Replaced

Claudio Valderrama C.

Hard-coded messages were replaced with the regular parameterised-style ones in tracemanager and nbackup.

Arbitrary Switch Syntax Clean-up

Claudio Valderrama C.

Switch options in qli and nbackup were made to check the correctness (or not) of the abbreviated switch options presented.

12. Compatibility Issues

In this section are features and modifications that might affect the way you have installed and used Firebird in earlier releases.

Where Are the Tools?

On Windows, you will find all of the executable programs, including the command-line and shell tools, in the Firebird installation (root) folder. In previous versions they were in a folder beneath the root folder, named .\bin.

On POSIX platforms, you will find the tools and other executable programs in similar locations to those used for earlier versions. Exactly where depends on the distribution you are using.

Other Windows Changes

  • Notice that the installer no longer gives the option to build fbclient.dll as gds32.dll or to locate it in the system folder. If your applications still require this, you will have to do it manually.

aliases.conf Is No More

The file aliases.conf is replaced by databases.conf in the Firebird root directory. The format for database aliases has not changed, so you can copy/paste the contents of your existing aliases.conf file into databases.conf successfully. The new file is capable of carrying a lot more configuration detail, however, to enable database-level configuration of many features that were previously available only at the global server level.

Embedded Connections

For an embedded connection, an authenticated login is no longer required on POSIX platforms. If you provide a user name and password, the password is ignored. Applications may still require a user name and possibly a role name, due to SQL privileges in databases.

This feature is new for embedded Firebird on POSIX, but it is the way embedded always worked on Windows, when the embedded engine was a separate executable.

Initializing the Security Database

By default, Firebird 3 is configured for the new authentication model which uses SRP to work with user passwords and generate unique session identifiers for traffic encryption. The security database (security3.fdb) has no predefined users. This is intentional.

However, when using the standard installers for Windows, Linux and MacOSX, the SYSDBA user is created during the final step of the installation, with a password that is either random or defined via a user input routine.

In a situation where the SYSDBA initialization step either fails, or is missing from an OS-specific install process, such as a Windows .zip kit install, or a dedicated POSIX platform port, it may be necessary to initialize the security database manually for use with the SRP plugins. You will need to create the user SYSDBA and set up the password for it using SQL CREATE USER command syntax in embedded mode as your first step to getting access to databases and utilities.

The gsec utility can be used instead but, having been deprecated, it is not discussed here.

This initialization is not required and should not be performed if you have configured the server to use legacy (pre-Firebird 3 style) authentication and user management. The legacy security plugins totally preserve the legacy behaviour and thus contain the legacy record for SYSDBA with “masterke” as the initial password.

Instructions for configuring firebird.conf for legacy authentication behaviour are in the next section.

Initialization Steps

Initialization is performed in embedded mode using the isql utility. For an embedded connection, an authentication password is not required, and will be ignored if you provide one. An embedded connection will work fine with no login credentials and “log you in” using your host credentials if you omit a user name. However, even though the user name is not subject to authentication, creating or modifying anything in the existing security database requires that the user be SYSDBA; otherwise, isql will throw a privilege error for the CREATE USER request.

The SQL user management commands will work with any open database. Because the sample database employee.fdb is present in your installation and already aliased in databases.conf, it is convenient to use it for the user management task.

  1. Stop the Firebird server. Firebird 3 caches connections to the security database aggressively. The presence of server connections may prevent isql from establishing an embedded connection.

  2. In a suitable shell, start an isql interactive session, opening the employee database via its alias:

    > isql -user sysdba employee
  3. Create the SYSDBA user:

    SQL> create user SYSDBA password 'SomethingCryptic';
    SQL> commit;
    SQL> quit;
    The SYSDBA user will have full administrator rights automatically.
    Do not assign the ADMIN role (`RDB$ADMIN`) to SYSDBA.
    Quick Tip

    From version 3.0.1, the statement CREATE OR ALTER USER SYSDBA PASSWORD <password> can be used to initialize an empty securityN.fdb security database.

  4. To complete the initialization, start the Firebird server again. Now you will be able to perform a network login to databases, including the security database, using the password you assigned to SYSDBA.

About Passwords
  • An effective password, using the default user manager Srp, can be up to 20 characters, although a password of up to 255 characters will be valid.

  • If you intend in future to configure the server to use legacy authentication (not recommended!), then only the first 8 characters of any password, including that of the SYSDBA, will be read.

  • masterkey is NOT a recommended password for SYSDBA!

Additional Notes About Security Databases

Since Firebird 2, users — including SYSDBA — could not log in to the security database directly. With Firebird 3 it is possible to establish a direct, embedded connection to the security database. With the appropriate configuration parameters, you can control the ability to connect remotely to other security databases.

The gsec utility, although deprecated in Firebird 3, is still available for access to security3.fdb only. It cannot be used to manage a custom user database. We recommend that database admins plan to move away from relying on gsec and become familiar with the newer SQL user management features.

Legacy Authentication

If you do not intend to use SRP encrypted log-ins right away and want to use the security database — security3.fdb — as you have done in previous Firebird versions, proceed as follows:

  1. Using a text editor, open firebird.conf and find the entry for the parameter UserManager:

    #UserManager = Srp

    Delete the “#” symbol and change the value to:

    UserManager = Legacy_UserManager
  2. Find the entry for the WireCrypt parameter:

    #WireCrypt = Enabled (for client) / Required (for server)

    Delete the “#” symbol and change the value to:

    WireCrypt = Enabled
    -- or, if you don't plan to use SRP encryption at all --
    WireCrypt = Disabled
  3. Find the entry for the AuthServer parameter:

    #AuthServer = Srp, WinSspi, Legacy_Auth

    Delete the “#” symbol and change the order of the arguments:

    AuthServer = Legacy_Auth, Srp, WinSspi
  4. Find the entry for the AuthClient parameter:

    #AuthClient = Srp, WinSspi, Legacy_Auth

    Delete the “#” symbol and change the order of the arguments:

    AuthClient = Legacy_Auth, Srp, WinSspi
  5. Save the changes.

  6. Stop and restart Firebird for the changes to take effect.

Legacy Passwords
  • The old masterke password is available for your first login as SYSDBA. It is known to the whole world and should be changed as soon as possible.

  • Reminder: Legacy authentication reads only the first 8 characters of any password.

Upgrading a v.2.x Security Database

Because of the new authentication model in Firebird 3, upgrading a version 2.5 security database (security2.fdb) directly for use under Firebird 3 is not possible. However, an upgrade procedure is in place to enable retention of the user account data — user name, first name, etc., but not passwords — from the security2.fdb database that was used under version 2.x servers.

The SYSDBA user is not touched: after the upgrade procedure, it will remain as it was before.

Steps

  • In the commands below, replace masterkey with the actual SYSDBA password for the server version, as appropriate.

  • The procedure requires running the script security_database.sql that is located in the misc/upgrade directory of your Firebird 3 installation. These instructions assume you have a temporary copy of this script in the same directory as the isql executable.

  1. Under the Firebird 2.5 server, back up the security database. Here, use the SYSDBA password for the version 2.5 installation:

    gbak -user sysdba -pas masterkey -b {host/path}security2.fdb security2.fbk
  2. Under the Firebird 3 server, restore a copy of the version 2.5 backup. Here, use the SYSDBA password for the version 3 installation:

    gbak -user sysdba -pas masterkey -c security2.fbk {host/path}security2db.fdb
  3. Under the Firebird 3 server, go to the directory where the isql utility is located and run the upgrade script:

    isql -user sysdba -pas masterkey -i security_database.sql {host/path}security2db.fdb

    "security2db.fdb" is just a sample name for the database: it can be any preferred name.

  4. The procedure will generate new, random passwords and will output them to screen afterwards. Capture the output and notify users of their new passwords.

  5. When you are ready, shut down Firebird. Rename security3.fdb or move it another location; then rename the upgraded database to security3.fdb.

Local Connections to Superserver on Windows

In previous Firebird versions, a serverless protocol known as “Windows Local” was available to local clients connecting to Superserver on a Windows platform, using the XNET subsystem. A typical connection string looked like this:

c:\Program Files\Firebird_2_5\examples\empbuild\employee.fdb

Under the new unified server, that form of connection is no longer valid for a serverless client connection to Superserver. It attempts to load an embedded server. If you try whilst Superserver is connected to your database, you will get a refusal message to the effect “File is in use by another process”.

This is not a bug. Since Superserver clients share resources, another server (in this case, an embedded server) cannot attach a client to the same database that Superserver has any clients attached to.

However, all is not lost. The XNET subsystem can still do local client sessions for Superserver. You just need a more elaborate connection string now:

xnet://alias-or-path-to-database

So, for our connection to the employee database:

xnet://c:\Program Files\Firebird_3_0\examples\empbuild\employee.fdb

or using an alias:

xnet://employee

New connection strings are available as alternatives for other local connection protocols, too. For more information, see New Connection Formats for Local Clients on Windows.

Configuration Parameters

The previously deprecated firebird.conf parameters CompleteBooleanEvaluation, OldColumnNaming and OldSetClauseSemantics are no longer supported anymore and have been removed. The lack of one or more of these parameters may break your application code, so please check these settings in firebird.conf on your older server version.

The parameters UsePriorityScheduler, PrioritySwitchDelay, PriorityBoost, LegacyHash and LockGrantOrder no longer have any use and have been removed.

Ensure that you study the chapter Configuration Additions and Changes in preparation for upgrading user software to Firebird 3.

System Tables are Now Read-only

The system tables (RDB$*) are now read-only. Attempting any DDL or insert/update/delete statement on them will be rejected.

SQL Language Changes

It will be necessary to pay attention to some changes in the SQL language implementation.

Support for Mixed-Syntax Joins is Gone

Improperly mixed explicit and implicit joins are no longer supported, in accordance with the SQL specification. It also means that, in the explicit A JOIN B ON <condition>, the condition is not allowed to refer to any stream except A and B. See for more details.

Enforcement of Length Limit for Local Alias and Variable Names

The names for column and table aliases and for local variables names in PSQL are now restricted to 31 bytes in length. This enforcement has been made to comply with the SQL specification that requires them to be regular SQL identifiers, in accordance with the implementation-dependent limits. In Firebird, SQL identifiers are limited to 31 bytes.

See also:

Changes for User Names

User names are treated as SQL identifiers and are accordingly now limited to 31 bytes in length.

Case-sensitive user names are also supported now. The CONNECT and CREATE DATABASE statements in isql will thus allow user names to be specified in double quotes.

Cursor Output Columns Must Be Named

The DECLARE CURSOR statement in PSQL now requires all of the output columns to be explicitly named or aliased. The same requirement applies to the FOR SELECT …​ AS CURSOR <cursor name> DO …​ statement in PSQL. This requirement is necessitated by the new capability to read cursor elements directly as pseudo columns, e.g. MY_CURSOR.COLUMN_A.

Illustration
create procedure sp_test
as
  declare c cursor for (select 1 /* as a */ from rdb$database);
  declare n int;
begin
  open c;
  fetch c into n;
  close c;
end
Statement failed, SQLSTATE = 42000
unsuccessful metadata update
-ALTER PROCEDURE SP_TEST failed
-Dynamic SQL Error
-SQL error code = -104
-Invalid command
-no column name specified for column number 1 in derived table C

“Cursor Stability” Effects

Some statements may now work differently due to the “cursor stability” improvement. Statements affected will be:

  • those that modify the table that is being explicitly or implicitly selected from within the same statement

  • (as a side effect) some MERGE statements, that might work differently if multiple matches are possible.

    The SQL standard stipulates that the MERGE statement must raise an error if multiple matches are found. Firebird is not so strict in this regard, but its behaviour should be considered undefined in these cases.

See also: , , .

Reserved Words

A number of new reserved keywords are introduced. Please refer to the chapter Reserved Words and Changes and ensure your DSQL statements and procedure/trigger sources do not contain any of those keywords as identifiers. Otherwise, it will be necessary either to use them quoted (in Dialect 3 only) or to rename them.

Pay special attention to the keywords INSERTING, UPDATING and DELETING in your PSQL modules. They are now reserved words and thus must not be used as identifiers.

Also check very carefully that the keyword BOOLEAN is not used as a domain name in your databases.

Trace Configuration Syntax

The format of sections and key-value pairs in the trace configuration file has changed:

Old format
<database %[\\/](test|azk2|rulez).fdb>
  enabled true
  time_threshold 100
  log_statement_finish true
</database>
New format
database = %[\\/](test|azk2|rulez).fdb
{
  enabled = true
  time_threshold = 100
  log_statement_finish = true
}

Pay attention to the “database” section header and equality signs (‘=’) between parameter names and values.

Unexpected GDSCODE Errors

An unexpected GDSCODE may be returned for UPDATE/DELETE statements, viz. isc_deadlock instead of isc_lock_conflict.

In fact, the complete error stack previously contained {isc_lock_conflict, isc_deadlock, isc_update_conflict}, whereas it now contains, more correctly, {isc_deadlock, isc_update_conflict}.

However, the GDSCODE system variable returns the first error element, thus causing possible compatibility issues. Be sure that, besides isc_lock_conflict, your error handlers for updates and deletes also check for the other error codes, isc_deadlock and isc_update_conflict.

Remote Client Recognition

From this version forward, the remote client address for TCPv4 and TCPv6 protocols in MON$ATTACHMENTS.MON$REMOTE_ADDRESS includes the port number, separated from the IP address with a slash:

<IP address>/<port>

A new column in that table also reports the host name of the remote client.

For details, see Changes to Client Address Reporting in the chapter entitled Changes to the Firebird API and ODS.

13. Bugs Fixed

Firebird 3.0.7 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.7 official release:

Core Engine

() — Memory leak was possible when a computed field containing COALESCE/CAST is selected.

fixed by A. dos Santos Fernandes


() — Adding NOT NULL column with DEFAULT value could cause default values to be volatile or have the wrong charset.

fixed by A. dos Santos Fernandes


() — Inter-dependent computed fields could be wrongly evaluated as NULL in some cases.

fixed by V. Khorsun


() — Alternate quoting did not work on some particular cases.

fixed by A. dos Santos Fernandes


Server Crashes/Hang-ups

() — Server would crash attempting to create a table with many (more than 5460) fields.

fixed by D. Yemanov


() — Server would crash receiving a malformed packet from the network.

fixed by A. Peshkov


() — Engine could hang due to races when starting the crypt thread and simultaneous shutdown.

fixed by A. Peshkov


() — Wire compression could cause the server to freeze.

fixed by A. Peshkov


() — New connections to the server could stall, when there is an existing connection to the database.

fixed by V. Khorsun


() — New connection to the server could hang inside the engine.

fixed by V. Khorsun


() — Locked fb_lock file could cause a server crash.

fixed by V. Khorsun


API/Remote Interface

() — Client library code could assert when closing the event thread in the debug build.

fixed by A. Peshkov


() — Wrong reference counting in the UDR trigger sample.

fixed by V. Khorsun


Utilities

isql

() — Script with extracted (using isql -x) metadata could fail creating the global mapping.

fixed by A. Peshkov


gbak

() — Space inside the database path could prevent backup using GBAK in the -SE mode.

fixed by A. Peshkov


fbsvcmgr

() — Error "Unknown tag (0) in isc_svc_query() result" could be reported when running trace with fbsvcmgr.

fixed by A. Peshkov


Firebird 3.0.6 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.6 official release:

Core Engine

() — Rolled back transaction could produce unexpected results leading to duplicate values in the primary key.

fixed by D. Yemanov


() — Memory leak was possible when executing some statements and expressions.

fixed by A. dos Santos Fernandes


() — File-system ID could be duplicated among databases located on different volumes.

fixed by D. Yemanov


() — Assigning RDB$DB_KEY to a multi-byte CHAR/VARCHAR variable/parameter did not enforce the target limit.

fixed by A. dos Santos Fernandes


() — Varchar length limit was not enforced when assigning a string with trailing spaces in multi-byte character sets.

fixed by A. dos Santos Fernandes


() — Many connections could result in poor performance.

fixed by V. Khorsun


() — Hex number used at the end of statement could read invalid memory and produce wrong values or exceptions.

fixed by A. dos Santos Fernandes


() — MERGE statement could lose parameters in the “WHEN [NOT] MATCHED” clause that will never be matched. This could also cause server crashes in some situations.

fixed by A. dos Santos Fernandes


() — Deleting records from MON$ATTACHMENTS using the ORDER BY clause didn’t close the corresponding attachments.

fixed by D. Yemanov


() — UNIQUE CONSTRAINT violation could be possible.

fixed by V. Khorsun


() — Signature mismatch error could be raised when creating package body on identical packaged procedure header.

fixed by A. dos Santos Fernandes


() — Output parameters could be wrongly described if there are too many number of columns in the result set.

fixed by A. dos Santos Fernandes


() — It was impossible to connect to a database if the security.db reference was removed from databases.conf.

fixed by A. Peshkov


() — Incorrect implementation of allocFunc() for zlib1: memory leak was possible.

fixed by A. Peshkov


() — 64-bit transaction IDs were not stored properly inside the status vector.

fixed by I. Eremin


() — RDB$CHARACTER_LENGTH in RDB$FIELDS was not being populated when the column was a computed VARCHAR without an explicit type.

fixed by A. dos Santos Fernandes


Server Crashes/Hang-ups

() — Server could crash on overflow of numeric division result.

fixed by A. Peshkov


() — Server could crash executing long GRANT statements.

fixed by A. dos Santos Fernandes


() — Server could crash when using SET TRANSACTION and ON TRANSACTION START trigger uses EXECUTE STATEMENT against current transaction.

fixed by V. Khorsun


() — Server would crash when built-in function LEFT or RIGHT is missing its 2nd argument.

fixed by A. dos Santos Fernandes


() — Server would crash during shutdown of XNET connection to a local database when events have been registered.

fixed by V. Khorsun


() — Server could crash due to double destruction of the rem_port object.

fixed by D. Kovalenko, A. Peshkov


() — Dangerous (possibly leading to a crash) work with pointer: delete ptr; ptr=new ;.

fixed by D. Kovalenko, A. Peshkov


() — External engine trigger would crash the server if the table has computed fields.

fixed by A. dos Santos Fernandes


Builds

() — Distributed header files were incomplete (missing some type declarations).

fixed by V. Khorsun


() — Building the bundled libtommath could fail.

fixed by A. Peshkov


API/Remote Interface

() — Result of isNullable() in message metadata, returned by metadata builder, did not match datatype set by setType() in metadata builder.

fixed by A. Peshkov


() — isc_info_svc_user_dbpath was always returning an alias of the main security database.

fixed by A. Peshkov


() — Authentication plugin on the server could read garbage data from the client instead of the empty packet.

fixed by A. Peshkov


POSIX Only

() — Server could ignore KeepAlive settings (server connection/process is not killed when its client endpoint is lost).

fixed by A. Peshkov


Utilities

gbak

() — GBAK with service_mgr and `Win_Sspi_ authentication could raise an error in clumplet API.

fixed by A. Peshkov


() — Existing mapping rules were removed by the backup/restore cycle.

fixed by A. Peshkov


() — GBAK with PIPE to stdout: invalid content in the 'service_mgr' mode.

fixed by A. Peshkov


() — Wrong dependencies of stored function on view were created after backup/restore.

fixed by A. dos Santos Fernandes


() — CREATE DATABASE permission would disappear from security.db after the backup/restore cycle.

fixed by A. Peshkov


() — Creating backup to STDOUT using the service manager was broken.

fixed by A. Peshkov


() — GBAK doesn’t return the error code in some cases.

fixed by A. Peshkov


gfix

() — gfix -online normal did not raise an error when there was another SYSDBA-owned session open.

fixed by A. Peshkov


isql

() — Warnings were not always displayed in ISQL.

fixed by A. Peshkov


() — Metadata script extracted with ISQL from a database restored from a version 2.5 backup was invalid if some table has COMPUTED BY fields.

fixed by A. dos Santos Fernandes


nBackup

() — NBACKUP called via the Services API was locking the database file after error.

fixed by V. Khorsun


fbsvcmgr

() — fbsvcmgr could not decode information buffer returned by GFIX list limbo service.

fixed by A. Peshkov


Firebird 3.0.5 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.5 official release:

Core Engine

() — “Error reading data from the connection” could be raised sometimes when using a non-default authentication plugin.

fixed by A. Peshkov


() — The client library could leak memory when resolving the remote host name.

fixed by A. dos Santos Fernandes


() — Firebird’s internal timer could incorrectly reset the other existing timer entries.

fixed by V. Khorsun


() — Error “No current record for fetch operation” could be raised for queries containing an aggregated sub-select.

fixed by D. Yemanov


() — Generator pages were not encrypted.

fixed by A. Peshkov


() — Bugcheck could be raised when some PK\UK\FK constraint checks a record already marked as damaged.

fixed by V. Khorsun


() — Inconsistent behaviour of the NEW context variable was observed in AFTER UPDATE OR DELETE triggers.

fixed by I. Eremin


() — Error “Multiple maps found for …​” was raised in not appropriate case.

fixed by A. Peshkov


() — Inconsistent behavior was observed regarding visibility of the master record while inserting the detail record.

fixed by V. Khorsun


() — Different error messages (“Your user name and password are not defined” vs “Missing security context”) could be reported for a failed login depending on whether the Win_Sspi plugin is present in the configuration.

fixed by A. Peshkov


() — Error “Datatypes are not comparable in expression” could be raised for a COALESCE function used inside an expression.

fixed by A. dos Santos Fernandes


() — Connection could not see itself in MON$ATTACHMENTS when Windows domain/username (using Win_Sspi authentication) is exactly 31 bytes long.

fixed by D. Yemanov


() — BLOB fields could be suddenly set to NULLs during UPDATE after a table format change.

fixed by D. Yemanov


() — BLOBs were unnecessarily copied during UPDATE after a table format change.

fixed by D. Yemanov


() — CAST function used inside a union could be ignored.

fixed by A. dos Santos Fernandes


() — CREATE OR ALTER series of statements were ignoring the existing permissions.

fixed by A. Peshkov


() — The client library could leak memory if DSQL cursors are closed after the transaction commit.

fixed by V. Khorsun


() — Firebird server could return the op_accept packet instead of the expected op_accept_data one.

fixed by D. Kovalenko


() — The client library was attempting to open files not existent on the client host when connecting to a remove version 2.5 server.

fixed by A. Peshkov


() — Global temporary tables created with the ON COMMIT DELETE ROWS clause were not releasing their used space in the temporary file(s) at COMMIT RETAINING or ROLLBACK RETAINING.

fixed by V. Khorsun


() — Srp user manager could sporadically create user accounts that cannot pass the authentication.

fixed by A. Peshkov


() — The “creator” user name was empty in user trace sessions.

fixed by V. Khorsun


() — Error message was missing inside firebird.log after failure to create the audit log file.

fixed by V. Khorsun


() — Trace could not work correctly with quoted file names in the trace configurations.

fixed by V. Khorsun


() — Condition NULL IS [NOT] {FALSE | TRUE} was evaluated incorrectly (NULL was returned instead of the expected FALSE/TRUE).

fixed by A. dos Santos Fernandes


() — ROLE option was ignored when specified for a remote EXECUTE STATEMENT command.

fixed by A. Peshkov


() — Error “no permission for SELECT access” could be raised when a BLOB is used as input or output parameter of a stored procedure.

fixed by D. Starodubov


() — Slow performance could be noticed when executing SQL scripts as non-SYSDBA user.

fixed by A. Peshkov


() — Firebird 3 optimizer could choose a less efficient plan than the Firebird 2.5 optimizer.

fixed by D. Yemanov


() — Bugcheck could happen when a read-only database with non-zero linger is converted into a read-write mode.

fixed by V. Khorsun


() — Bugcheck 165 (“cannot find tip page”) could happen under load in Classic and SuperClassic.

fixed by V. Khorsun


() — CREATE MAPPING with non-ASCII user name could lead to the “Malformed string” error message.

fixed by A. Peshkov


() — Memory statistics ('used' / 'mapped' counters) was not always accurate.

fixed by A. Peshkov


() — Trace session could not be started if the 'database' section contains a pattern including curvy brackets.

fixed by A. Peshkov


() — Firebird server on Android had problems with NUMERIC values (scale was not reported properly).

fixed by A. Peshkov


() — Error “no permission for SELECT access” was raised when selecting a BLOB field from a stored procedure.

fixed by D. Starodubov


() — ORDER BY clause on compound index could prevent from using other available indices.

fixed by D. Yemanov


() — Lock folder inizialization was not multi-process safe.

fixed by A. Peshkov


() — Mapping created for the Win_Sspi plugin did not cause the Windows user group to be converted to a Firebird role.

fixed by A. Peshkov


() — Error “Invalid data type for negation” was raised for a query containing a negated parameter.

fixed by A. dos Santos Fernandes, D. Yemanov


() — Charset ID of the array element could be lost (zero was always reported) when preparing a query selecting from that array field.

fixed by D. Kovalenko


() — Compound index could not be used for filtering in some ORDER/GROUP BY queries.

fixed by D. Yemanov


() — Bugcheck could be raised after attempting to create a self-referencial FK that violates the existing data.

fixed by V. Khorsun


Server Crashes/Hang-ups

() — Server could crash while creating some complex stored procedure (containing multiple CTEs).

fixed by V. Khorsun, D. Yemanov


() — Server could hang while compiling a large package body containing a syntax error.

fixed by A. dos Santos Fernandes


() — Server could hang at the incoming attachment if it happened immediately after the trace session has been stopped.

fixed by A. Peshkov


() — Server crash could happen while using the computed index based on a complex expression.

fixed by A. Peshkov


() — Firebird could crash during concurrent operations with expression indices.

fixed by V. Khorsun


() — External engine trigger was causing the server to crash if the underlying table has computed field(s).

fixed by A. dos Santos Fernandes


() — Server could deadlock when some user attaches a bugchecked database.

fixed by A. Peshkov


() — The server could crash preparing a query having DISTINCT or ORDER BY together with a non-field expression in the select list.

fixed by D. Yemanov


() — The Firebird server could crash at the end of a gbak backup.

fixed by V. Khorsun


Builds

() — Firebird would not build on MacOS with the fresh toolchain installed, compilation failed while building the editline library.

fixed by A. Peshkov


() — Linking the binaries could fail due to missing libiconv_open reference if the libiconv library is installed.

fixed by A. Peshkov


() — gpre_boot was failing to link using CMake, reporting undefined reference dladdr and dlerror.

fixed by A. Peshkov


API/Remote Interface

() — Error “connection lost to database” could be raised when application creates multiple local attachments (using the XNET protocol) simultaneously.

fixed by V. Khorsun


() — Connection could remain not encrypted using some non-standard authentication plugins even if a correct key was provided.

fixed by A. Peshkov


() — Events processing was not working between Windows and MacOS hosts.

fixed by A. dos Santos Fernandes, V. Khorsun


POSIX Only

() — The first connection to any database was causing a small memory leak related to the binreloc library initialization.

fixed by A. Peshkov


() — Incorrect alignment was used for the BIGINT data type transferred from/to UDR libraries in 32-bit Linux builds.

fixed by A. dos Santos Fernandes


() — The binreloc library could fail to initialize properly with ld versions 2.31 and above, causing the Firebird binaries to wrongly detect its root directory.

fixed by R. Simakov


MacOS Only

() — An attempt to call the properly declared UDF was raising the “module name or entrypoint could not be found” error on MacOS Mojave (10.14.3).

fixed by P. Beach


Windows Only

() — Firebird could return incorrect (outdated) current time/timestamp after the time zone was changed at the operating system level.

fixed by V. Khorsun


() — VC runtime installer could reboot the system before the Firebird installation process finishes.

fixed by B. Walker, P. Reeves


Utilities

gbak

() — User-defined triggers declared for system tables were included into backups, and it is impossible to alter/drop them due to system tables protection in Firebird 3. Starting with version 3.0.5, gbak skips such triggers during both backup and restore operations.

fixed by A. Peshkov


() — gbak was reporting the authentication error “Your user name and password are not defined” when the “-fe(tch_password)” command switch is specified and gbak was running as a service.

fixed by A. Peshkov


() — Indices on computed fields were broken after a restore, with all keys being set to NULL.

fixed by D. Yemanov


() — Restoring a database with a computed index on a computed field was providing the wrong index and wrong query results.

fixed by D. Yemanov


gpre

() — Incorrect error “invalid request handle” was raised by the gpre-generated code when an embedded user does not have proper permissions for the table.

fixed by D. Yemanov


isql

() — Metadata script extracted using isql -x was invalid if some table contains a COMPUTED BY column.

fixed by A. dos Santos Fernandes


nBackup

 

fbsvcmgr

() — fbsvcmgr action_repair rpr_list_limbo_trans was not showing the expected list of 'in limbo' transactions.

fixed by A. Peshkov


fbtracemgr

() — Segmentation fault could happen in fbtracemgr while closing it using the Ctrl-C keystroke.

fixed by A. Peshkov


Firebird 3.0.4 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.4 official release:

Core Engine

() — Fixed some inconsistencies that could occur when using ALTER FUNCTION to replace a legacy UDF with a PSQL function having the same name.

fixed by A. dos Santos Fernandes


() — An attempt to create global mapping with a FROM field length greater than the SQL identifier length would fail.

fixed by A. Peshkov


() — Login attempts during engine shutdown would cause unnecessary delays in the shutdown process.

fixed by A. Peshkov


() — ROLE was not passed in EXECUTE STATEMENT …​ ON EXTERNAL.

fixed by V. Khorsun


() — The wrong events were released when a dead process was detected, i.e. valid events of a valid connection were released instead of the events of the broken connection.

fixed by A. Peshkov


() — An initial global mapping from the Srp plugin would not work.

fixed by A. Peshkov


() — Network server would ignore any error that occurred in the KeyHolderPlugin when establishing the initial callback with a client.

fixed by A. Peshkov


() — Unhandled: C0000005.ACCESS_VIOLATION could cause the server to crash. The issue was related with events processing and could happen when POST_EVENT ran from an ON DISCONNECT trigger. The same exception could occur if the application failed to cancel events when disconnecting.

fixed by V. Khorsun


() — Error reading/writing data to connection: appears to be related to CORE-5880 (above).

fixed by V. Khorsun


() — The system trigger RDB$TRIGGER9 checks that an object exists. However, existence checks for generators and exceptions were not implemented for checking when granting the USAGE privilege on generator and exception objects. As a result, it was possible to grant USAGE on a non-existent object.

fixed by R. Simakov


() — The error message returned from a primary key violation was returning “Malformed string” instead of the key value.

fixed by V. Khorsun


() — CREATE VIEW was issuing a faulty error “Implementation of text subtype 512 not located”.

fixed by A. dos Santos Fernandes


() — Firebird was not cancelling events when the client disconnected. See also CORE-5879 and CORE-5880, above.

fixed by V. Khorsun


() — Failures in a TRANSACTION START trigger were not being handled correctly.

fixed by V. Khorsun


() — A user was able to create a table that had a foreign key reference to another table to which he had no access privilege.

fixed by R. Simakov


() — Inconsistent results when working with GLOBAL TEMPORARY TABLE ON COMMIT PRESERVE ROWS.

fixed by V. Khorsun


() — The client would get the error SQLSTATE = 28000 ("user/password not defined") if WireCrypt = Disabled was used on the client side.

fixed by A. Peshkov


() — An attachment could persist after it was deleted from MON$ATTACHMENTS.

fixed by V. Khorsun


() — Fixed several legacy errors and anomalies in the implementation of the REVOKE operator.

fixed by R. Simakov


(CORE-5801, link not available) — Unauthorized BLOB access vulnerability was fixed.

fixed by D. Starodubov


() — The gstat utility could produce a faulty report concerning the presence of some non-encrypted pages in a database.

fixed by A. Peshkov


() — The error returned from DbCryptPlugin::setKey() (“Missing correct crypt key”) was not helpful for diagnostics.

fixed by A. Peshkov


() — The ODS 12.0 for Windows and Linux-x64 builds differs from Linux-x86. The fix results in a minor ODS change, from 12.0 to 12.2, for some but not all platforms. This has certain implications for compatibility, so please read these notes.

fixed by A. Peshkov


() — User with DROP DATABASE privilege could not drop the database.

fixed by R. Simakov


() — Security Patch: Replacement of use of SHA-1 in the SRP Client Proof with SHA-256. See V. 3.0.4 SRP Security Patch in the Security chapter of these notes.

fixed by A. Whyman, A. Peshkov


() — EXECUTE STATEMENT was ignoring the text of the SQL query string after a comment of the double-hyphen form (‘--’).

The bug was solved by ensuring that the parser properly handled a newline symbol.

However, when EXECUTE STATEMENT gets the SQL query text from a binary BLOB variable, the engine uses a special blob filter (filter_text) to convert the binary blob to text. For some reason, this blob filter treats \n (new-line) as a segment boundary and removes it from the output. In a case where the line-endings in the binary BLOB were stored as \r\n, the SQL query text received by EXECUTE STATEMENT thus loses the newline directive, and the PSQL parser is left with no way to detect the end of the single-line comment.

The BLOB filter filter_text will not be changed to cater for the edge case of applying a binary BLOB to EXECUTE STATEMENT. The solution is to avoid using a binary BLOB to store data that is to be used as text.

fixed by V. Khorsun


() — The server would hang when a client tried to send DB encryption keys that were too long to fit in the TCP buffer that FB allocates for sending/receiving messages, TcpRemoteBufferSize. The fix is to drop the attachment as soon as the server detects a fragmented packet in the async receiver.

fixed by A. Peshkov


() — Error “Input parameter mismatch” could be thrown after altering a function to replace an external function with a PSQL function.

fixed by V. Khorsun


() — A PSQL (explicit) cursor could not see an inserted record, because cursor stability based on savepoints and the Undo log could not be applied to explicit cursors. A different approach was taken to implement stability for explicit cursors.

fixed by V. Khorsun


() — Wrong transaction number in system table RDB$PAGES relation could cause infinite recursion in the engine and result in a segfault.

fixed by A. Peshkov


() — Deadlock with events.

fixed by A. Peshkov


() — Granting a privilege on a non-existent object produced no error or warning.

fixed by R. Simakov


() — Since Firebird does not have explicit privileges for triggers, ALTER TRIGGER would check the privileges for ALTER DATABASE, which was correct for database triggers but wrong for table-level triggers.

fixed by R. Simakov


() — The parser allowed the use of WITH GRANT OPTION when granting the EXECUTE privilege to functions and packages. This was wrong.

fixed by R. Simakov


() — A user could grant USAGE privileges to self.

fixed by R. Simakov


() — A conversion error could be thrown when both GROUP BY/ORDER BY expressions and WHERE expressions contained literals.

fixed by A. dos Santos Fernandes


() — Firebird on Windows was leaking handles.

fixed by V. Khorsun


() — Genuine errors during connection to the security database were hidden by the Srp user manager.

fixed by A. Peshkov


() — An old pitfall: if an index contained garbage and automatic GC was disabled, it would not be possible to create unique indexes or PRIMARY, UNIQUE or FOREIGN KEY constraints.

fixed by V. Khorsun


Server Crashes/Hang-ups

() — A Classic process could crash when the client had disconnected.

fixed by V. Khorsun


() — The encryption interface could crash the Firebird process when working on big database file.

fixed by A. Peshkov


() — The server could hang for 60/120 seconds if the client disconnected during the database encryption key transfer callback.

fixed by A. Peshkov


() — Server could crash when trying to recreate a table that is in use for a DML operation.

fixed by D. Yemanov


() — SuperClassic server would hang under load.

fixed by V. Khorsun


API/Remote Interface

() — Connection could hang if there was no activity for 60 seconds.

fixed by V. Khorsun


() — A client connection could hang when working with events under high load.

fixed by A. Peshkov


Builds

() — An alignment error would occur on x86_64 when using clang to compile Firebird, due to an optimization in that environment. It was known to affect builds for MacOSX and LibreOffice Base.

fixed by A. Peshkov


() — Segmentation fault during install on Linux.

fixed by A. Peshkov


() — Firebird would not compile with ICU 60 or higher.

fixed by A. Peshkov


() — install.sh would fail if the -path argument contained 'firebird'.

fixed by A. Peshkov


() — Database crypt plug-in sample in Pascal was broken.

fixed by A. Peshkov


() — Missing directives in Firebird.pas.

fixed by A. Whyman, A. Peshkov


() — Installation of Firebird 3.0.3 on SLES 12 SP3 would fail with “Could not find acceptable ICU library”.

fixed by A. Peshkov


Utilities

gbak

() — Using gbak as a service could cause memory leakage.

fixed by A. Peshkov


() — gbak could crash the database engine when mixing a database alias with the full database path.

fixed by A. Peshkov


() — Metadata triggers for some object types (views, exceptions, roles, indexes, domains) were lost in the backup-restore process.

fixed by A. dos Santos Fernandes


() — The server process could crash while restoring a database.

fixed by V. Khorsun


() — SQL privileges error during database restore.

fixed by A. Peshkov


gfix

() — Database validation would report false errors “Record XXX has bad transaction” and\or “Record XXX is wrong length” when the record’s transaction number exceeded 232.

fixed by V. Khorsun


() — Sweep was failing to increment the OIT value past 232, when Next Transaction value was already higher than 232.

fixed by V. Khorsun


gsec

() — The gsec utility was not releasing the user manager plug-in.

fixed by A. Peshkov


gstat

() — Output of gstat from an encrypted database was not user-friendly.

fixed by A. Peshkov


isql

() — The SHOW DATABASE command in isql was returning zero for statistics with a value exceeding the range of 32-bit integer.

fixed by V. Khorsun


() — An incorrect error message was returned in isql when trying to create a database when logged in with the wrong password.

fixed by A. Peshkov


() — Invalid parameters for gds transaction in isql would cause it to hang when trying to show various system objects while another attachment had uncommitted changes to those objects.

fixed by A. Peshkov


nBackup

() — nbackup would stop working after 32K iterations of backups.

fixed by A. Peshkov


Firebird 3.0.3 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.3 official release:

Core Engine

() — An additional keyholder could open unauthorized connections to an encrypted database, creating a vulnerability.

fixed by A. Peshkov


() — Sweep in Classic mode could run too slowly when there was a big load on the server.

fixed by V. Khorsun


() — A field alias could disappear in a complex query

fixed by A. dos Santos Fernandes


() — The function POSITION did not consider the collation for BLOBs.

fixed by V. Khorsun


() — Duplicate values could occur in columns with the UNIQUE constraint.

fixed by V. Khorsun


() — Error “no current record for fetch operation” was raised while deleting a record from MON$ATTACHMENTS using ORDER BY clause

fixed by D. Yemanov


() — An access violation could occur when an external statement was executed and the local transaction was rolled back.

fixed by V. Khorsun


() — isc_vax_integer() and isc_portable_integer() did not work properly with short negative numbers.

fixed by D. Sibiryakov


() — Unique constraint would not work in an encrypted database on first command.

fixed by A. Peshkov


() — Message “CTE 'X' has cyclic dependencies” would appear when ‘X’ was the alias for a result set and there was a previous CTE part in the query with the same name ‘X’.

fixed by V. Khorsun


() — The optimizer was generating a bad plan under certain conditions.

fixed by D. Yemanov


() — isc_info_sql_relation_alias was reporting incorrect aliases for CTE or nested queries.

fixed by V. Khorsun


() — A wrong transaction could be passed to the external engine.

fixed by A. dos Santos Fernades


() — Message “Operating system call munmap failed. Error code 12” could appear in firebird.log under heavy load in Classic or Superclassic.

fixed by A. Peshkov


() — Wrong result with index on case-insensitive collation using NUMERIC-SORT.

fixed by A. dos Santos Fernandes


() — Some portions of the pages of second-level blobs were not released when dropping relations.

fixed by D. Logashov


() — When running with a plug-in to access a version 2.5 database, the engine would incorrectly check for existing MAP if a pre-FB3 security database was used.

fixed by A. Peshkov


() — An invalid blob id would be reported when adding a new blob column of type text and updating another field in the same operation.

fixed by V.Khorsun


() — Error “Block size exceeds implementation restriction” could occur while inner joining large datasets with a long key using the HASH JOIN plan.

fixed by D. Yemanov


() — DbInfo interface was not being passed to ancillary instances of a plug-in.

fixed by A. Peshkov


() — Signatures of packaged functions were not being checked for mismatch with the [NOT] DETERMINISTIC attribute.

fixed by A. dos Santos Fernandes


() — Queries containing a WITH LOCK clause could trigger a bugcheck.

fixed by R. Simakov


() — Direct system table modifications were not completely prohibited.

fixed by A. dos Santos Fernandes


() — Error handling for SELECT WITH LOCK broke compatibility with Firebird 2.5.

fixed by V. Khorsun


() — A database could not be encrypted if the DatabaseAccess was set to None.

fixed by A. Peshkov


() — A computed decimal field in a view could be stored with the wrong RDB$FIELD_PRECISION value.

fixed by A. dos Santos Fernandes


() — Errors could occur when building or running Firebird on hardware that did not have SSE enabled.

fixed by A. Peshkov


() — Database-level triggers related to TRANSACTION events (start/commit/rollback) were not taking their POSITION index into account.

fixed by A. dos Santos Fernandes


() — The value of RDB$FIELD_SUB_TYPE in RDB$FUNCTION_ARGUMENTS would be garbage after altering the function.

fixed by A. dos Santos Fernandes


() — Internal Firebird consistency check (limbo impossible (184), file: vio.cpp line: 2379)

fixed by V. Khorsun


() — External routines were not receiving default values for their output parameters.

fixed by A. dos Santos Fernandes


() — External routines could receive parameters with incorrect length or data type.

fixed by A. dos Santos Fernandes


() — Classic server could raise the error “Global mapping memory overflow” periodically and require restarting to accept new connections.

fixed by A. Peshknov


() — Memory leak: Firebird did not release memory after load tests.

fixed by V. Khorsun


() — Found and fixed an elusive bug whereby an unexpected memory overwrite could cause the in-memory image of a random database page to be corrupted.

fixed by V. Khorsun, with testing assistance from D. Kovalenko


() — With OR or IN predicates for RDB$DBKEY, the optimizer failed to use an INDEX plan when it should have.

fixed by D. Yemanov


() — Bugcheck: page N, page type 5 lock conversion denied (215).

fixed by V. Khorsun


() — Recursive EXECUTE STATEMENT was not working properly.

fixed by V. Khorsun


() — Cyclic dependency errors in CTE, involving aliases.

fixed by V. Khorsun


() — Records were left in RDB$PAGES after rollback of CREATE TABLE statement.

fixed by R. Simakov & D. Yemanov


Server Crashes/Hang-ups

() — The connection to the server could hang when working with encrypted databases over network protocol other than TCP.

fixed by A. Peshkov


() — Firebird 3 engine could crash when restoring from a backup made by Firebird 2.5.

fixed by A. dos Santos Fernandes


() — Begin and end of physical backup in the same transaction could crash the engine.

fixed by V. Khorsun


() — Trace config with misplaced ‘{’ could lead to an engine crash.

fixed by A. dos Santos Fernandes


() — The engine could hang during transaction start, soon after certain errors.

fixed by A. Peshkov


() — Firebird could crash when a UDF was loaded and unloaded frequently.

fixed by A. Peshkov


() — The server could crash while compiling a stored procedure with nested references to query-based computed fields.

fixed by D. Yemanov


() — The engine could crash when a database contained a database-level trigger.

fixed by A. dos Santos Fernandes


() — Firebird service could crash when stopping.

fixed by V. Khorsun


Builds

() — Intermediate build files for plugins examples were mixed up.

fixed by A. Peshkov


API/Remote Interface

() — Information items isc_info_length and fb_info_crypt_state had the same code.

Software using the fb_info_crypt_state constant should be recompiled to make it work with version 3.0.3 and later.

fixed by A. Peshkov


() — Firebird 3.0.2 would allow any protocol version >= 10 to connect, including InterBase protocol 14

fixed by M. Rotteveel


() — isc_add_user / isc_modify_user could produce an unusable user record.

fixed by A. Peshkov


() — A race condition could arise between event notification and event registration.

fixed by V. Khorsun


() — The engine could send a status vector with more than 20 items to old clients and cause a disconnection.

fixed by R. Simakov


POSIX Only

() — Dropping a procedure on the Classic server could cause a segmentation fault.

fixed by A. Peshkov


() — An old bug was fixed, whereby backslashes in path strings for use on POSIX file systems were not always converted to forward slashes.

fixed by A. Peshkov


Utilities

gbak

() — gbak restore with a large number of small blobs was very slow using Linux Classic.

fixed by R. Simakov


() — During restore, the shadow file could not be created.

fixed by A. Peshkov


() — A request synchronization error could occur in gbak during a restore.

fixed by V. Khorsun


() — The server could crash while restoring from a backup if the shadow file already existed.

fixed by D. Yemanov


isql

() — isql would incorrectly pad UNICODE_FSS/UTF8 columns when they used a collation.

fixed by A. dos Santos Fernandes


() — Negative infinity (double) was being shown incorrectly without the sign in isql.

fixed by A. dos Santos Fernandes


nBackup

() — Superserver could hang when changing physical backup state under high load

fixed by V. Khorsun


() — Alternate executions of an isql command from the command line in embedded mode would be 5 to 10 times slower, i.e. fast-slow-fast-slow…​

fixed by V. Khorsun


User Management

() — Problem migrating users to version 3.0.

fixed by A. Peshkov


Firebird 3.0.2 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.2 official release:

Core Engine

() — Diagnostics form gstat were unclear when a damaged page in DB file appeared encrypted.

fixed by A. Peshkov


() — It was not possible to create a connection for events processing.

fixed by V. Khorsun


() — Creating SRP SYSDBA with explicit admin (-admin yes in gsec or grant admin role in create user) was creating two SYSDBA accounts.

fixed by A. Peshkov


() — Performance for NULLs filtering inside a navigational index scan was poor.

fixed by D. Yemanov


() — The setting “Restrict UDF” for configuration parameter UdfAccess was ineffective because of fbudf.so being dynamically linked against libc.

fixed by A. Peshkov, A. dos Santos Fernandes


() — Problems with dbcrypt key transfer.

fixed by A. Peshkov


() — Trace INCLUDE_FILTER with [[:WHITESPACE:]]+ would not work when a statement was issued that contained a newline character.

fixed by V. Khorsun


() — An access violation would occur in fbclient when reading a BLOB stored in incompatible encoding.

fixed by V. Khorsun


() — Bugcheck 167 (invalid SEND request).

fixed by D. Yemanov


() — When attempting to migrate a database from Firebird 2.5.2 to 3.0.1, gbak restore would always crash at the end.

fixed by D. Yemanov


() — A segmentation fault would occur when the engine’s dynamic library was unloaded right after closing worker threads (GC and/or cache writer).

fixed by A. Peshkov


() — Bugcheck oocurred when creating a view on a table that has a column with character set NONE, database default charset is UTF8, and the default collation is UNICODE.

fixed by V. Khorsun


() — EXECUTE STATEMENT <e> when <e> starts with ‘--’ (comment) issues “Unexpected end of command” error with a wrong column number.

fixed by A. dos Santos Fernandes


() — Classic Server suffered a performance regression: too much time was being spent during context switches.

fixed by D. Yemanov


() — The error “Decompression overran buffer” could occur after rollback. It affected all platforms but would appear particularly on Windows Superserver, when GCPolicy was set to Combined or Cooperative and CpuAffinityMask was configured to more than a single core.

fixed by V. Khorsun


() — Performance degradation in FB 3.0.2 compared to FB 2.5.7.

fixed by D. Yemanov


() — Error restoring on FB 3.0 from FB 2.5 backup: bugcheck 221 (cannot remap).

fixed by D. Yemanov


() — Dependencies were not stored when certain types of construction were used in subroutines.

fixed by A. dos Santos Fernandes


() — Optimization was poor for some operations with views containing subqueries.

fixed by D. Yemanov


() — BUGCHECK 179 (decompression overran buffer) or an unexpected lock conflict could happen during back-out of record versions.

fixed by D. Yemanov


() — isql would crash with corrupt memory when statement length exceeded 10 MB.

fixed by A. dos Santos Fernandes


() — Plug-in manager was failing to provide the appropriate config information when the name of the configuration file was set in plugins.conf.

fixed by A. Peshkov


() — Dependencies in PSQL packages were not being recognised.

fixed by A. dos Santos Fernandes


() — An SQL statement longer than 10MB could be processed (truncated) incorrectly.

fixed by D. Yemanov


() — Regression: A (select from view with nested view) query could not be executed.

fixed by D. Yemanov


() — Regression: a client application could hang or throw an access violation on selecting from multiple threads using one embedded connection.

fixed by A. Peshkov


() — Regression: A BOOLEAN state as a parameter for a search condition was incorrectly rejected as invalid.

fixed by A. dos Santos Fernandes


() — Regression: A CASE expression with more than 255 conditions was being rejected.

fixed by A. dos Santos Fernandes


() — An assertion was exhibited on Windows in CachedMasterInterface::set.

fixed by V. Khorsun


() — UDFs declared with large varchars were taking excessive time to execute.

fixed by A. dos Santos Fernandes


() — BETWEEN operand/clause was not supported for COMPUTED BY columns — “feature is not supported”. Now it is.

fixed by D. Yemanov


Server Crashes/Hang-ups

() — An old bug could occasionally cause a segmentation fault in Classic when the structure of a table with triggers was modified (DDL) while another attachment was preparing a request to work with that table. In rare cases, pointers to the old triggers were cached by the request compiler, and the fault would occur the next time they were used.

fixed by A. Peshkov


() — A load test was hanging the server.

fixed by A. Peshkov


() — The engine could hang if an automatic sweep was starting during a database shutdown.

fixed by V. Khorsun


() —  (Old bug) Circular dependencies between computed fields would crash the engine.

fixed by A. dos Santos Fernandes


Builds

() — Issues with the function snprintf (in autoconfing_msvc.h) and user-defined literals in VS 2015 (MSVC 14).

fixed by V. Khorsun


API

() — New API - MessageMetadata's getScale() was returning INTEGER, while MetadataBuilder’s setScale() expected unsigned.

fixed by A. Peshkov


POSIX Only

() — Classic Server was limited to 64 connections when using systemd init scripts

fixed by A. Peshkov


() — A database on an NFS share was unavailable when accessed using the mount point path.

fixed by A. Peshkov


() — Connections were failing due to dead NFS mount points.

fixed by A. Peshkov


Windows Only

() — Trace was duplicating ascii_char(13) in its output on Windows.

fixed by V. Khorsun


Firebird 3.0.1 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.1 official release:

Core Engine

() — XpbBuilder was failing to create a new TPB.

fixed by A. Peshkov


() — LEFT JOIN would incorrectly push a UDF into the inner stream, causing wrong results.

fixed by D. Yemanov


() — A user-defined collation failed to work with BLOBs.

fixed by A. dos Santos Fernandes


() — An assertion could occur in createDatabase() when doing the overwrite check.

fixed by A. Peshkov


() — A subquery in the insert list expression was ignoring changes made earlier in the same executable block.

fixed by V. Khorsun


() — A database could suffer partial corruption in the “use all space” (no-reserve) mode.

fixed by V. Khorsun & D. Yemanov


() — BUGCHECK 165 (cannot find tip page)

fixed by V. Khorsun


() — Cascade deletion in a self-referencing table could raise the error “no current record for fetch operation”.

fixed by V. Khorsun


() — A “Data type unknown” error was occurring with the LIST operation.

fixed by A. dos Santos Fernandes


() — Random “invalid transaction handle” errors were being returned from the engine.

fixed by D. Yemanov


() — ON UPDATE CASCADE was failing for a self-referencing Foreign Key.

fixed by V. Khorsun


() — REGRESSION: Performance was degraded when bulk inserting into a table with indices.

fixed by V. Khorsun


() — Trying to encrypt a database in the absence of the appropriate key could corrupt it.

fixed by A. Peshkov


() — A small memory leak could occur when an empty result set was returned.

fixed by V. Khorsun


() — Granting of privileges to views was broken.

fixed by A. Peshkov


() — Parameters with multi-byte character sets were being allowed to exceed the length limit for VARCHAR fields.

fixed by A. dos Santos Fernandes


() — Expression index could become inconsistent if CREATE INDEX was interrupted after b-tree creation but before committing.

fixed by D. Yemanov


() — Nested OR conditions could lead to incorrect results.

fixed by D. Yemanov


() — Consistency between roles and privileges in GRANT syntax needed improving in accordance with the SQL standard.

fixed by A. Peshkov


() — An attempt to execute a SELECT statement using isc_dsql_execute_immediate() would produce the wrong error message.

fixed by A. Peshkov


() — Affected rows were not being counted for some update operations with views.

fixed by D. Yemanov


() — IN/ANY/ALL predicates could cause sub-optimal execution (late filtering) of joins.

fixed by D. Yemanov


() — Unaligned memory access in MET_format() would cause segfault on big-endian machines.

fixed by James Clarke


() — 32-bit locks on big-endian systems would lead to deadlock.

fixed by James Clarke


() — EXECUTE STATEMENT would throw a BLR error if the number of output parameters exceeded 256.

fixed by D. Yemanov


() — An incorrect result set (missing records) could be returned by an ORDER plan query navigating on a descending index.

fixed by D. Yemanov


() — Authentication was ending with first plug-in that had the user but would fail there, instead of continuing with the next plug-in.

fixed by A. Peshkov


() — SELECT WITH LOCK could raise unexpected update conflict errors under concurrent load.

fixed by D. Yemanov


() — A database could get decrypted after changing a couple of bytes in the database header without 'agreement' from the crypt plug-in.

fixed by A. Peshkov


() — POST_EVENT was not working with the Fb 3.0 server and client.

fixed by V. Khorsun


() — The operator IS exhibited a precedence problem.

fixed by A. dos Santos Fernandes


() — REGRESSION: line/column numbering in error messages were being garbled if alias.name syntax was used in the offending statement.

fixed by D. Yemanov


() — The error “no current record to fetch” was thrown if some record was to be deleted both by the statement itself and by some trigger fired during statement execution.

fixed by V. Khorsun


() — An expression index could be created while inserts into the table were in progress.

fixed by V. Khorsun


Server Crashes

() — Access violation from certain UDFs would cause the server to crash.

fixed by V. Khorsun


API/Remote Interface

() — Race conditions could arise when initializing wire compression.

fixed by A. Peshkov


() — A network protocol error was exhibited when performing a callback to a client for the database crypt key.

fixed by A. Peshkov


() — Some SPB parameters were failing in FB 3.0.

fixed by A. Peshkov


() — Transaction id tags for the Services API did not support new 48-bit transaction ids.

fixed by D. Yemanov


() — The Services API would not work with non-ASCII database names.

fixed by D. Sibiryakov


Utilities

fbsvcmgr

() — fbsvcmgr was not producing any error when it attempted to shut down a database without a specified timeout (prp_force_shutdown N).

fixed by A. Peshkov


gbak

() — Restore could hang if the database contained more than 4 billion records.

fixed by D. Yemanov


gfix

() — Validation could read after the end-of-file when operating on a multi-file database.

fixed by V. Khorsun


isql

() — The BLOBVIEW command was causing isql to hang.

fixed by A. Peshkov


() — isql would leak memory when SHOW GRANTS was used on a new, empty database.

fixed by A. Peshkov


() — isql -[e]X[tract_metadata] output was omitting double quotes for the COLLATE <collation> clause of a CREATE DOMAIN statement when <collation> was for any non-ascii character set.

fixed by A. dos Santos Fernandes


() — isql -[e]X[tract_metadata] was not exporting explicitly defined names for NOT NULL constraints for Dialect 3 databases.

fixed by D. Yemanov


() — isql -[e]X[tract_metadata] could crash while exporting an exception with message text length longer than 127 bytes.

fixed by D. Yemanov


() — isql -[e]X[tract_metadata] could generate invalid GRANT USAGE statements for domains.

fixed by D. Yemanov


Monitoring and Trace

() — A string truncation error would occur while selecting from the MON$ tables if some user-defined context variable was longer than 255 bytes.

fixed by D. Yemanov


() — A trace session could fully utilize a single CPU core.

fixed by A. Peshkov


nBackup

() — A database located on a raw device could not be unlocked from an nBackup-locked state.

fixed by A. Peshkov


Builds

() — Databases created during the process of the Windows builds were owned by a Windows administrator, not SYSDBA.

fixed by V. Khorsun


() — Firebird failed to build with USE_VALGRIND

fixed by A. Peshkov


POSIX-Specific

() — A segfault could occur when op_que_events was used on a port that was not prepared for events processing.

fixed by A. Peshkov


() — A segfault could occur if an attachment was closed while requests were still open.

fixed by A. Peshkov


() — Incorrect use of the system call readlink().

fixed by A. Peshkov


Firebird 3.0.0 Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.0 official release:

Core Engine

() — An assertion would occur in the cloop dispatcher when trying to save exception information in the status interface.

fixed by A. Peshkov


() — Failure to find a record on re-fetch would produce inconsistent effects on subsequent statement execution.

fixed by D. Yemanov


() — Unique key violation on a BOOLEAN column was returning an erroneous message.

fixed by A. dos Santos Fernandes


() — HAVING COUNT(*) NOT IN ( <Q> ) would erroneously exclude a record from the outer result set. (<Q> represents a result set without nulls)

fixed by A. dos Santos Fernandes


() — It was possible to create a unique index on non-unique data.

fixed by V. Khorsun


() — Regression: the engine could throw transliteration errors when running from a non-ASCII system path.

fixed by A. dos Santos Fernandes


() — The server could go into an infinite loop when data was being sent from a client to the server in small packets.

fixed by A. Peshkov


() — Regression: LEFT JOIN was incorrectly pushing COALESCE into the inner stream, causing wrong results.

fixed by D. Yemanov


() — CREATE TRIGGER was failing with the error “Ambiguous field name between table B and table A”.

fixed by A. dos Santos Fernandes


() — Join order was sub-optimal if one table had a selective predicate and MIN was calculated for the other one.

fixed by D. Yemanov


() — Deadlock could occur when a database was being encrypted or decrypted under high parallel load.

fixed by A. Peshkov


() — ALTER SEQUENCE RESTART WITH was failing to change the initial value.

fixed by D. Yemanov


() — Minor regression: wrong error message was being returned on an attempt to compile a view with both a subquery and the WITH CHECK OPTION clause in its definition — an illegal combination.

fixed by A. dos Santos Fernandes


() — The optimizer could avoid using an expression index if the query was running in a different connection character set to the one that was in effect when the index was created.

fixed by D. Yemanov


() — Internal Firebird consistency check (cannot find tip page (165), file: tra.cpp line: 2375) would occur if too many transactions were created for one connection to a read-only database. This was a very old bug, dating back to InterBase 6.

fixed by V. Khorsun


Server Crashes

() — Regression: The server would crash when aggregate functions were used together with the NOT IN predicate.

fixed by A. dos Santos Fernandes


Utilities

gbak

() — gbak restore could fail when there was some SQL function accessing a table while the switch -O(NE_AT_A_TIME) was in use.

fixed by D. Yemanov


gfix

() — An attempt to set the number of page buffers to a non-supported value would return the wrong error message.

fixed by V. Khorsun


isql

() — An invalid COMPUTED BY definition was being generated by the 'extract metadata' operation.

fixed by A. dos Santos Fernandes


Monitoring

() — Database-level statistics (MON$STAT_GROUP = 0) could contain multiple rows under concurrent load.

fixed by D. Yemanov


User Management

() — The codes of operation for user management plug-ins were missing from the public API.

fixed by A. Peshkov


() — Users with the same name can co-exist in the security database if they are created by different plug-ins. However, tags or attributes for a user created using one plug-in were being displayed in the SEC$ virtual tables for the same-name user created by the other plug-in.

fixed by A. Peshkov


() — [CREATE OR] ALTER USER statement would fail if the PASSWORD parameter was present and was not specified immediately after the USER parameter.

fixed by A. Peshkov


Builds

() — The build would fail if HAVE_MMAP was not defined.

fixed by A. Peshkov


() — Root rights were unnecessarily required to build a Linux package.

fixed by A. Peshkov


() — The samples of dbcrypt-related plug-ins contained references to internal files.

fixed by A. Peshkov


Firebird 3.0 Release Candidate 2: Bug Fixes and Minor Improvements

The following tweaks and bug fixes were reported prior to version 3.0.0 release candidate 2:

Minor Improvements

() — NEW: SECURITY2.FDB → SECURITY3.FDB upgrade feature

Implemented by A. Peshkov


() — NEW: An option was added to the API function isc_database_info() to return the number of free pages in a database.

Implemented by V. Khorsun


() — IMPROVEMENT: The error messages for a malformed SQDLA were improved.

Implemented by D. Sibiryakov


() — IMPROVEMENT: Improved control over database crypt keys, see Encryption Tasks.

Implemented by A. Peshkov


() — IMPROVEMENT: A method was provided to set the correct offsets in user-implemented IMessageMetadata.

Implemented by A. Peshkov


() — IMPROVEMENT: In isql, SHOW DATABASE includes the encryption status of the database, as does gstat -h[eader].

Implemented by A. Peshkov


() — IMPROVEMENT: In isql, SHOW DATABASE includes the numbers of used and free pages in the database.

Implemented by V. Khorsun


() — IMPROVEMENT: The remote port number was added to the monitoring table MON$ATTACHMENTS.

Implemented by V. Khorsun


() — IMPROVEMENT: Validation and repair of lost data pages have been improved.

Implemented by R. Simakov, V. Khorsun


() — IMPROVEMENT: The IP address of the remote host is now appended to error messages in firebird.log for TCP connections.

Implemented by V. Khorsun


Bugs

Core Engine

() — A false primary or foreign key violation could be reported when an attachment used the isc_dpb_no_garbage_collect flag.

fixed by V. Khorsun


() — SELECTs from an external data source would become slow.

fixed by V. Khorsun, A. Peshkov


() — COMPUTED BY expressions were not being converted to their data type inside the engine.

fixed by A. dos Santos Fernandes


() — ALTER COLUMN was not changing the type of a computed column where it should.

fixed by A. dos Santos Fernandes


() — The server was not validating the correctness of the user/password pair provided in the EXECUTE STATEMENT operation.

fixed by A. Peshkov


() — The “Invalid BLOB ID” error was being thrown unexpectedly after a large VARCHAR column was changed to text BLOB.

fixed by V. Khorsun


() — ON DISCONNECT trigger with EXECUTE STATEMENT was not working and would lead to a server crash when it was recreated.

fixed by V. Khorsun


() — Function CHAR_TO_UUID on an indexed column would throw the error “expression evaluation not supported”.

fixed by A. dos Santos Fernandes


() — CREATE VIEW that selected from a system table would fail, despite all grants being available.

fixed by A. Peshkov


() — The write lock on the database file was being cleared when retrieving the database statistics via the Services API.

fixed by A. Peshkov


() — Byte-length for view columns was being calculated incorrectly.

fixed by A. dos Santos Fernandes


() — mapUser() routine would establish a secondary attachment using the wrong provider.

fixed by A. Peshkov


() — Connecting to a service with an invalid service name would yield an incorrect error message.

fixed by A. Peshkov


() — A delay of at least 5 seconds could occur on disconnect if the disconnect followed closely on an Event Manager initialization.

fixed by V. Khorsun


() — The file fb_user_mapping could be left behind in the Firebird lock directory after a correct shutdown of all Firebird worker processes.

fixed by V. Khorsun


() — Connection to a database with its own security database was not possible in the absence of security3.fdb.

fixed by A. Peshkov


() — Firebird authentication would give up after failing to establish authentication via the first common plugin.

fixed by A. Peshkov


() — An ORDER BY clause on the key columns of a compound index could disable usage of other indices.

fixed by D. Yemanov


() — Non-indexed predicates might not have been applied immediately after retrieval when tables were being joined.

fixed by D. Yemanov


() — Interrupt of aux connection during TCP setup phase would cause a server exit due to an unhandled exception.

fixed by A. Peshkov


() — Interrupt of aux connection during TCP setup phase was producing unclear error messages in firebird.log.

fixed by A. Peshkov


() — Races would occur when two threads tried simultaneously to establish the first embedded connection to the engine.

fixed by A. Peshkov


() — The Keyholder plug-in was being loaded on connection to a database but not on creating a database.

fixed by A. Peshkov


() — The REVOKE operator could modify the privileges granted to system tables at database creation time.

fixed by A. Peshkov


() — The SEC$USERS table became unavailable when an error occurred in any configured UserManager plug-in.

fixed by A. Peshkov


() — Real errors during connection to a security database were being hidden by the Srp user manager.

fixed by A. Peshkov


() — Retrieving the current user name when using Windows trusted authentication and an OS user name containing non-ASCII characters would cause an error.

fixed by A. Peshkov, D. Yemanov


() — Transactions with isc_tpb_autocommit could hang the server.

fixed by D. Yemanov


() — Dependencies with views were exhibiting a problem.

fixed by A. dos Santos Fernandes


Server Crashes

() — SET AUTODDL OFF and sequence of CREATE + ALTER + DROP VIEW would lead to a server crash.

fixed by A. dos Santos Fernandes


() — A database shutdown could cause a server crash if multiple attachments were running EXECUTE STATEMENT.

fixed by V. Khorsun


() — The server could crash during garbage collection when a DELETE is executed after adding a new referencing column.

fixed by D. Yemanov


() — The server could crash when an event was posted from an ON DISCONNECT trigger.

fixed by V. Khorsun


() — The server could crash during a select from a procedure if an exception was thrown while backing out the current savepoint.

fixed by D. Yemanov


Database Encryption

() — Deadlocks could occur when a database was initially encrypted.

fixed by A. Peshkov


() — Pages were being encrypted for each shadow separately.

fixed by A. Peshkov


() — A database could be corrupted after ALTER DATABASE ENCRYPT/DECRYPT was issued multiple times per database open time.

fixed by A. Peshkov


() — Trying to access a database with an invalid crypt key would cause a deadlock.

fixed by A. Peshkov


() — Databases were not being decrypted completely.

fixed by A. Peshkov


API/Remote Interface

() — fbclient.dll API calls were (wrongly) modifying the FPU control word.

fixed by V. Khorsun


() — New connections were being blocked as a consequence of sweeps of the security database taking too long.

fixed by V. Khorsun


() — The version 3 client could not create a database on a version 2.5 server.

fixed by A. Peshkov


() — A crash could occur in fbclient when a multi-threaded application was using the XNET protocol.

fixed by V. Khorsun


() — Neither client nor server could close the connection after failed authentication.

fixed by V. Khorsun


Utilities

gbak

() — Restore was slow when the database contained many small tables with indices.

fixed by V. Khorsun


gstat

() — A segmentation fault could occur while gstat -h was closing.

fixed by A. Peshkov


isql

() — The EXTRACT command would lose the data types of COMPUTED BY columns.

fixed by A. dos Santos Fernandes


() — Metadata extraction (isql -X): A CREATE PROCEDURE/FUNCTION statement would retain the TYPE OF reference to a table column that no longer existed, if the procedure or function had a parameter of that type declared at the time it was created.

fixed by A. dos Santos Fernandes


() — Plan output was being unexpectedly truncated after a query was simplified by shortening it.

fixed by D. Yemanov


Monitoring

() — Content of MON$EXPLAINED_PLAN in MON$STATEMENTS was being truncated if it exceeded the 32KB limit.

fixed by D. Yemanov


() — The MON$ tables became unavailable when there was intensive disk-bound load.

fixed by D. Yemanov


Builds and Installers

() — (POSIX): changeServerMode.sh could mess with configuration.

fixed by A. Peshkov


() — (WINDOWS): The Windows installer exhibited some character set confusion (ANSI vs. UTF8) in its GUI for non-English environments.

fixed by P. Reeves


() — (WINDOWS): The Windows build would fail if a path to source files contained spaces.

fixed by D. Sibiryakov


() — (WINDOWS): ibase.h from the x64 package included a nonexistent file.

fixed by D. Yemanov


() — (POSIX): the installer script would fail in some Linux environments.

fixed by A. Peshkov, M. Masyk


Firebird 3.0 Release Candidate 1: Bug Fixes and Minor Improvements

The following bug fixes and tweaks were reported prior to version 3.0.0 release candidate 1:

Minor Improvements

() — IMPROVEMENT: Most errors that occur in providers are fatal and it is pointless to try other providers in the hope another will succeed when that one failed. As an example, if the remote redirector detects that the format of the database connection string matches a certain protocol, trying other providers will be futile if connection on that protocol failed. Or, suppose the engine reports some internal database error (other than invalid ODS version), it will be no use trying to open the same database through the loopback provider.

A better option appeared to be to enumerate the codes for those errors after which it would make sense to try other providers — primarily isc_unavailable and isc_no_priv. Given the desirability of retaining the content of the status vector as-is, the solution was to add some more error codes.

Implemented by A. Peshkov


() — IMPROVEMENT: Creation and loading of functions was slow when there were many functions in the database. This process has been given a speed boost.

Implemented by A. dos Santos Fernandes


() — IMPROVEMENT: Creating packages containing many functions was slow. This process has been given a speed boost.

Implemented by A. dos Santos Fernandes


() — IMPROVEMENT: I/O operations on the nbackup backup file at are now aligned at the page size boundary

Implemented by V. Khorsun


() — IMPROVEMENT: Internal contention in the background garbage collector has been reduced

Implemented by V. Khorsun


() — IMPROVEMENT: Performance was improved on backing out a dead record version or undoing a newly created primary record version when the back-version chain gets too long

Implemented by V. Khorsun


Bugs

Core Engine

() — Ordering by a compound index together with a range condition was giving wrong results

fixed by D. Yemanov


() — Packaged procedure with a default input value would not be called if the default parameter was not specified

fixed by A. dos Santos Fernandes


() — SuperServer could hang on database shutdown

fixed by V. Khorsun


() — The message “Incorrect parameters provided to internal function INF_transaction_info Invalid token” was reported when trying to commit a two-phase-commit transaction

fixed by A. Peshkov


() — Statistics could not be recalculated or updated for any system index

fixed by A. dos Santos Fernandes


() — A compound ALTER TABLE statement that tried to ADD and DROP the same check constraint would fail

fixed by A. dos Santos Fernandes


() — Casting a Dialect 1 date to string would break when a domain with a check constraint was present

fixed by A. dos Santos Fernandes


() — Operations (get statistics, change forced wites, make backup) could not be completed on a database when working in embedded mode

fixed by V. Khorsun


() — PSQL source with “ELSE IF ( <expr> ) THEN” block containing commands to manipulate an explicit cursor would not compile

fixed by D. Yemanov


() — Predicate IS [NOT] DISTINCT FROM was not being pushed into unions/aggregates, resulting in sub-optimal plans

fixed by D. Yemanov


() — ALTER DOMAIN …​ TO new_name would allow the new_name to be specified in the same pattern as domains generated by the system, i.e. matching to 'RDB$[[:DIGIT:]]*'

fixed by A. dos Santos Fernandes


() — Moving a database into the 'full shutdown' state could corrupt indexes on long-key text columns

fixed by V. Khorsun


() — MERGE / HASH JOINs would produce incorrect results when VARCHAR join keys differed only by trailing spaces

fixed by D. Yemanov


() — An access violation would occur if the server was shut down with active trace session(s)

fixed by V. Khorsun


() — Index corruption could occur while inserting data into a long-key-indexed field

fixed by V. Khorsun


() — Failure of assertion (m_format→fmt_length == m_data.getCount())

fixed by D. Yemanov, V. Khorsun


() — The fbsvcmgr could produce a broken log when restoring a database

fixed by A. Peshkov


() — Placing a comment (single-lined or multi-lined) between the final END statement and the terminator character (e.g. ^) in stored procedure code produced a compiling error if the utility tool used the legacy API

fixed by A. Peshkov


() — Using fbsvcmgr with action_trace_start under SuperSsrver would cause a livelock and prevent attachments using local protocol

fixed by V. Khorsun


() — An AFTER CREATE/ALTER PACKAGE DDL trigger would run before inserts and updates of RDB$PROCEDURES and RDB$FUNCTIONS, whereas it should run after all changes to the system tables.

fixed by A. dos Santos Fernandes


() — A script containing EXECUTE BLOCK with multiple nested BEGIN…​END statements would cause a crash during parsing

fixed by A. dos Santos Fernandes


() — The version 3 fbclient could not work with servers older than version 2.5 via the remote protocol

fixed by D. Yemanov


() — A segmentation fault could occur when working with a saved exception in a request

fixed by A. Peshkov


() — The client library was handling non-UTF8 representation of international characters incorrectly in the SPB

fixed by A. Peshkov


() — MERGE …​ WHEN NOT MATCHED …​ RETURNING was returning wrong (non-NULL) values when no insert was performed

fixed by A. dos Santos Fernandes


() — Altering a trigger to attempt to change the table to which it belongs did not succeed, but failed with bizarre results and an unhelpful error message

fixed by A. dos Santos Fernandes


() — GRANT UPDATE(c) ON T TO U01 WITH GRANT OPTION: user U01 would be unable to revoke this privilege if some DML was executed before REVOKE

fixed by A. Peshkov


() — “Invalid request BLR at offset 361 context already in use (BLR error)” was the wrong message for the error concerned, occurring only when the context count was near the 256 limit

fixed by D. Yemanov


() — Deadlock could occur during attachment

fixed by V. Khorsun


() — A column CHECK constraint could not be created if the column was based on a domain

fixed by A. dos Santos Fernandes


() — Regression: wrong handling in FOR-cursor when a “NOT EXISTS( select from <`view`> )” expression was used to check results obtained from a stored procedure

fixed by D. Yemanov


() — The server would ignore asynchronous requests (monitoring or cancellation) while preparing a query with a lot of windowed functions

fixed by A. dos Santos Fernandes


() — Invalid error message: “CHARACTER SET OCTETS is not defined” when creating a database specifying isc_dpb_lc_ctype = OCTETS

fixed by A. dos Santos Fernandes


() — A database could be created with an invalid character set as its default

fixed by A. dos Santos Fernandes


() — CREATE TABLE would throw an error if a column was being defined with the character set DOS775

fixed by A. dos Santos Fernandes


() — (subquery + derived table + union) would produce a wrong result set

fixed by D. Yemanov


() — Anomalies existed between the three methods of user management, regarding valid user names and passwords

fixed by A. Peshkov


() — Validation of domain constraints in PSQL was inconsistent

fixed by A. dos Santos Fernandes


POSIX-Specific

() — On AIX and Solaris, all processes on a Classic server would hang after the death of one process

fixed by A. Peshkov


Windows-Specific

() — Embedded server could not work without the folder C:\ProgramData\firebird

fixed by V. Khorsun


() — Online validation on Windows would fail if the dbname argument contained forward slash (‘/’) and a concurrent attachment existed that also used ‘/

fixed by V. Khorsun


Services API

() — When using the Services API to manage users (from fbsvcmgr or by building a SPB in some program, for example) it was required — but not documented — to place the user name parameter immediately after the desired action (add/modify). If any other parameter was placed before the user name, the Services Manager on the server would go crazy and build a wrong command line for the gsec service.

fixed by A. Peshkov


Utilities

fbsvcmgr

() — Online validation during DML activity in another connection would lead to errors: “Error while trying to read from file” and “Page in use during flush (210)”

fixed by V. Khorsun


() — A named trace session launched by fbsvcmgr with a non-empty value for the parameter trc_name could not be stopped using its name

fixed by V. Khorsun


gbak

() — The connection information in an ON CONNECT trigger could not be saved if the connection was established by gbak

fixed by A. dos Santos Fernandes


gfix

() — Using gfix -online in Classic returned the message “IProvider::attachDatabase failed when loading mapping cache” if access was using a remote protocol

fixed by A. Peshkov


() — Online validation could return a false report of double-allocated pages

fixed by V. Khorsun


() — gfix would return an error after correctly shutting down a database

fixed by A. Peshkov


isql

() — The isql INPUT command (or the -i option at the command line) was reading large (> 64K) lines incorrectly

fixed by A. dos Santos Fernandes


() — CREATE DATABASE fails in isql

fixed by A. Peshkov


() — In isql, SET COUNT ON reports the wrong number of affected rows when issuing UPDATE on a view which was created WITH CHECK OPTION

fixed by A. dos Santos Fernandes


Firebird 3.0 Second Beta Release: Bug Fixes

The following bug fixes were reported prior to the version 3.0.0 Beta 2 release:

Core Engine

() — failure of the engine to check the validity of RETURNING_VALUES for EXECUTE PROCEDURE and INTO for EXECUTE STATEMENT could lead to bugchecks.

fixed by A. dos Santos Fernandes


() — HASH/MERGE JOIN was not used for more than two streams if they were joined via USING/NATURAL clauses and the join was based on DBKEY concatenations.

fixed by D. Yemanov


() — Regression: generators could be seen and modified by unprivileged users.

fixed by D. Yemanov


() — Regression: MIN() or MAX() with a join would ignore possible index navigation.

fixed by D. Yemanov


() — Timestamps that were out of valid range were not prevented from being cast to VARCHAR or CHAR.

fixed by A. dos Santos Fernandes


() — When an attempt was made to insert a duplicate value into a primary or unique key column, where the length of the key was 127 characters or more, the problematic key was not shown in the error message.

fixed by D. Yemanov


() — Maximum string length (32765 bytes) was not being validated by the parser.

fixed by A. dos Santos Fernandes


() — Table aliasing was being required unnecessarily when doing UPDATE …​ RETURNING with RDB$ pseudo-columns.

fixed by A. dos Santos Fernandes


() — CREATE USER …​ TAGS ( argument_1 = 'value1', …​, argument_N = 'valueN' ) would return wrong results from the statement when there were many arguments.

fixed by A. Peshkov


() — In CREATE USER …​ TAGS ( attr = 'prefix suffix' ) the character “” in the attribute value would cause the subsequent characters to be dropped from storage.

fixed by A. Peshkov


() — A user with non-ASCII (multi-byte) characters in the name could not be created.

fixed by A. Peshkov


() — A request synchronization error could occur when building an expression index.

fixed by V. Khorsun


() — A parameterized exception would produce wrong output when the number of arguments exceeded seven.

fixed by A. dos Santos Fernandes


() — Manipulations with a GTT from several attachments (using EXECUTE STATEMENT with an ON EXTERNAL clause and different roles) could lead to the error “Internal Firebird consistency check (invalid SEND request (167), file: JrdStatement.cpp line: 325)”.

fixed by V. Khorsun


() — Firebird could hang in the embedded mode.

fixed by V. Khorsun


() — EXECUTE STATEMENT using BLOB parameters would result in an “Invalid BLOB ID” error.

fixed by A. dos Santos Fernandes


() — The error “Invalid BLOB ID” could occur when retrieving MON$STATEMENTS.MON$SQL_TEXT using EXECUTE STATEMENT with an ON EXTERNAL clause with the db_connect argument unspecified.

fixed by V. Khorsun


() — The statement ALTER DATABASE SET DEFAULT CHARACTER SET exhibited two faults: 1) it would take effect only once for the current attachment; 2) it was not checking that the new character set existed until it was actually used.

fixed by A. dos Santos Fernandes


() — A granted role would not work with a non-ASCII user name.

fixed by A. Peshkov


() — SIMILAR TO with quantifier {n,} in the pattern was failing in version 2.5 with the error “Invalid pattern” and was producing strange results in version 3.0.

fixed by A. dos Santos Fernandes


() — The command ALTER TABLE <T> ALTER TYPE <C> DOMAIN_WITH_NOT_NULL was not verifying the data in column <C> and would make incorrect assignments in <C> to ZERO / JULIAN_DATE / ASCII(0) for types INT, TIMESTAMP and VARCHAR.

fixed by A. dos Santos Fernandes


() — Inconsistencies were exhibited with ALTER DOMAIN and ALTER TABLE involving DROP NOT NULL and PRIMARY KEYs.

fixed by A. dos Santos Fernandes


() — The message “Statement failed, SQLSTATE = 00000 + unknown ISC error 0” would appear when issuing REVOKE ALL ON ALL FROM <existing_user>.

fixed by A. Peshkov


() — A “BLOB not found” error would be thrown at rollback after inserting into a table with an expression index.

fixed by V. Khorsun


() — Messages stating “Error in isc_release_request() …​ when working with legacy security database” were appearing in firebird.log for the Classic server when connecting with legacy authentication.

fixed by A. Peshkov


() — Join order was less optimal than in version 2.x.

fixed by D. Yemanov


() — An error was being thrown while preparing a complex query (“Too many Contexts of Relation/Procedure/Views. Maximum allowed is 256”.)

fixed by D. Yemanov


() — Conditions like WHERE <field> = <cursor>.<field> would not use an existing index.

fixed by D. Yemanov


() — A computed index based on a computed column would store NULL for all its keys.

fixed by D. Yemanov


() — Constraint violation errors could be swallowed in some cases.

fixed by D. Yemanov


() — SELECT FROM MON$TABLE_STATS did not work on Superclassic and Classic.

fixed by D. Yemanov


() — Search criteria would produce wrong results on WHERE <field_C> STARTING WITH <:value> ORDER BY <field_N> when field_C was the leading part of a compound index key consisting of (field_C, field_N).

fixed by D. Yemanov


() — A bug was introduced when implementing support for multiple user managers, whereby the error “Missing security context for …​..\SECURITY3.FDB” could appear when performing some operations where per-database security was configured.

fixed by A. Peshkov


() — The server could hang while chasing dead record versions.

fixed by D. Starodubov


() — MSVC8 and MSVC9 builds of Beta 1 were failing.

fixed by D. Starodubov


() — A user with the RDB$ADMIN role was denied the privilege for CREATE DATABASE.

fixed by A. Peshkov


() — Security context and open database errors could present under heavy concurrent load.

fixed by A. Peshkov


() — The error “No current record for fetch operation” could occur when the same indexed column was specified in both the WHERE and the ORDER BY clauses.

fixed by D. Yemanov


() — When the shared memory region backing the lock table could not be remapped, the status vector would return the error message “Lock manager out of room” without any low-level details (e.g. OS level error), thus hiding the real cause of the problem.

fixed by D. Yemanov


() — Firebird’s handling of ‘:’ characters in mount table entries was invalid.

fixed by A. Peshkov


() — PSQL variables based on “Domain” and “Type Of” referring to BLOB with sub_type < 0 were no longer working.

fixed by A. dos Santos Fernandes


() — Triggers with the UPDATE OR INSERT statement and IIF() were not working as expected.

fixed by D. Yemanov


() — Rollback was failing to undo changes when a MERGE statement was updating the same target rows multiple times and the optimizer used PLAN MERGE.

fixed by D. Yemanov


() — EXECUTE STATEMENT was inflating the CHAR_LENGTH() size for VARCHARs.

fixed by A. dos Santos Fernandes


() — The REPLACE() function was not working correctly with multi-byte character sets.

fixed by A. dos Santos Fernandes


() — The embedded engine would try to load the ICU DLLs from PATH folders.

fixed by A. Peshkov


() — During the “linger” period, problems would occur when attempting to change some database properties.

fixed by A. Peshkov


() — The embedded server was trying to load UDFs from wrong place.

fixed by V. Khorsun


() — The incorrect error message was returned for PSQL functions when the number of actual arguments did not match the number of formal arguments.

fixed by D. Yemanov


() — Selecting from a table with ICU columns would fail where the database had been created on Linux with ICU 4.2.1 and then copied to Windows.

fixed by A. Peshkov


() — BUGCHECK(183) was thrown when using a cursor with ORDER BY ID+0 and FOR UPDATE WITH LOCK.

fixed by D. Yemanov


() — The server would not accept the right plan.

fixed by D. Yemanov


() — Procedure source could not be deleted.

fixed by D. Yemanov


() — The message “Modifying function <F> which is currently in use” was displayed when <F> was called from an internal function declared in another unit.

fixed by D. Yemanov


() — Problems would occur when a table grew beyond 65535 pointer pages.

fixed by D. Yemanov


() — Index and BLOBs garbage collection was not working for in-place updates.

fixed by D. Sibiryakov & D. Yemanov


() — User savepoints were not being released on COMMIT.

fixed by D. Sibiryakov


() — BUGCHECK(177) was being thrown for a MERGE involving multiple matches.

fixed by D. Yemanov


() — Stored PSQL functions would accept duplicate input arguments.

fixed by D. Yemanov


() — It was not possible to insert records into a table having a column GENERATED BY DEFAULT AS IDENTITY in its DDL.

fixed by D. Yemanov


() — It was possible to store a 31-character string in a VARCHAR(25) column.

fixed by A. dos Santos Fernandes


() — Field position would be wrong after ALTER POSITION.

fixed by A. dos Santos Fernandes


() — Page-level “lock conversion denied” or “lock denied” errors could occur under high load when no deadlocks were apparent.

fixed by V. Khorsun


() — COUNT (DISTINCT …​) was too slow.

fixed by D. Yemanov


Server Crashes

() — An access violation would occur on an attempt to manage the users list using EXECUTE STATEMENT on behalf of a non-SYSDBA user having the RDB$ADMIN role.

fixed by V. Khorsun


() — Superclassic would crash at disconnect after running EXECUTE STATEMENT as non-current user.

fixed by D. Yemanov


() — Infinite mutual stored procedure calls were causing the server to crash instead of returning the appropriate error, “Too many concurrent executions of the same request”.

fixed by V. Khorsun


() — The server would bugcheck or crash on an exception in a calculated index.

fixed by D. Yemanov


() — The application would crash after calling fork() in a process using libfbembed.so.

fixed by A. Peshkov


API/Remote Interface

() — fbclient was executing the event callback function twice.

fixed by A. Peshkov


() — isc_cancel_events() was returning the error “invalid events id (handle) (code: 335545021)” if an event was no longer queued.

fixed by A. Peshkov


() — The Superclassic server could hang when receiving a network packet.

fixed by A. Peshkov


() — Bad op_execute packet would kill the server.

fixed by A. Peshkov


() — Events were broken in Firebird 3.

fixed by A. Peshkov


() — Attachment would fail when using LegacyAuth on the server and the default configuration on the client.

fixed by A. Peshkov


() — Linux servers could reject connections with a transliteration error.

fixed by D. Yemanov


() — The Legacy_Auth plugin would not connect from a FB3 to a FB2.5 server when tried after Win_Sspi.

fixed by A. Peshkov


Utilities

isql

() — SHOW TABLE command in isql would fail when the table contained a field with Unicode collation in its DDL.

fixed by A. dos Santos Fernandes


() — isql was padding BLOB columns wrongly when the column alias had more than 17 characters.

fixed by A. dos Santos Fernandes


() — In isql, INPUT files were not being properly closed.

fixed by A. Peshkov


() — In isql, the command SHOW USERS would display only the current user.

fixed by A. Peshkov


() — The call to setlocale(LC_CTYPE, "") should not be made in isql for Windows.

fixed by F. Schlottmann-Goedde & D. Yemanov


gbak

() — Restore of a shadowed database would fail if the -k (“restore without shadow”) switch was used.

fixed by A. Peshkov


() — Permissions for generators and exceptions were being corrupted after backup/restore.

fixed by D. Yemanov


gsec

() — Typing add -user SYSDBA -pw masterkey -admin yes into gsec would crash firebird.exe.

fixed by V. Khorsun & A. Peshkov


Firebird 3.0 Second Beta Release: Improvements

The following improvements were implemented prior to the version 3.0.0 Beta 2 release:

() — INSERTING, UPDATING and DELETING were made reserved words to fix ambiguity with Boolean expressions.

implemented by A. dos Santos Fernandes


() — An expression “where bool_field IS true | false” can now use the same index as “where bool_field = true | false” if such an index exists.

implemented by A. dos Santos Fernandes


() — Issuing DML or DDL statements on the RDB$* system tables is prohibited in Firebird 3.

implemented by A. Peshkov


() — A flag was added to MON$DATABASE to assist in determining what type of security database is used — default, self or other.

implemented by A. dos Santos Fernandes


() — Flushing dirty pages to disk after creation of a temporary table index is now avoided.

implemented by V. Khorsun


() — Resolved some risky issues on POSIX when aliases in databases.conf were pointing to symbolic or hard links, allowing invalid database accesses.

implemented by A. Peshkov


() — Internal temporary blobs are released early, to free up some memory/disk space.

implemented by V. Khorsun


() — Diagnostics will now report a tag name when transliteration errors occur in Parameter Block values.

implemented by A. Peshkov


() — Support added for having multiple UserManagers in firebird.conf and for using them from SQL.

implemented by A. Peshkov


() — The configuration parameter CryptPlugin was renamed to WireCryptPlugin.

implemented by A. Peshkov


() — The data type of the result returned by the functions CHAR_LENGTH(), BIT_LENGTH() and OCTET_LENGTH() for BLOBs was changed to BIGINT.

implemented by A. dos Santos Fernandes


() — “Object in use” errors now come with more details.

implemented by D. Yemanov


() — Support was added for WHEN SQLSTATE error handlers in PSQL.

implemented by D. Yemanov


() — Support was added for text BLOBs >= 32K as the first argument for the TRIM() function.

implemented by A. dos Santos Fernandes


() — IPv6 support for the remote interface.

implemented by M. Kubecek


() — Compression of data over the network.

implemented by A. Peshkov


Firebird 3.0 First Beta Release

The following improvements and bug fixes were reported prior to the version 3.0.0 Beta 1 release:

Core Engine

() — The Cache Writer thread could not start.

fixed by V. Khorsun


() — Regression: Incorrect result in subquery with aggregate.

fixed by A. dos Santos Fernandes


() — ALTER PACKAGE was returning a wrong error.

fixed by A. dos Santos Fernandes


() — Incorrect size of the output parameter or argument when EXECUTE BLOCK, procedure or function used a system field in the metadata character set.

fixed by A. Peshkov


() — GDSCODE could have value = 0 in WHEN-section under some concurrent environments. This bug affected Superclassic and Classic models but not Superserver.

fixed by V. Khorsun


() — A DDL trigger was remaining active after being dropped.

fixed by A. dos Santos Fernandes


() — A DB_KEY based join of two tables could be ineffective.

fixed by D. Yemanov


() — DDL permissions were not protecting against removal of BLOB filters.

fixed by R. Simakov


() — Regression: trace was reporting UPDATEs in statistics when doing INSERT into <some_table>.

fixed by D. Yemanov


() — Use of a named cursor would fail if a statement was not executed.

fixed by A. Peshkov


() — A FOR SELECT <L> FROM <T> AS CURSOR <C> seems to return a wrong result if table <T> is modified inside the cursor’s BEGIN…​END block. Cursor references, which are not variables, should represent the current state of the record. If it was updated “in place” (via “where current of”), then cursor references should return the new values. The first example reported in should return NULLs.

fixed by A. dos Santos Fernandes


() — Failure to load a provider or plugin was not reported anywhere.

fixed by A. Peshkov


() — The field RDB$MAP_TO_TYPE was missing from the system table RDB$TYPES.

fixed by A. Peshkov


() — CREATE USER GRANT ADMIN ROLE did not work.

fixed by A. Peshkov


() — Duplicate tags for CREATE/ALTER USER were not handled correctly.

fixed by A. Peshkov


() — Regression: The NOT NULL constraint, if declared in a domain, did not work.

fixed by A. dos Santos Fernandes


() — A positioned UPDATE statement would preclude its index usage for the subsequent cursor field references.

fixed by D. Yemanov


() — Engine could hang and block all attachments in an out-of-disk-space condition during physical backup.

fixed by V. Khorsun


() — GlobalRWLock could not downgrade an EX lock to SH if readers were present.

fixed by V. Khorsun


() — After calling release() instead of detach() for an attachment to a database in embedded mode, the attachment would remain interminably active.

fixed by A. Peshkov


() — Properties of a user created in Legacy_UserManager were padded with spaces up to 10 characters.

fixed by A. Peshkov


() — Pointless extraction of generic DDL trigger.

fixed by A. dos Santos Fernandes


() — A query executed via EXECUTE STATEMENT was returning the wrong result.

fixed by A. dos Santos Fernandes


() — EXECUTE STATEMENT ON EXTERNAL was not finding a Firebird 2.5 database.

fixed by A. Peshkov


() — “Cursor not found” error when using the legacy API.

fixed by A. Peshkov


() — SELECT WITH LOCK could enter an infinite loop for a single record.

fixed by D. Yemanov


() — Run-time errors were returning incorrect line/column information.

fixed by A. dos Santos Fernandes


() — Explicit cursors containing correlated subqueries in the select list were performing poorly.

fixed by D. Yemanov


() — Preparation of an erroneous DDL statement was not indicating that the main command failed.

fixed by A. dos Santos Fernandes


() — A procedure would execute infinitely if it contained more than 32767 statements inside any BEGIN…​END block.

fixed by D. Yemanov


() — Truncation error when using EXECUTE STATEMENT with a BLOB.

fixed by A. dos Santos Fernandes


() — Package compilation was not checking for duplicate names.

fixed by A. dos Santos Fernandes


() — Deadlock could occur when two data pages contained record fragments pointing to each other.

fixed by V. Khorsun


() — A CREATE FUNCTION or CREATE PROCEDURE statement that referred to a non-existent exception would return the error message “Error while parsing function’s BLR” instead of “exception not defined”.

fixed by D. Yemanov


() — A WHERE predicate containing NULL IS NOT DISTINCT FROM (select min(NULL) from …​) was returning the wrong result.

fixed by D. Yemanov


() — Equality predicate distribution was not working for some complex queries.

fixed by D. Yemanov


() — SELECT from derived table which contains GROUP BY on a column with a literal value was returning wrong results.

fixed by D. Yemanov


() — Parsing of a recursive query would return the error “Column does not belong to referenced table” when the source table did have such a column.

fixed by A. dos Santos Fernandes


() — Sorting records were larger than was really necessary.

fixed by D. Yemanov


() — Error “no current record for fetch operation” when table inner joins procedure inner joins table.

fixed by D. Yemanov


() — Resources (e.g. sort files) owned by a trigger could be left unreleased when the trigger was interrupted asynchronously.

fixed by D. Yemanov


() — LAG, LEAD and NTH_VALUE would raise an error when the second argument was NULL.

fixed by A. dos Santos Fernandes


() — The function LAG returned an incorrect result if the OFFSET value was assigned from a table.

fixed by A. dos Santos Fernandes


() — The keyword SET was required in the ALTER USER statement when it should have been optional.

fixed by A. Peshkov


() — Regression: Predicates involving PSQL variables or parameters were not pushed inside the aggregation.

fixed by D. Yemanov


() — Error “Attempt to reopen an open cursor” could be raised if the query handle was reused in a different transaction.

fixed by D. Yemanov


() — The 'Cache Writer' record in MON$ATTACHMENTS would vanish when deleting, via delete from MON$ATTACHMENTS, another connection that was running a heavy update on a big table.

fixed by V. Khorsun


() — Any fields present only in the WHERE clause of a view WITH CHECK OPTION would cause an invalid CHECK CONSTRAINT violation.

fixed by A. dos Santos Fernandes


() — Possible races during service destruction.

fixed by A. Peshkov


() — Non-ASCII data in SEC$USERS was not read correctly.

fixed by A. dos Santos Fernandes


() — The error “Statement already has a cursor assigned” would be thrown when trying to execute another SQL statement using a different cursor name.

fixed by A. Peshkov


() — Expression index might be overlooked for derived fields or view fields.

fixed by D. Yemanov


() — A “BLOB not found” error would be returned after creating or altering an invalid trigger.

fixed by D. Yemanov


() — An over-long column name for a SELECT alias was not being rejected, as it should have been.

fixed by A. dos Santos Fernandes


() — A database which had active attachments could not be replaced from a backup file even after the database was shut down.

fixed by D. Yemanov


Server Crashes

() — The server would crash in the garbage collector thread at disconnect of the last attachment.

fixed by V. Khorsun


() — The server could crash while disconnecting from the database under load.

fixed by D. Yemanov


() — A database validation bug could cause the server to crash.

fixed by V. Khorsun


() — The server would crash when executing almost any “show …​” commands after a reconnect.

fixed by A. Peshkov


() — Firebird would crash after an unsuccessful remapping of the lock table’s shared memory.

fixed by D. Yemanov


() — The server would crash when getting an explained plan for a DBKEY-based retrieval.

fixed by D. Yemanov


() — The server would crash when using ROW_NUMBER() OVER (PARTITION BY x) in an ORDER BY clause.

fixed by D. Yemanov


() — The server could crash while sorting records longer than 128KB.

fixed by D. Yemanov


() — The engine would crash when aggregate or window functions were used in a recursive query.

fixed by A. dos Santos Fernandes


() — Regression: isql was not destroying the SQL statement.

fixed by A. Peshkov


() — The engine would crash when the Trace config contained the line “connection_id=NN” and an attempt was made to connect to a non-existent database or alias.

fixed by V. Khorsun


() — The engine would crash when an attempt to recreate a table with a foreign key was made after a syntax error that preceded the RECREATE attempt.

fixed by A. dos Santos Fernandes


API/Remote Interface

() — CREATE DATABASE would fault if fbclient.dll was loaded from another directory (Providers = Engine12).

fixed by V. Khorsun


Utilities

isql

() — isql would issue the warning “Bad debug info format” when connecting to a database with stored functions after a restore.

fixed by D. Yemanov


() — isql would crash without connecting when executing the command SHOW VERSION.

fixed by A. Peshkov


() — isql would truncate a BLOB when reading an empty segment.

fixed by A. dos Santos Fernandes


() — Regression: isql would crash when receiving statistics from the execution of a SQL query.

fixed by V. Khorsun


gbak

() — gbak restore would fail on a database containing dependency between views and packaged functions.

fixed by A. dos Santos Fernandes


() — User collations based on UNICODE were not being upgraded to a newer ICU version on restore.

fixed by A. dos Santos Fernandes


() — gbak refused to commit the index for a primary key with characters accented with an umlaut.

fixed by A. dos Santos Fernandes


() — Version 3 gbak was unable to restore backups made on earlier server versions.

fixed by D. Yemanov


nBackup

() — nbackup was printing error messages to stdout instead of stderr.

fixed by A. Peshkov


qli

() — qli was throwing an error when copying NULL blobs between databases.

fixed by A. Peshkov


Firebird 3.0 Second Alpha Release

The following improvements and bug fixes were reported as fixed prior to the version 3.0.0 release:

Core Engine

() — Descending index could be very inefficient for some keys

fixed by V. Khorsun


() — A NOT NULL field from a derived table could become NULL when referred to from outside the derived table

fixed by D. Yemanov


() — TYPE OF arguments of stored functions could cause the server to hang if depending on a domain or column that had been changed

fixed by A. dos Santos Fernandes


() — A subquery involving a windowed function and a where <field> IN(select …​) condition could cause an error

fixed by D. Yemanov


() — An unexpected lock conflict error could be raised while connecting to a heavily loaded database

fixed by D. Yemanov


() — Context parsing errors could occur with derived tables and CASE functions

fixed by D. Yemanov


() — JOIN result could be wrong when joined fields had been created via the row_number() function

fixed by D. Yemanov


() — The boundary for the minimum value for BIGINT/DECIMAL(18) was wrong

fixed by A. dos Santos Fernandes


() — The Guardian service could write garbage after the end of a message in the Event Log

fixed by V. Khorsun


() — Access violation could occur in Guardian at process shutdown

fixed by V. Khorsun


() — Metadata being reported from system table queries for UDF return arguments was different to that returned in Firebird 2.5

fixed by A. dos Santos Fernandes


() — IF (subfunc()) would throw an error when subfunc returned a BOOLEAN

fixed by A. dos Santos Fernandes


() — Bidirectional cursor was not being positioned by the first call of FETCH LAST

fixed by D. Yemanov


() — A parser conflict was causing wrong evaluation of BETWEEN and Boolean expressions

fixed by A. dos Santos Fernandes


() — Memory leak with TRIGGER ON TRANSACTION COMMIT

fixed by V. Khorsun


() — The embedded engine would hang for 5 seconds when closing, with errors about timeout in shutdown process and invalid mutex being written into firebird.log

fixed by A. Peshkov


() — A computed field would return NULL inside a BEFORE INSERT trigger

fixed by D. Yemanov


() — An incorrect “token unknown” error would occur when an SQL string ended with a hex number literal

fixed by A. dos Santos Fernandes


() — Some Boolean expressions were not being allowed

fixed by A. dos Santos Fernandes


() — Incorrect memory statistics were being reported

fixed by D. Yemanov


() — RDB$GET_CONTEXT/RDB$SET_CONTEXT parameters were being described incorrectly as CHAR NOT NULL instead of VARCHAR NULLABLE

fixed by A. dos Santos Fernandes


() — Bad performance and slow response were exhibited when many concurrent sorts were executed

fixed by D. Yemanov


() — With bugcheckabort=1 and sweep starting at gap ~21000, “Bugcheck 186 (record disappeared)” and 100% CPU load would occur

fixed by V. Khorsun


() — Unnecessary index reads could occur when using a strict inequality condition

fixed by V. Khorsun


Server Crashes

() — The server could crash on a SELECT with a long or complex list of compound AND/OR'd predicates

fixed by D. Yemanov


() — Recreation of an errant package body could cause the engine to crash

fixed by A. dos Santos Fernandes


() — Disconnecting from a database could cause a server crash

fixed by D. Yemanov


() — Sweeping a database could cause a server crash

fixed by D. Yemanov


() — The server could crash when trace activity was attempted on a database having a database-level trigger

fixed by V. Khorsun


() — Server crashes could occur, reporting “invalid lock id (NNNNN)”

fixed by A. Peshkov


API/Remote Interface

() — “Resource temporarily unavailable” errors could occur while events were being registered simultaneously

fixed by A. Peshkov


() — Database shutdown was being reported as successfully completed before all active connections had actually been interrupted

fixed by D. Yemanov


() — The new API was still returning obscure historical definition artifacts of data fields, instead of proper metadata properties that would make the interface actually usable

fixed by A. Peshkov


() — Warnings were not being returned from calls to attachDatabase()

fixed by A. Peshkov


Security/User Management

() — Log-in could succeed with an empty password

fixed by A. Peshkov


() — An uncommitted SELECT from the pseudo table SEC$USERS would block new database connections

fixed by A. Peshkov


Procedural Language

() — Positioned DELETE …​ WHERE CURRENT OF <CURSOR> could fail for tables with newly added fields

fixed by A. dos Santos Fernandes


() — Creating a procedure could be a problem if it involved adding text in DOS864 character set

fixed by A. dos Santos Fernandes


() — An error would be raised while executing an empty EXECUTE BLOCK with NOT NULL output parameter

fixed by A. dos Santos Fernandes


() — A parameterized exception would mishandle non-ASCII characters passed as the parameter

fixed by A. dos Santos Fernandes


() — Preparing an EXECUTE BLOCK that used domains was causing a memory leak

fixed by A. dos Santos Fernandes


Data Definition Language

() — Global temporary tables were able to reference permanent relations, which they should not be able to do

fixed by V. Khorsun


() — Dropping a foreign key on a Global temporary table would cause a server crash

fixed by V. Khorsun


() — Packaged routines with CHAR or VARCHAR parameters could not be created

fixed by A. dos Santos Fernandes


() — CREATE COLLATION was not verifying the base collation character set

fixed by A. dos Santos Fernandes


() — Setting a generator value twice in a single transaction would set it to zero

fixed by D. Yemanov


() — External routines DDL in packages was wrongly reporting termination with semi-colon as an error

fixed by A. dos Santos Fernandes


Data Manipulation Language

() — Wrong output would be produced when a field with the result of a windowed function was used in a query with a useless WHERE 0=0 and GROUP BY clause

fixed by D. Yemanov


() — Parameterized queries using RDB$DB_KEY would not work

fixed by D. Yemanov


() — Recursive query would return incorrect results if passage through more than one branch was requested

fixed by D. Yemanov


() — LIKE with ESCAPE clause was not working

fixed by A. dos Santos Fernandes


Utilities

gfix

() — gfix would crash when the size of the description of a limbo transaction was larger than 1 KB

fixed by V. Khorsun


fbsvcmgr

() — fbsvcmgr was not recognising sts_record_versions and other sts switches

fixed by A. Peshkov


isql

() — Bug in the isql command setlocale(LC_CTYPE, "") on Windows due to a reference to editline, which is not available on that platform

fixed by F. Schlottmann-Goedde


() — ISQL -x was failing to output the START WITH clause of generators/sequences

fixed by A. dos Santos Fernandes


() — New permission types were not being displayed by isql

fixed by D. Yemanov


() — It was impossible to enter certain characters in isql

fixed by F. Schlottmann-Goedde


gbak

() — Backup/restore from an older version to version 3.0 would fail with a BLR error

fixed by D. Yemanov


() — A backup containing procedures or triggers that selected from external tables could not be restored with ExternalFileAccess = None

fixed by D. Yemanov


() — Owner name was missing for generators and exceptions restored from a backup

fixed by D. Yemanov


nbackup

() — nbackup's delta file was ignoring the Forced Writes setting of the database

fixed by V. Khorsun


Database Monitoring (MON$)

() — Deadlock could occur while accessing the monitoring tables under concurrent load

fixed by D. Yemanov & V. Khorsun


() — Monitoring tables were returning incomplete information in Classic and Superclassic configurations

fixed by D. Yemanov


Trace

() — Regular expressions with double-slash would fail in trace

fixed by A. Peshkov


() — Configuration file fbtrace.conf contained syntax errors

fixed by A. Peshkov


International Language Support

 

Installation Issues

() — Attempting to use Legacy_Auth directly after install would not work without restarting the service

fixed by P. Reeves


Firebird 3.0 First Alpha Release

The following improvements and bug fixes were reported as fixed prior to the version 3.0.0 release:

Core Engine

() — Sweep was blocking the establishment of concurrent attachments in Superserver.

fixed by V. Khorsun


() — A race condition could occur when auto-sweep was started.

fixed by V. Khorsun


() — COMPUTED BY columns and POSITION function could produce garbled results.

fixed by A. dos Santos Fernandes


() — Creating a table with computed fields containing SELECT FIRST could produce a corrupted result.

fixed by A. dos Santos Fernandes


() — The SQLDA for an aliased column in a grouped query was missing the original table name, column name and owner.

fixed by A. dos Santos Fernandes


() — Wrong results were produced when a column in the WHERE clause used the collation option (NUMERIC-SORT=1).

fixed by A. dos Santos Fernandes


() — A unique expression index would exhibit a memory alignment problem.

fixed by A. dos Santos Fernandes


() — The invalid error “attempted update of read-only column” would appear when selecting MINVALUE from a list of more than 255 elements.

fixed by A. dos Santos Fernandes


() — When an attempt was made to reduce the size of a CHAR or VARCHAR column, the numbers delivered in the error message were incorrect.

fixed by A. dos Santos Fernandes


() — A computed column would appear in non-existent rows output from a left join.

fixed by A. dos Santos Fernandes


() — Some character sets were duplicated in the system table RDB$TYPES.

fixed by A. dos Santos Fernandes


() — SIMILAR TO was not working correctly.

fixed by A. dos Santos Fernandes


() — An unprivileged user could delete from the system tables RDB$DATABASE, RDB$COLLATIONS and RDB$CHARACTER_SETS.

fixed by D. Yemanov


() — “Internal consistency check” would occur in a query with grouping by subquery+stored procedure+aggregate.

fixed by A. dos Santos Fernandes


() — It was not possible to use the SUBSTRING function to create a computed index for large columns.

fixed by A. dos Santos Fernandes


() — Some collation tweaking: FR_CA_CI_AI collation was introduced; FR_FR was changed to be identical to FR_CA and FR_FR_CI_AI was changed to be identical to the new FR_CA_CI_AI.

fixed by A. dos Santos Fernandes


() — The LIST function was concatenating binary blobs as though they were text.

fixed by A. dos Santos Fernandes


() — Collation errors could occur with the use of [type of] <domain> and type of <column>.

fixed by A. dos Santos Fernandes


() — It was possible to store a string of length 31 characters into a VARCHAR(25) column.

fixed by A. dos Santos Fernandes


() — Regression: Code changes had disabled support for expression indexes with COALESCE, CASE and DECODE.

fixed by A. dos Santos Fernandes


() — Success of row deletion depended on the order of insertion of the rows.

fixed by V. Khorsun


() — A complex expression involving RDB$GET_CONTEXT and BETWEEN worked in DSQL but failed with a conversion error when selected in a view definition.

fixed by A. dos Santos Fernandes


() — interlock.h was not portable.

fixed by A. Peshkov


() — The Firebird server could not be started under any user name other than “root”, “firebird”, “interbas” or “interbase”.

fixed by A. Peshkov


() — The collation UTF8 UNICODE_CI could not be used in a compound index.

fixed by A. dos Santos Fernandes


() — A constraint violation error involving CAST was not being raised inside views.

fixed by A. dos Santos Fernandes


() — Comparisons involving multiple index segments could produce wrong result sets.

fixed by A. dos Santos Fernandes


() — The concurrent transaction number was not being reported when a lock timeout occurred.

fixed by N. Samofatov


() — COUNT(*) from a big table could return a negative result.

fixed by D. Yemanov


() — Character class names in SIMILAR TO expressions could be case-sensitive or case-insensitive, depending on the collation of the left part, whereas they should be unequivocally case-insensitive.

fixed by D. Sibiryakov


() — An ALTER TABLE …​ ALTER COLUMN …​ ALTER POSITION operation could result in wrong column positions.

fixed by A. dos Santos Fernandes


() — The character set used in a constant was not being registered as a dependency.

fixed by A. dos Santos Fernandes


() — COLLATE expressions were being applied incorrectly.

fixed by A. dos Santos Fernandes


() — Plan output lacked the names of views when selecting from views that contained procedure calls.

fixed by D. Yemanov


() — DB_KEY was always zero for rows in external tables.

fixed by D. Yemanov


() — A full outer join could not use available indices, resulting in very slow execution sometimes.

fixed by D. Yemanov


() — Use of certain choices of character in double-quoted index names, for example a bracket character, could defeat the parsing logic when generating a human-readable plan.

fixed by D. Yemanov


() — A join of a stored procedure with a view or a table could fail with the error “No current record for fetch operation”.

fixed by D. Yemanov


() — A buffer overrun error was being caught erroneously in a DOUBLE PRECISION to VARCHAR conversion in a Dialect 1 database.

fixed by C. Valderrama C.


() — An aggregated query was causing “Bugcheck 232 (invalid operation)”.

fixed by A. dos Santos Fernandes


() — An unnecessary index scan was executed when the same index is mapped to both WHERE and ORDER BY clauses.

fixed by D. Yemanov


API/Remote Interface

() — The client library could hang after an unsuccessful attempt to connect to the remote auxiliary (events) port.

fixed by A. Peshkov


() — Parameters inside the CAST function were being wrongly described in the SQLDA as non-nullable.

fixed by A. dos Santos Fernandes


() — The client would perform detach incorrectly when the server became unavailable.

fixed by A. Peshkov


() — An erroneous “Success” message would be returned in the error status vector when failing to connect to a trash database file.

fixed by C. Valderrama C.


() — String values in error messages were not converted to the connection character set.

fixed by A. dos Santos Fernandes


Procedural Language

() — Use of a system domain in declarations of arguments or return values in a stored procedure could prevent the procedure from being modifiable.

fixed by A. dos Santos Fernandes


() — EXECUTE BLOCK parameter definitions were not being respected and could cause wrong behavior with respect to character sets.

fixed by A. dos Santos Fernandes


() — Validation of domain CHECK constraints when used in PSQL declarations was inconsistent: it was using the type of the expression, instead of the type of the variable.

fixed by A. dos Santos Fernandes


() — The names of variables or arguments could be wrong or absent in error messages when more than 256 variables were used.

fixed by A. dos Santos Fernandes


() — Resolution of EXECUTE BLOCK parameter collations was using wrong logic.

fixed by A. dos Santos Fernandes


() — Constraints on stored procedure output parameters were checked even when the procedure returned no rows.

fixed by A. dos Santos Fernandes


() — Incorrect error message (an absurd column number) was returned if an empty SQL string was prepared for EXECUTE STATEMENT.

fixed by D. Yemanov


Data Definition Language

() — Attempting to drop a non-existent generator (sequence) would result in a serious exception.

fixed by A. dos Santos Fernandes


() — Problems could occur if further DDL commands were issued in the same transaction following a CREATE COLLATION command.

fixed by A. dos Santos Fernandes


() — The ALTER TABLE command allowed the addition of a column with a NOT NULL definition, allowing a non-savvy DBAdmin to wreck the table.

fixed by A. dos Santos Fernandes


() — Unrestorable backup: a problem which would occur if ALTER TABLE …​ ADD COLUMN added a column with a NOT NULL constraint. The fix for has now made it impossible to do this.

fixed by A. dos Santos Fernandes


() — Adding a non-nullable column to a populated table would render the table inconsistent. The fix for has now made it impossible to do this.

fixed by A. dos Santos Fernandes


() — Client tools tended to be confused about how to interpret a NULL that is returned from a non-nullable column. The fix for has now made it impossible to add a non-nullable column to a populated table.

It is not clear, though, whether this part of the fix makes it mandatory to specify a default value for a non-nullable column.

fixed by A. dos Santos Fernandes


() — Bad behaviour of DELETE when the WHERE clause was a subquery involving FIRST/SKIP: the operation would zap every row in the table.

fixed by V. Khorsun


() — Any user could alter or drop generators and exceptions, a metadata security hole.

fixed by D. Yemanov


Data Manipulation Language & DSQL

() — When when preparing a query with UNION, the error “context already in use (BLR error)” was wrongly being thrown.

fixed by V. Khorsun


() — Recursive CTEs were returning a wrong error message.

fixed by V. Khorsun


() — Inserting a word containing the 8-bit character ‘ä’ into a CHARACTER SET ASCII column would succeed instead of throwing a transliteration error.

fixed by A. dos Santos Fernandes


() — The internal function ATAN2 was returning an incorrect value with arguments (0, 0).

fixed by A. dos Santos Fernandes


() — An expression index involving TRIM could lead to an incorrect indexed lookup.

fixed by A. dos Santos Fernandes


() — A common table expression context could be used with parameters.

fixed by A. dos Santos Fernandes


() — A multi-byte CHAR value requested as VARCHAR was returned with padded spaces.

fixed by A. dos Santos Fernandes


() — With UTF8 and large varchar fields, IS DISTINCT FROM would cause the error “Implementation limit exceeded”.

fixed by D. Yemanov


() — STARTING WITH ? (where the parameter value supplied is an empty string) would fail if the plan used a compound index.

fixed by A. dos Santos Fernandes


() — Infinite insertion cycle: INSERT INTO THIS_TABLE SELECT …​ FROM THIS_TABLE would loop forever until resources were exhausted.

fixed by V. Khorsun


Command-line Utilities

() — Utilities did not always honour the minimum number of characters required to recognise an option.

fixed by C. Valderrama C.


Other old bugs that were fixed in utilities:

FbGuard

() — Guardian would keep creating more and more threads each time FBServer died.

fixed by C. Valderrama C.


() — Firebird Guardian’s tray icon would disappear after a Windows Explorer crash.

fixed by C. Valderrama C.


isql

() — isql was generating metadata script output with syntax errors in the CHARACTER SET clause, e.g., CHARACTER SETISO8859_1.

fixed by A. dos Santos Fernandes


() — isql was padding UTF-8 data incorrectly.

fixed by A. dos Santos Fernandes


() — isql would extract the array dimensions after the character set name.

fixed by C. Valderrama C.


gbak

() — gbak did not support backup volumes of size greater than 4GB.

fixed by A. Peshkov


() — gbak would restore invalid views without any warning to the user.

fixed by C. Valderrama C.


() — Several validations were lacking in gbak.

fixed by C. Valderrama C.


nbackup

() — nbackup could hide the real cause of a failure.

fixed by C. Valderrama C.


International Language Support

() — The “Sharp-S” character was being treated incorrectly in the UNICODE_CI_AI collation.

fixed by A. dos Santos Fernandes


14. Firebird 3.0 Project Teams

Table 6. Firebird Development Teams
Developer Country Major Tasks

Dmitry Yemanov

Russian Federation

Full-time database engineer/implementor; core team leader

Alex Peshkov

Russian Federation

Full-time security features coordinator; buildmaster; porting authority

Claudio Valderrama

Chile

Code scrutineer; bug-finder and fixer; ISQL enhancements; UDF fixer, designer and implementor

Vladyslav Khorsun

Ukraine

Full-time DB engineer; SQL feature designer/implementor

Adriano dos Santos Fernandes

Brazil

International character-set handling; text and text BLOB enhancements; new DSQL features; code scrutineering

Roman Simakov

Russian Federation

Engine contributions

Paul Beach

France

Release Manager; HP-UX builds; MacOS Builds; Solaris Builds

Pavel Cisar

Czech Republic

QA tools designer/coordinator

Philippe Makowski

France

QA tester

Paul Reeves

France

Win32 installers and builds

Mark Rotteveel

The Netherlands

Jaybird implementor and co-coordinator; Documentation writer

Jiri Cincura

Czech Republic

Developer and coordinator of .NET providers

Alexander Potapchenko

Russian Federation

Developer and coordinator of ODBC/JDBC driver for Firebird

Stephen Boyd

Canada

GPRE contributions

Alexey Kovyazin

Russian Federation

Website coordinator

Paul Vinkenoog

The Netherlands

Coordinator, Firebird documentation project; documentation writer and tools developer/implementor

Norman Dunbar

U.K.

Documentation writer

Pavel Menshchikov

Russian Federation

Documentation translator

Tomneko Hayashi

Japan

Documentation translator

Umberto (Mimmo) Masotti

Italy

Documentation translator

Helen Borrie

Australia

Release notes editor; Chief of Thought Police

Appendix A: Licence Notice

The contents of this Documentation are subject to the Public Documentation License Version 1.0 (the “License”); you may only use this Documentation if you comply with the terms of this Licence. Copies of the Licence are available at http://1688shucai.com/pdfmanual/pdl.pdf (PDF) and http://1688shucai.com/manual/pdl.html (HTML).

The Original Documentation is entitled Firebird 3.0 Release Notes.

The Initial Writer of the Original Documentation is: Helen Borrie. Persons named in attributions are Contributors.

Copyright © 2004-2020. All Rights Reserved. Initial Writer contact: helebor at users dot sourceforge dot net.