August 2021

• Finished feature #6815 - Support multiple rows for DML RETURNING
• Fixed BLR format problem present at least since v1.5
• Add some utility code (GenericMap, RefPtr, AutoPtr) to improve codebase
• Add VSCode C++ config for better editing experience
• Add noreturn attribute to raise functions
• Task #6933 - Remove blr_fetch, never generated by Firebird and its tools
• Adjust build for MSVC 19
• Started working in #6681 - Support for WHEN NOT MATCHED BY SOURCE for MERGE statement [CORE6448].

• Tested and reviewed PR #6914 - Allow keywords in locales when creating ICU collations
• Improvement #6915 - Allow attribute DISABLE-COMPRESSIONS in UNICODE collations
• Backported fix for #6913 - Buffer overflows in getInfo APIs
• Tests and discussion about time zones and ICU installation in the client
• Fixed error in batch example.

• Fixed build in GitHub Actions
• Pull requests review.

July

• Fixed #6873 - SIMILAR TO does not use index when pattern starts with non-wildcard character (in contrary to LIKE)
• Working on feature #6815 - Support multiple rows for DML RETURNING
• Fixed problem happening when epp file is modified after initial build
• Fixed #6913 - Buffer overflows in getInfo APIs.

• Fixed #6875 - Significant performance regression of SIMILAR TO and SUBSTRING(SIMILAR) when pattern is taken from variable (rather than directly specified)
• Fixed #6887 - Invalid SIMILAR TO patterns may lead memory read beyond string limits
• Fixed warning deleting pointer to incomplete type 'Jrd::PatternMatcher' may cause undefined behavior
• Fixed BLR BLOB filter for blr_marks following blr_erase
• Fixed BLR BLOB filter for blr_receive_batch
• Feature #6910 - Add way to retrieve statement BLR with Statement::getInfo and ISQL's SET EXEC_PATH_DISPLAY BLR

• Design and discussion: New getInfo2 API
• Design and discussion: Local temporary tables internals

June

• Removed QLI and adjust the builds and TCS tests
• Created PR for #6839 - Weird string comparison with index
• Fixed #6801 - Error recompiling a package with some combination of nested functions
• Do not remove versioned filed with make clean
• Improvement #6740 - Allow parenthesized query expression for standard-compliance [CORE6511].

• Improvement #6872 - Indexed STARTING WITH execution is very slow with UNICODE collation
• Started fix for #6875 - Significant performance regression of SIMILAR TO and SUBSTRING(SIMILAR) when pattern is taken from variable (rather than directly specified).

• Fixed #5534 - String truncation exception on UPPER/LOWER functions, UTF8 database and some multibyte characters
• Fixed #6854 - Crash occurs when use SIMILAR TO and compare string with pattern that contains non-ascii character with suppressed exception handling (or "Invalid SIMILAR TO pattern" raises otherwise)
• Created PR for backported fix for CORE-5611 - Higher memory consumption for prepared statements
• Replace deprecated ubuntu-16.04 by ubuntu-20.04 in GitHub Actions.

May

Master branch

• Fixed #6787 - MON$ATTACHMENTS.MON$TIMESTAMP should use its session original time zone
• Improvement #6788 - Extend EXTRACT to extract time zone strings
• Improvement #6786 - Add session time zone to system context
• Fixed #6790 - MON$ATTACHMENTS.MON$TIMESTAMP is incorrect when DefaultTimeZone is configured with time zone different from the server's default
• Reviewed PR #315 - Compile an external trigger before executing if it was deleted
• Fixed memory leaks in external triggers examples
• Improvement #6794 -Add MON$SESSION_TIMEZONE to MON$ATTACHMENTS
• Fixed #6796 - Buffer overflow when padding line with national characters causes ISQL crash
• Dynamic libncurses versus static libtermcap patch and discussion
• Fixed #6800 - Client config DefaultTimeZone is not passed to server when isc_dpb_session_time_zone is not set
• Improvement #5588 - Support full SQL standard binary string literal syntax
• Improvement #5589 - Support full SQL standard character string literal syntax
• Feature #6798 - Add built-in functions UNICODE_CHAR and UNICODE_VAL to convert between Unicode code point and character
• Some code cleanup
• Updated build to C++17 standard
• Updated re2 library to latest stable version
• Fixed TCS tests for v4 and forked it for v5.

• Backported #5913 - Increase number of formats/versions of views from 255 to 32K (CORE-5647).

• Continuing INTL refactoring, integration in the new API
• Researching usage of smart pointer in cloop generated code
• Reviewing RDB$KEYWORDS and RDB$BLOB_UTIL pull requests
• Research about increasing max. VARCHAR length to 64K
• Proposed CALL statement.

April

• Fixed CORE-6527 - Regression: inline comment of SP parameter with closing parenthesis leads to incorrect SQL when trying to extract metadata
• Fixed some TCS tests
• Frontported fix for CORE-6517 - CREATE DATABASE fails with 'Token unknown' error when DB name is enclosed in double quotes and 'DEFAULT CHARACTER SET' is specified after DB name
• Fixed CORE-6494 - Inconsistent translation "string->timestamp->string->timestamp" in dialect 1
• Asked Dmitry question about SKD_* and time zones
• Fixed CORE-6542 - Implementation of SUBSTRING for UTF8 character set is inefficient
• Fixed TCS tests for CORE-6494.

• Backported fix for CORE-6527 - Regression: inline comment of SP parameter with closing parenthesis leads to incorrect SQL when trying to extract metadata
• Fixed CORE-6517 - CREATE DATABASE fails with 'Token unknown' error when DB name is enclosed in double quotes and 'DEFAULT CHARACTER SET' is specified after DB name
• Backported fix for CORE-6494 - Inconsistent translation "string->timestamp->string->timestamp" in dialect 1
• Integrating Ubuntu build in GitHub Actions.

• Continuing INTL refactoring, integration in the new API and improvements
• Working in unit tests in the core code
• Memory manager cleanup / external libraries problem investigation.

• Working on Jira tracker migration to GitHub issues