MariaDB Connector/ODBC 3.0.3 Changelog

You are viewing an old version of this article. View the current version here.

Note: This page describes features in the source repository for MariaDB Connector/ODBC. There are currently no official packages or binaries available for download which contain the version of the driver described here. If you want to try out any of the new features described here you will need to get and compile the code yourself.

For the highlights of this release, see the release notes.

The revision number links will take you to the revision's page on GitHub. On GitHub you can view more details of the revision and view diffs of the code modified in that revision.

  • Revision #ed35eb6 2018-02-06 12:55:20 +0100
    • Added missing dependency in windows build
  • Revision #ed92516 2018-02-04 23:01:24 +0100
    • [ODBC-134] Fix and testcase. Fetch would fail, if unbound column contained NULL, and for that column some arbitrary descriptor field was set by application(but not value/len/ind buffer ptrs). The bug affected ADO, as that is something it can do in some cases with CursorLocation adUseClient
  • Revision #b7520af 2018-02-01 10:20:13 +0100
    • Fix and testcase of the bug, that was found during investigation on ODBC-109. It's not clear atm if the bug causes the reported error in the connector's work, since I can't reproduce the bug. But looks probable. For the info type SQL_SCHEMA_TERM/SQL_OWNER_TERM the connector returned wrong length. Or more exactly - did not write anything into application's length buffer
  • Revision #f0fc11a 2018-01-29 14:06:30 +0100
    • Slightly changed the test failing on Travis - using KILL on travis instead of timeout emulation, mnd ade some better output.
  • Revision #740c862 2018-01-28 23:21:04 +0100
    • [ODBC-119] The fix and testcase. The connector ordered SQLStatistics results using worng columns. That could cause MS Access to pick wrong column as a unique index. And that, in its turn, could cause mangled data shown for the linked table. Fix of testcase in catalog2 that failed against 10.3 due to new privelege. Added test of SQL_GROUP_BY info to ODBC-123 testcase. It had similar problem(and tested in similar way) Fixed build in VS2017
  • Revision #256bf8d 2018-01-26 16:21:06 +0100
  • Revision #1c97fd9 2018-01-21 23:38:16 +0100
    • [ODBC-131] Fix and the testcase. While linking table in MS Access, it threw error when it received unexpected length for SQLSMALLINT and SQLINTEGER columns from SQLColumns resultset. New graphic images for the installer on Windows.
  • Revision #4b507cd 2018-01-17 14:48:56 +0100
    • Small fix of the testcase for SQLProcedureColumns - reference values for RADIX field were wrong for 2 parameters. Has to be 10(or theoretically 2) for numeric types
  • Revision #554afe2 2018-01-13 18:02:15 +0100
    • Temporary turned off tests atainst MaxScale in Travis
  • Revision #cfb8e53 2018-01-12 17:50:40 -0500
    • Addition to the fix of ODBC-123 and the testcase for that problem - for SQL_CATALOG_LOCATION info type the connector incorrectly wrote to the buffer SQLUINTEGER, while it has to be SQLUSMALLINT
  • Revision #acdd5cd 2018-01-10 20:54:38 +0100
    • [ODBC-126] Fix and testcases. Some of internal structures weren't reallocated in accordance with the next resultset fields count. Beyond the case described in the report, they cover the case of statements batch. It had similar and own problems. Removed unused property from the statement structure.
  • Revision #a1e8995 2017-12-27 21:46:48 +0100
    • [ODBC-123] Fix and testcase. LibreOffice sets SQL_ATTR_USE_BOOKMARKS attribute, but does not actually use bookmarks. The connector was not ready for that, and would throw wrongly error, and then crash. Fixed a number of compilation warnings.
  • Revision #9f9ca38 2017-12-13 12:07:27 +0000
    • Don't space pad parameters with non-zero scale
  • Revision #99272f0 2017-12-07 14:35:42 +0100
    • Correction in the travis config to actually run tests against MaxScale
  • Revision #4300efe 2017-06-22 14:12:19 +0200
    • correct include guard
  • Revision #7d5f4fa 2017-12-03 16:25:51 +0100
    • Adding MariaDB 5.5 server to Travis tests. Skipping one tests with 5.5 servers(and MySQL <5.7), since it doesn't make sense there.
  • Revision #050ebc8 2017-12-19 19:16:55 +0100
    • Fix of waring from recent pull request
  • Revision #740206f 2017-12-13 10:49:18 +0000
    • Return correct scale for SQL_C_NUMERIC
  • Revision #c7059c4 2017-11-27 22:32:02 +0100
    • Fixing remaining issues in Travis Made (most of) testframework global variables static, as one of them clashed with variable in connector's env, and that caused problems in some cases. Also fixed few tests in connstring, that expected test connection to use password. Skipped one of tests there(in connstring) on travis.
  • Revision #d0523b2 2017-11-23 01:30:48 +0100
    • [ODBC-120] Fix of performance issue. We did redundant calls of mysql_stmt_data_seek. They are neede for different type of cursors, positioned operation, array fetch etc. But in general forward_only cursor it only significantly slows down execution.
  • Revision #c5659d5 2017-11-17 18:40:02 +0100
    • Fixed connection procedure call in couple of tests, since 1 parameter type was changed in previous commit, and these references were overlooked. Changed in travis' script ODBCINI and ODBCSYSINI initialization to use PWD(not sure why:)). Added env var export for tests to detect Travis. Skipping connstring in Travis. I'd say this is harmless - it is kinda a special testsuite, and also to check if that is simply a linking problem(libodbcinst.so is the suspect). Fixed SQL_API SQLTablePrivilegesW - it would always return empty resultset in CatalogName is NULL. That was not intended, and its ANSI counterpart doesn't behave like that. Need to file bug report about that. But I suspect other catalog functions have the same problem.
  • Revision #cae4d21 2017-11-14 13:02:13 +0200
    • Fix of testcases failing in travis. Mostly that is adding cursor closing, as older UnixODBC versions require it even there it is not really required. One test in catalog2(bug50195) changed more profoundly, as it implied, that db server is on the same host with tests. Added Travis detection in the tests framework.
  • Revision #4aae120 2017-11-09 13:48:59 +0200
    • Changed DSN parsing/saving/reading tests(connstring.c) in order they do not fail with unixODBC. UnixODBC has buggy ini cache(which seemingly will be fixed in 2.3.5): if dsn is changed, and then read again, application will get old cached value. That doesn't occur if unixODBC is built with `--enable-inicaching=no` Thus tests have been changed to create and use individual DSN's, rather then one common for all tests DSN. I guess what some of tests are not failing is rather the result of that bug :) Probably it'd be better to simply skip failing tests if run with (old version of) unixODBC
  • Revision #d31b4b0 2017-10-19 15:51:54 +0200
    • Initial travis setup
  • Revision #945dacf 2017-10-19 08:20:57 +0200
    • Version bump -> 3.0.3

Comments

Comments loading...
Content reproduced on this site is the property of its respective owners, and this content is not reviewed in advance by MariaDB. The views, information and opinions expressed by this content do not necessarily represent those of MariaDB or any other party.