release.html

来自「postgresql-odbc,跨平台应用」· HTML 代码 · 共 629 行 · 第 1/2 页

   .The outfit of SQLDescribeParam
   .PREPARE/EXECUTE is changed from the command base to the protocol base.
   .FETCH command is changed to an Execute request as it has come out.
   .Moreover, bytea and lo type coexistence is possible.
   .Others...

2. Enhancement in declare/fetch mode (7.4 or later)
   Even if a certain renewal occurs on the way by holdable cursor exploitation, 
   a commitment is possible in comfort.
   Moreover, the static update cursor was supported.
   (Only for SQLSetPos correspondence, SQLBulkOperation are un-supporting)  
   Bug fix of the cursor which can be updated.

3. Rollback support of a syntactical unit at the time of an error.(8.0 or later)
   Can choose from three Transaction module,Statement module,Nothing.
   These can be specified by data source setup.

4. Output parameter support (8.1 or later)
   Corresponded to the output parameter of the function introduced by 8.1.
   (Only the return value of a function was supported before.)
</pre>
<hr>
<h2>psqlODBC 08.01.0200 Release</h2>
<p>psqlODBC 08.01.0200 contains only documentation changes over the 08.01.0108
development build.</p>
<br>
<li>Notes: psqlODBC 08.01.0108</li><br>
1) fix most warnings during build in x86_64 environment<br>
2) fixed point of failure on x86_64 boxes<br>
3) fixed autocommit = off behaviour with starting transactions when needed<br>
4) quick hack againist access violation with Disallow premature - 
   this rarely used feature is still broken, but should no longer crash<br>
5) report correct SQLSTATE in SQLGetDiagField and SQLGetDiagRec<br>
<br>
<li>Notes: psqlODBC 08.01.0107</li><br>
1) fix next problem with calling implicit rollback in bad time<br>
2) detecting transaction commands into if branch which mean backend
   successfuly done the command. add support for "start transaction"<br>
3) fix bug when mylog output can't be created<br>
4) fix problem with SQLSetPos<br>
5) fix problem with column length in CC_mapping<br>
6) fix - typelen could be less then zero not only -1<br>
<br>
<li>Notes: psqlODBC 08.01.0106</li><br>
1) fix problem with getting long text in unicode driver<br>
2) fix sqlstate iin case of error in statement<br>
3) doesn't call rollback in autocommit off or in manual transaction<br>
4) remove obsolete parameter protocol<br>
<br>
<li>Notes: psqlODBC 08.01.0105</li><br>
Changes:<br>
1) fix problem with rollback aborted queries<br>
2) SQLColumns or SQLColAttribute may return bad answer when Show system tables is turned off<br>
3) adds support for ; delimited options in ConfigDSN<br>
4) fixes problem with binding parameters and SQLCancel without SQLFreeStmt<br>
<br>
<li>Notes: psqlODBC 08.01.0104</li><br>
Changes:<br>
1) add commit after set basic options in SQLDriverConnect when autocommit is turned off<br>
2) driver doesn't fail when there isn't bind enough parameters<br>
3) collattributes doesn't work since switch to libpq.<br>
   There is still problem with select '' as xx from table<br>
4) SQLTables after SQLExecDirect has problem with GetData for all rows<br>
5) grow MAXKEYLEN for SQLConfigDataSource<br>
6) repair behaviour of more queries in one statement<br>
<br>
<li>Notes: psqlODBC 08.01.0103</li><br>
Changes:<br>
1) RowCount problem in SELECT<br>
2) Server side prepare - but only first run in allocated statement<br>
   (http://pgfoundry.org/tracker/index.php?func=detail&aid=1000460&group_id=1000125&atid=538)<br>
3) Table names repeated in BDE with usedeclarefetch=1<br>
4) Binding parameters are now freed in SQLCancel<br>
5) Skip fetching data for commands<br>
<br>
<li>Notes: psqlODBC 08.01.0102</li><br>
Changes:<br>
1) Server side prepare now works correctly.<br>
2) A bug in SQLTables that shows up in Borland SQL Explorer in<br>
3) Declare/Fetch mode has been fixed.<br>
4) A number of inconsistent internal APIs have been cleaned up. These<br>
issues only showed up on some OS's.<br>
5) Correct value returned by SQLGetInfo for<br>
SQL_CATALOG_LOCATION/SQL_QUALIFIER_LOCATION. This prevented use of the<br>
driver in Visual Studio 2005.<br>
<br>
The last four numbers will represent the build. If the last 2 digits are<br>
both zero, a stable version is indicated. If the last two digits are<br>
non-zero, it is a development snapshot. <br>
So, starting with the next stable release, we will have:<br>
08.01.0200 - Stable<br>
08.01.0201 - Development<br>
...<br>
...<br>
08.01.0299 - Development<br>
08.01.0300 - Stable<br>
Of course, we won't necessarily have 99 snaphosts between stable releases!<br>
<br>
<li>Notes: psqlODBC 08.01.0101</li><br>
Changes:<br>
This release fixes a serious problem in non-declare/fetch mode where<br>
recordsets over the configured declare/fetch cache size were returned<br>
incorrectly. It is advised that all users of v08.01.0100 upgrade as soon<br>
as possible.<br>
<br>
<h2>psqlODBC 08.01.0100 Release</h2>
This release represents a major change to the driver over previous versions;<br>
in particular, the network code has been replaced with the PostgreSQL libpq<br>
library. This allows us to more easily take advantage of updates and new<br>
features that might be introduced to the PostgreSQL wire protocol, and also<br>
allowed us to upgrade straight away to the version 3 protocol which was<br>
previously unsupported. We were also able to add SSL support, as well as<br>
Kerberos authentication support.<br>
<br>
>From this release, two versions of the driver are provided for Windows,<br>
'PostgreSQL ANSI' which supports single and multibyte applications through<br>
the ANSI ODBC API, and 'PostgreSQL Unicode' which provides Unicode support<br>
through the Unicode ODBC API. On Unix systems, the driver type may be<br>
selected via a configure option.<br>
<br>
psqlODBC may be downloaded from http://www.postgresql.org/ftp/odbc/versions/<br>
in source, Windows Installer, merge module, and basic zip file formats.<br>
<br>
Please post any bug reports to the [e-mail] mailing list.<br>
<br>
I'd like to take this opportunity to thank all those involved with the<br>
development, testing and bug fixing of the updated driver.<br>
PostgreSQL Core Team<br>
<br>
<li>Notes: psqlODBC 08.00.0102</li><br>
This is a bug fix release<br>
of the pre-libpq version of the driver, specifically to fix a bug<br>
affecting Unix users<br>
(http://gborg.postgresql.org/project/psqlodbc/bugs/bugupdate.php?1176)<br>
<br>
<h2>psqlODBC 08.00.0100 Release</h2>
The driver is available in source form, as well as a dll, merge module<br>
and msi installer for Windows.<br>
<br>
Regards, Dave.<br>
<br>
<li>Notes: psqlODBC 08.00.0005</li><br>
This version<br>
contains a some recent fixes, and is packaged in a new installer built<br>
entirely using open source tools (specifically, the Wix toolkit).<br>
<br>
<li>Notes: psqlODBC 08.00.0002</li><br>
This version of the driver is a minor update<br>
with a fix to the string sizing code in SQLGetInfo, and removal of the<br>
need for tables to be in the search path or public schema for them to be<br>
returned by SQLTables.<br>
<br>
<li>Notes: psqlODBC 08.00.0001</li><br>
The purpose of this version is to try to get something stable, with<br>
proper schema/without OID support ready for the 8.0 release<br>
(particularly the win32 release). To that end, this release is mainly a<br>
quick cleanup of the following:<br>
<br>
- GUI cleanup. Dialogues should behave better now and not have unlabeled<br>
buttons or move around the screen in strange ways.<br>
<br>
- Updateable Cursor support. This is largely experimental and is (I<br>
believe) the cause of a least a couple of stability issues in the recent<br>
snapshots. In this release it is marked 'experimental' on the setup<br>
dialogue (only in English, if anyone can do the Japanese version, please<br>
send a patch) and turned off by default.<br>
<br>
I'd appreciate it if ppl could test it a little with Updateable Cursors<br>
*turned off* and report any bugs.<br>
<br>
<h2>psqlODBC 07.05.0001</h2>
This is version 07.05.0001. For those that don't know or don't recall<br>
how the versioning works, the major and minor number track the latest<br>
release of PostgreSQL that the driver is expected to work with. The<br>
revision number is a bit more complex. The first two digits are the<br>
version number, where even numbers are development versions, and odd are<br>
release. The last two digits are the build number.<br>
<br>
In other words, this driver is build 01 of development version 00, and<br>
is expected to work with PostgreSQL 7.5dev. :-) The leading zeros are an<br>
ODBC thing...<br>
<br>
<h2>psqlODBC 07.03.0100 Release</h2>
I'm pleased to announce the long awaited new version of psqlODBC. This<br>
release represents a major change in the packaging and distribution of<br>
psqlODBC. In particular the following points should be noted:<br>
<br>
1) The build number will now be used to indicate both the release number<br>
(first 2 digits), and snapshot build number (last 2 digits). this<br>
release is 0100, or release 1, snapshot 0. The first development<br>
snapshot will become 0101, the second 0102 and so on. The next official<br>
release should be 07.03.0200.<br>
<br>
2) With this release, the ODBC 3 driver is now the default. The<br>
filenames, driver names and their capabilities are as follows:<br>
<br>
psqlodbc.dll PostgreSQL ODBC3.0 Level<br>
driver<br>
psqlodbc25.dll PostgreSQL Legacy ODBC2.5 Level<br>
driver<br>
psqlodbc30w.dll PostgreSQL Unicode (Beta) Beta quality ODBC3.0<br>
level driver with Unicode support<br>
<br>
3) The 3 drivers are no longer self contained. psqlodbc.dll is used as<br>
the setup dll for all three drivers, hence no matter how you<br>
redistribute or install the drivers, you should always include this<br>
file.<br>
<br>
In addition, this driver fixes an annoying problem with recent builds of<br>
PostgreSQL on Redhat systems that have a particularly long compiler<br>
version string that caused a buffer overflow resulting in a GPF on<br>
Windows upon connection to the database.<br>
The driver is released under the LGPL, and can be downloaded from the<br>
Downloads page on the website at<br>
http://gborg.postgresql.org/project/psqlodbc/projdisplay.php<br>
<br>
<h2>=== gborg.postgresql.org(This is the past history.) ===</h2>
1.33 by hinoue Annotate  <br>
Thu Feb 5 09:11:44 2004 (2 weeks, 3 days ago)<br>
Changed since 1.32: +3 -3 [Select for Diff] <br>
Diffs to version 1.32 (long / unified) <br>
Diffs ignoring whitespace to version 1.32 (long / unified) <br>
 [7.3.0208]<br>
1) Change catalog functions to search 'public' schema if a user<br>
tries to search the user named schema and finds no record to<br>
return.<br>
2) Set the statement status to STMT_FINISHED(not STMT_PREPARED)<br>
if the pre execution causes an error.<br>
3) Adjust null terminator for SQL_C_WCHAR type.<br>
4) Change getCharColumnSize to return max size when the<br>
statement has no result.<br>
5) Cache the number of parameters for statements. <br>
<br>
1.32 by hinoue Annotate   <br>
Tue Jan 6 09:58:22 2004 (46 hours, 12 minutes ago)<br>
Changed since 1.31: +3 -3 [Select for Diff] <br>
Diffs to version 1.31 (long / unified) <br>
Diffs ignoring whitespace to version 1.31 (long / unified) <br>
 1) Correct the handling of SQL_C_ULONG (for *nix).<br>
2) Prevent plan deallocation errors when the transaction<br>
is in abort status.<br>
3) Avoid a connection failure when notice message arrives.<br>
4) Improve the handling of mutex(critical section).<br>
5) Correct the cursor open check.<br>
6) Change the Unicode driver to set the NULL terminator<br>
of SQL_C_WCHAR type data properly.<br>
7) Add some m(re)alloc error check.<br>
8) Add proper cleanup steps for some functions.<br>
9) Return proper min & max scale for timestamp data type.<br>
10) Return proper scale for timestamp data fields.<br>
11) Fix the bug that .009 numeric generates .9 output. (George A.J) <br>
<br>
1.31 by hinoue Annotate  <br>
Tue Dec 9 05:01:38 2003 (44 hours, 22 minutes ago)<br>
Changed since 1.30: +3 -3 [Select for Diff] <br>
Diffs to version 1.30 (long / unified) <br>
Diffs ignoring whitespace to version 1.30 (long / unified) <br>
 [7.3.0205]<br>
(for *nix)<br>
1. Detect -lsocket for some platforms.<br>
2. Handle unix domain sockets.<br>
3. EINTR safety for send/recv.<br>
4. Set TCP_NDEALY option for inet connections.<br>
<1-4: thanks to Hartmut Raschick><br>
5. Use recursive mutex if possible.<br>
6. Add descriptor.c to Makefile.<br>
(for Windows)<br>
7. Automatic setting of client encoding for latin1(9).<br>
(common)<br>
8. Avoid those notorious core dumps when logging on hitting<br>
"null" data...<br>
<8: thanls to Hartmut Raschick><br>
9. Removed a static varible for thread-safety.<br>
10. Improve the performance of keyset-driven fetch operation.<br>
11. Map (var)char type to SQL type more properly.<br>
12. Change create_params for NUMERIC type to (precison, scale).<br>
13. Add type cast to remove some compiler warnings. <br>
<br>
1.30 by hinoue <br>
Wed Nov 12 05:15:21 2003 (8 days, 1 hour ago)<br>
Changed since 1.29: +3 -3 [Select for Diff] <br>
Diffs to version 1.29 (long / unified) <br>
Diffs ignoring whitespace to version 1.29 (long / unified) <br>
 [7.03.0204]<br>
1) Clear statements' cursors on commit/rollback.<br>
2) Change SQLColumns, SQLTablePrivileges to take show_system_tables<br>
option into account.<br>
3) When an ODBC 2.x application working with an ODBC 3.x driver<br>
1.calls SQLExecDirect, SQLExecute, or SQLParamData, and a searched<br>
2.update or delete statement was executed but did not affect any<br>
3.rows at the data source, the ODBC 3.x driver should return<br>
4.SQL_SUCCESS. When an ODBC 3.x application working with an ODBC<br>
5.3.x driver calls SQLExecDirect, SQLExecute, or SQLParamData<br>
6.with the same result, the ODBC 3.x driver should return SQL_NO_DATA.<br>
<br>
1.29 by hinoue   <br>
Mon Nov 10 02:10:07 2003 (10 hours, 1 minute ago)<br>
Changed since 1.28: +3 -3 [Select for Diff] <br>
Diffs to version 1.28 (long / unified) <br>
Diffs ignoring whitespace to version 1.28 (long / unified) <br>
 [7.03.0203]<br>
1) Add conversion from text to SQL_C_BINARY(bug report from Brad Wilson).<br>
2) Change SQLColumns, SQLTables etc to initialize the statement handle<br>
properly(bug report from Yutaka Tanida).<br>
3) Improve the handling of mutex under *nix(bug report from Steve Wampler). <br>
 <br>
1.28 by hinoue   <br>
Wed Nov 5 20:35:02 2003 (4 days, 15 hours ago)<br>
Changed since 1.27: +3 -3 [Select for Diff] <br>
Diffs to version 1.27 (long / unified) <br>
Diffs ignoring whitespace to version 1.27 (long / unified) <br>
 [7.03.0202]<br>
1) Improve SQLColAttributes(SQL_COLUMN_DISPLAY_SIZE) (bug report from Benjamin Riefenstahl).<br>
2) Correct EN_set_odbc3 macro (bug report from [e-mail](Tomas Sk