Dave Page [Tue, 2 Aug 2005 08:09:39 +0000 (08:09 +0000)]
Post release version number bump
Dave Page [Tue, 2 Aug 2005 07:38:04 +0000 (07:38 +0000)]
correct version number which seems tohave been inadvertantly bumped.
Dave Page [Mon, 1 Aug 2005 20:58:50 +0000 (20:58 +0000)]
Fix memory leak per gborg bug 1356 [pauldaugherty]
Dave Page [Mon, 1 Aug 2005 20:39:04 +0000 (20:39 +0000)]
Fix SQL_MAX_IDENTIFIER_LEN per gborg bug ref: 1348
Dave Page [Mon, 1 Aug 2005 20:18:25 +0000 (20:18 +0000)]
Include win_unicode.c in Unix build because it contains functions that are used under unix. (per gborg bug ref 1176)
Dave Page [Wed, 13 Jul 2005 19:51:29 +0000 (19:51 +0000)]
Fix format string [Marko Ristola]
Dave Page [Wed, 13 Jul 2005 14:28:50 +0000 (14:28 +0000)]
Bump version number to avoid confusion with earlier snapshots.
Dave Page [Thu, 30 Jun 2005 09:30:38 +0000 (09:30 +0000)]
According to MSDN docs for SQLSetStmtAttr, if you try to request a
cursor type which isn't supported, the driver should fall back to a
supported type, set an appropriate SQLState and return
SQL_SUCCESS_WITH_INFO.
Currently, if a SQL_ATTR_CURSOR_TYPE of SQL_CURSOR_DYNAMIC is requested,
we fall back to SQL_CURSOR_FORWARD_ONLY and no warning is given; this is
simply because we don't explicitly check for SQL_CURSOR_DYNAMIC.
A better solution is to check for SQL_CURSOR_DYNAMIC, fall back to
SQL_CURSOR_KEYSET_DRIVEN and set a warning accordingly.
Also, currently if SQL_CURSOR_KEYSET_DRIVEN is requested and only static
cursors are supported, we fall back without returning a warning.
[Mark G. Adams]
Dave Page [Wed, 27 Apr 2005 10:34:00 +0000 (10:34 +0000)]
Bug ref: 1262
When specifiying a client_encoding, the driver will always use UTF-8.
The problem is in connection.c, line number 1024, the code that reads:
if (!self->client_encoding || stricmp(self->client_encoding, "UNICODE"))
should read:
if (!self->client_encoding || !stricmp(self->client_encoding, "UNICODE"))
(the check with stricmp will always be true when client_encoding is different from UNICODE, not when it is UNICODE, wich is what's intended).
[melkor <NOSPAM> odyssey.com.uy]
Dave Page [Wed, 27 Apr 2005 10:27:29 +0000 (10:27 +0000)]
Bug ref: 1276
When I attempt to bind a null-terminated Unicode string parameter using
SQLBindParameter, the database engine receives a string representation of
today's date instead of the data I passed. The same issue does not occur if
I explicitly specify the length of the string instead of using SQL_NTS.
[jeremy <NOSPAM> accessdata.com]
Dave Page [Mon, 25 Apr 2005 08:08:22 +0000 (08:08 +0000)]
This patch returns the remarks (description) on a table, if available. If there is no description, it returns an empty string when SQLTables API is called. This patch works for versions with schema support.
[Anoop Kumar]
Dave Page [Mon, 25 Apr 2005 07:54:17 +0000 (07:54 +0000)]
Bump version to 08.01.0102
Peter Eisentraut [Fri, 18 Mar 2005 15:34:48 +0000 (15:34 +0000)]
When inserting SHRT_MIN into an int2 column, the ODBC driver comes up
with the following SQL:
insert into dbc_test ( i16 ) values ( -32768::int2 );
This fails with an out of range error. Changed to:
insert into dbc_test ( i16 ) values ( (-32768)::int2 );
patch by Kelly Burkhart
Dave Page [Sat, 5 Mar 2005 11:46:00 +0000 (11:46 +0000)]
Fix driver registration
Dave Page [Sat, 5 Mar 2005 10:33:02 +0000 (10:33 +0000)]
Fix version number
Dave Page [Wed, 2 Mar 2005 15:11:09 +0000 (15:11 +0000)]
The following patch allow users to configure the Unix domain socket path, using the "Uds" property.
I took some code from the unixODBC PG driver and performed the necessary adjustments.
Please note that unlike unixODBC, I have followed the current convention, which means that:
1) "Uds" is considered only if "Servername" is empty.
2) Default path is "/tmp".
3) The provided path should point the folder that comtains the UDS node.
[Ramon Danieli]
Dave Page [Wed, 2 Mar 2005 15:02:24 +0000 (15:02 +0000)]
Post-release version number bump
Dave Page [Wed, 2 Mar 2005 14:37:52 +0000 (14:37 +0000)]
Include installer.
Dave Page [Wed, 2 Mar 2005 14:11:30 +0000 (14:11 +0000)]
Version bump for release.
Dave Page [Wed, 23 Feb 2005 20:59:33 +0000 (20:59 +0000)]
Correct version number
Dave Page [Wed, 23 Feb 2005 20:49:21 +0000 (20:49 +0000)]
Show version number in installer.
Dave Page [Wed, 23 Feb 2005 16:49:41 +0000 (16:49 +0000)]
Remove unused icon.
Dave Page [Wed, 23 Feb 2005 16:42:32 +0000 (16:42 +0000)]
We don't implement SQLSetScrollOptions
[Joost Kraaijeveld]
Dave Page [Wed, 23 Feb 2005 16:41:50 +0000 (16:41 +0000)]
Don't barf when parsing queries with functions in the table list (eg. SELECT * FROM version() )
[Joost Kraaijeveld]
Dave Page [Mon, 14 Feb 2005 13:07:11 +0000 (13:07 +0000)]
Increase SOCK_BUFFER_SIZE to 8192 per report from Karol Szkudlarek
Dave Page [Sun, 13 Feb 2005 18:09:40 +0000 (18:09 +0000)]
Fix directory used for debug builds
Dave Page [Sun, 13 Feb 2005 18:06:29 +0000 (18:06 +0000)]
Fix missing " win32.mak
Dave Page [Thu, 10 Feb 2005 15:41:30 +0000 (15:41 +0000)]
Note the need for vcvars32.bat when compiling in Windows on the command line.
Dave Page [Tue, 8 Feb 2005 15:11:21 +0000 (15:11 +0000)]
Bump version to 08.00.0005
Dave Page [Sat, 29 Jan 2005 23:34:23 +0000 (23:34 +0000)]
New WiX based installer.
Dave Page [Fri, 28 Jan 2005 12:03:41 +0000 (12:03 +0000)]
08.00.0004 Version bump
Dave Page [Mon, 17 Jan 2005 10:27:33 +0000 (10:27 +0000)]
Bump version for new snapshot
Dave Page [Mon, 17 Jan 2005 10:12:52 +0000 (10:12 +0000)]
With CVS HEAD, numeric values are not read correctly.
The SQL_NUMERIC_STRUCT is filled with
.precision=3, .scale=5 for a number like 500.00123, which MSDASQL will
interpret as .00123 (noticed when MSDASQL had to convert it to a
string). Obviously the precision should be 8 or more.
The attached patch corrects this.
[Andreas Pflug]
Dave Page [Tue, 21 Dec 2004 15:07:41 +0000 (15:07 +0000)]
There is a 16-byte leak for every statement handle that has SQLFetch called on it.
I have tracked this leak down to its source:
SC_Fetch calls extend_getdata_info, which allocates memory for the stmt.getdata_info.
However, this memory is never freed when the statement is deallocated.
[Scot Loach]
Dave Page [Thu, 11 Nov 2004 09:21:19 +0000 (09:21 +0000)]
Return type names without parens. Originally changed by Hiroshi for unknown reasons. Fixes reported SQL problems such as "CREATE TABLE foo(bar varchar()(12));" [Scot Loach]
Dave Page [Wed, 27 Oct 2004 07:48:03 +0000 (07:48 +0000)]
Updated Japanese dialog from Hiroshi Saito
Dave Page [Fri, 8 Oct 2004 09:44:05 +0000 (09:44 +0000)]
Ignore search_path in SQLTables
Dave Page [Fri, 8 Oct 2004 09:43:48 +0000 (09:43 +0000)]
Bump version number
Dave Page [Fri, 8 Oct 2004 08:50:08 +0000 (08:50 +0000)]
OK, finally think this SQLInfo unicode string length bug can be put to rest.
Tested and working in Access 2K3, MS ODBC Testsuite, and some of my own test and live project code.
Dave Page [Thu, 7 Oct 2004 21:49:31 +0000 (21:49 +0000)]
2nd attempt at the previously rolled back fix. This bug causes errors in MS Access when linking tables.
These occur when *pcbInfoValue returns a value that Access doesn't believe is correct (http://msdn.microsoft.com/archive/default.asp?url=/archive/en-us/dnaraccess/html/msdn_odk.asp)
Previously, string size was assumed to be WCLEN * len in Unicode. This is wrong as it meant we included 1 extra character.
Dave Page [Wed, 6 Oct 2004 11:16:15 +0000 (11:16 +0000)]
Back out previous revision. Can't see or recall why it was made and it just doesn't seem right somehow.
Dave Page [Mon, 4 Oct 2004 12:39:15 +0000 (12:39 +0000)]
Bump version number to 08.00.0001 in prep for PostgreSQL 8.0
Dave Page [Mon, 4 Oct 2004 12:30:27 +0000 (12:30 +0000)]
Turn off updateable cursors by default.
Clarify the Updateable Cursors option as being experimental.
Various GUI cleanups.
Peter Eisentraut [Fri, 30 Jul 2004 21:08:12 +0000 (21:08 +0000)]
Add explicit cast to bound int2 parameters (see bug ID 882)
Dave Page [Fri, 30 Jul 2004 12:50:38 +0000 (12:50 +0000)]
Support setting of SQL_ATTR_ANSI_APP w/ODBCVER >= 3.51
Dave Page [Mon, 26 Jul 2004 14:10:38 +0000 (14:10 +0000)]
Don't include the terminating character when returning pcbInfoValue from SQL_GetInfo.
Peter Eisentraut [Sun, 25 Jul 2004 17:27:47 +0000 (17:27 +0000)]
Update copyright year
Dave Page [Wed, 21 Jul 2004 12:29:59 +0000 (12:29 +0000)]
Make the driver ODBC 3 only.
Dave Page [Wed, 21 Jul 2004 11:39:16 +0000 (11:39 +0000)]
Remove unicode #ifdefs and configure option.
Dave Page [Wed, 21 Jul 2004 10:46:29 +0000 (10:46 +0000)]
Remove unused file.
Dave Page [Tue, 20 Jul 2004 20:24:11 +0000 (20:24 +0000)]
Bump build number
Dave Page [Tue, 20 Jul 2004 20:23:32 +0000 (20:23 +0000)]
Change Insight Distribution Systems the PGDG.
Dave Page [Tue, 20 Jul 2004 19:43:25 +0000 (19:43 +0000)]
Add missing psqlodbc.rc file.
Peter Eisentraut [Tue, 20 Jul 2004 08:30:17 +0000 (08:30 +0000)]
Do dynamic allocation when saving the locale, since some people seem to
have amazingly long locale values.
Dave Page [Tue, 20 Jul 2004 07:42:48 +0000 (07:42 +0000)]
Minor tidy up
Dave Page [Mon, 19 Jul 2004 15:58:48 +0000 (15:58 +0000)]
Cleanup formatting, remove some unicode/odbcver ifdefs.
Dave Page [Mon, 19 Jul 2004 15:21:11 +0000 (15:21 +0000)]
Bump build number
Dave Page [Mon, 19 Jul 2004 15:20:46 +0000 (15:20 +0000)]
Remove unicode & odbc3 ifdefs, correct driver file name.
Dave Page [Mon, 19 Jul 2004 10:19:58 +0000 (10:19 +0000)]
Bump version to 07.05.0001
Peter Eisentraut [Sat, 17 Jul 2004 21:01:07 +0000 (21:01 +0000)]
Some bug fixes in the function mapping table in convert.c. Removed
odbc.sql, because convert.c does its job.
Dave Page [Fri, 16 Jul 2004 08:22:54 +0000 (08:22 +0000)]
Updated a couple of comments
Dave Page [Fri, 16 Jul 2004 07:52:55 +0000 (07:52 +0000)]
Return the required SQLGetInfo buffer size in bytes not characters to prevent memory stomping in Unicode apps.
Peter Eisentraut [Sun, 11 Jul 2004 17:10:22 +0000 (17:10 +0000)]
Make ODBC 0x0300 and Unicode the default on Unix, and fix the compilation in
these modes.
Peter Eisentraut [Sat, 10 Jul 2004 22:26:27 +0000 (22:26 +0000)]
Remove traces of code that tries to compile on Unix without a driver manager.
Look automatically for a driver manager if none is requested explicitly.
Peter Eisentraut [Fri, 9 Jul 2004 18:50:51 +0000 (18:50 +0000)]
Fix buffer overflows related to unchecked make_string().
Dave Page [Wed, 7 Jul 2004 07:55:03 +0000 (07:55 +0000)]
I've looked at the warnings in info.c. The reason of the warnings was macro NULL_IF_NULL defined in info.c:49. (Karol Szkudlarek)
Also removed an unused variable (Dave)
Dave Page [Mon, 5 Jul 2004 09:55:16 +0000 (09:55 +0000)]
Remove unused files. These files are only required for the non-iodbc/non-unixodbc build of the driver on unix which is useless anyway.
Dave Page [Thu, 1 Jul 2004 09:20:56 +0000 (09:20 +0000)]
This patch fixes several misguided assumptions that sizeof(long) == 4, which is not true on 64-bit architectures (Peter Eisentraut)
Dave Page [Mon, 14 Jun 2004 07:40:35 +0000 (07:40 +0000)]
Cleaned up installer.
Dave Page [Sat, 12 Jun 2004 22:02:57 +0000 (22:02 +0000)]
Fix a directory reference that wasn't using an alias.
Dave Page [Sat, 12 Jun 2004 21:38:16 +0000 (21:38 +0000)]
Fix a directory reference that wasn't using an alias.
Dave Page [Wed, 14 Apr 2004 09:17:20 +0000 (09:17 +0000)]
Update version to 07.03.0209
Make the Unicode version the primary and remove others.
Dave Page [Wed, 14 Apr 2004 09:11:56 +0000 (09:11 +0000)]
Update version to 07.03.0209
Make the Unicode version the primary and remove others.
Dave Page [Fri, 9 Apr 2004 18:35:32 +0000 (18:35 +0000)]
Add WIN1250 support - Ludek Finstrle
Dave Page [Fri, 9 Apr 2004 18:13:12 +0000 (18:13 +0000)]
the following patch (against 7.03.0200) contains two things:
- The first hunk is necessary to build psqlodbc using Borland C++ Builder 6.0 [I successfully built the whole DLL, which requires a few other patches]
- The second hunk is necessary as Borland's sscanf would do all sorts of things with rest[] when CodeGuard is active (I wonder if it's not their way of hinting that we have an opportunity for a buffer overrun here).
Cyrille Ch?p?lov <cyrille@softek.fr>
Bruce Momjian [Tue, 6 Apr 2004 14:22:02 +0000 (14:22 +0000)]
When compile and start using psqlodbc.so, it crashes with missing export
QR_get_value_backend_row().
I think there is a missing #include "qresult.h" in multibyte.c
Best Regards,
Georgi Kolev
Dave Page [Mon, 5 Apr 2004 13:07:08 +0000 (13:07 +0000)]
Fix for invalid input syntax for type tid: '(,)' - Jose' Cruanyes
Hiroshi Inoue [Thu, 5 Feb 2004 09:11:44 +0000 (09:11 +0000)]
[7.3.0208]
1) Change catalog functions to search 'public' schema if a user
tries to search the user named schema and finds no record to
return.
2) Set the statement status to STMT_FINISHED(not STMT_PREPARED)
if the pre execution causes an error.
3) Adjust null terminator for SQL_C_WCHAR type.
4) Change getCharColumnSize to return max size when the
statement has no result.
5) Cache the number of parameters for statements.
Dave Page [Wed, 4 Feb 2004 08:55:28 +0000 (08:55 +0000)]
[Jon Willeke] I'm attaching a few tweaks to build psqlodbc with Unicode support on Unix. I changed some code that assumes two-byte SQLWCHAR and I changed configure.ac and Makefile.am, as necessary.
I tested this with unixODBC 2.2.7 on SuSE Linux 8.2. I built unixODBC --with-iconv-ucode-enc=UCS-4LE and with SQL_WCHART_CONVERT defined, so that SQLWCHAR is wchar_t. I have not tested it with UCS-2 / unsigned short.
I also tested this with iODBC 3.51.1, which always uses wchar_t on Unix.
One caveat, if anyone actually plays with this: sqlext.h should include sqlucode.h, so I haven't patched psqlodbc to do so. The next release of unixODBC will contain this fix, and I've mentioned it to the iODBC maintainer.
Hiroshi Inoue [Tue, 6 Jan 2004 09:58:22 +0000 (09:58 +0000)]
1) Correct the handling of SQL_C_ULONG (for *nix).
2) Prevent plan deallocation errors when the transaction
is in abort status.
3) Avoid a connection failure when notice message arrives.
4) Improve the handling of mutex(critical section).
5) Correct the cursor open check.
6) Change the Unicode driver to set the NULL terminator
of SQL_C_WCHAR type data properly.
7) Add some m(re)alloc error check.
8) Add proper cleanup steps for some functions.
9) Return proper min & max scale for timestamp data type.
10) Return proper scale for timestamp data fields.
11) Fix the bug that .009 numeric generates .9 output. (George A.J)
Hiroshi Inoue [Tue, 9 Dec 2003 10:01:38 +0000 (10:01 +0000)]
[7.3.0205]
(for *nix)
1. Detect -lsocket for some platforms.
2. Handle unix domain sockets.
3. EINTR safety for send/recv.
4. Set TCP_NDEALY option for inet connections.
<1-4: thanks to Hartmut Raschick>
5. Use recursive mutex if possible.
6. Add descriptor.c to Makefile.
(for Windows)
7. Automatic setting of client encoding for latin1(9).
(common)
8. Avoid those notorious core dumps when logging on hitting
"null" data...
<8: thanls to Hartmut Raschick>
9. Removed a static varible for thread-safety.
10. Improve the performance of keyset-driven fetch operation.
11. Map (var)char type to SQL type more properly.
12. Change create_params for NUMERIC type to (precison, scale).
13. Add type cast to remove some compiler warnings.
Hiroshi Inoue [Wed, 12 Nov 2003 10:15:21 +0000 (10:15 +0000)]
[7.03.0204]
1) Clear statements' cursors on commit/rollback.
2) Change SQLColumns, SQLTablePrivileges to take show_system_tables
option into account.
3) When an ODBC 2.x application working with an ODBC 3.x driver
calls SQLExecDirect, SQLExecute, or SQLParamData, and a searched
update or delete statement was executed but did not affect any
rows at the data source, the ODBC 3.x driver should return
SQL_SUCCESS. When an ODBC 3.x application working with an ODBC
3.x driver calls SQLExecDirect, SQLExecute, or SQLParamData
with the same result, the ODBC 3.x driver should return
SQL_NO_DATA.
Hiroshi Inoue [Mon, 10 Nov 2003 07:10:07 +0000 (07:10 +0000)]
[7.03.0203]
1) Add conversion from text to SQL_C_BINARY(bug report from Brad Wilson).
2) Change SQLColumns, SQLTables etc to initialize the statement handle
properly(bug report from Yutaka Tanida).
3) Improve the handling of mutex under *nix(bug report from Steve Wampler).
Hiroshi Inoue [Thu, 6 Nov 2003 01:35:02 +0000 (01:35 +0000)]
[7.03.0202]
1) Improve SQLColAttributes(SQL_COLUMN_DISPLAY_SIZE) (bug report from Benjamin Riefenstahl).
2) Correct EN_set_odbc3 macro (bug report from tomas@nocrew.org(Tomas Skäre)).
Hiroshi Inoue [Sat, 25 Oct 2003 04:19:23 +0000 (04:19 +0000)]
[7.03.0201]
1) Revise the handling of descriptors and implement SQLCopyDesc.
2) Handle data_at_execution columns for SQLSetPos or SQLBulkOperations.
Dave Page [Wed, 22 Oct 2003 13:32:58 +0000 (13:32 +0000)]
Version number bump.
Dave Page [Wed, 22 Oct 2003 13:08:49 +0000 (13:08 +0000)]
Version number bump.
Hiroshi Inoue [Mon, 20 Oct 2003 02:37:42 +0000 (02:37 +0000)]
[7.03.0116]
1. Fix a memory leak in case of updatable KEYSET_DRIVEN cursors.
2. Change SQLMoreResults to initialize the cursor position.
3. Hopefully, SQLDescribeCol() returns a more appropriate
error code(SQLSTATE).
Bruce Momjian [Wed, 24 Sep 2003 03:47:30 +0000 (03:47 +0000)]
Remove trailing commas in structs; reported by jason dang
Hiroshi Inoue [Fri, 19 Sep 2003 09:25:17 +0000 (09:25 +0000)]
1) Correct the return info for SQLGetInfo(.., SQL_CURSOR_COMMIT
(ROLLBACK)_BEHAVIOR) call.
2) Explictly reject the SQLBrowseConnect call.
Hiroshi Inoue [Mon, 15 Sep 2003 08:25:03 +0000 (08:25 +0000)]
1. Serialize the connection use for communication(multi-thread).
2. Let escape strings case-insensitive.
Hiroshi Inoue [Tue, 2 Sep 2003 11:25:53 +0000 (11:25 +0000)]
Fix a bug when handling data_at_execution parameter array
reported by Jol Valentine <jolv@genaware.com>.
Hiroshi Inoue [Wed, 27 Aug 2003 10:17:53 +0000 (10:17 +0000)]
1) Fix a bug on SQLSetStmtAttr(.., SQL_CONCURRENCY) reported by Shachar Shemesh.
2) Apply a patch to changes the storage type of the debug "func" variables
to be stored in the constant area of the program(Shachar Shemesh).
3) Correct the column size in case of bytea as LO.
Hiroshi Inoue [Thu, 31 Jul 2003 01:57:50 +0000 (01:57 +0000)]
1) Distinguish the rowset size for SQLFetch(Scroll) from that for
SQLExtendedFetch.
2) Implement SQLGetDiagFieldW() for the Unicode driver.
Hiroshi Inoue [Fri, 11 Jul 2003 10:47:17 +0000 (10:47 +0000)]
Improve SQLGetDiagField so that an MFC appliaction reports
error messages properly(bug report from Silvio Scarpati).
Hiroshi Inoue [Wed, 9 Jul 2003 08:09:11 +0000 (08:09 +0000)]
Add quotes around table names to use ::regclass in parse.c .
Hiroshi Inoue [Mon, 30 Jun 2003 03:51:24 +0000 (03:51 +0000)]
Change SQLGetFunctions to contain ODBC3 APIs.
Dave Page [Thu, 19 Jun 2003 16:10:20 +0000 (16:10 +0000)]
When doing an UPDATE or DELETE via ODBC, if no records are affected then SQLExecute should return SQL_NO_DATA_FOUND and not SQL_SUCCESS.
The following patch to RH9 applies to the version in CVS (with an offset). This patch has been tested on RH9.
[Tim Woodall]
Dave Page [Wed, 18 Jun 2003 13:24:23 +0000 (13:24 +0000)]
Fix incorrect patch.
Dave Page [Wed, 18 Jun 2003 12:57:58 +0000 (12:57 +0000)]
Fix misplaced comment end.