]> granicus.if.org Git - postgresql/log
postgresql
23 years agoDon't refer to odbcinst.ini by absolute path. SQLGetPrivateProfileString
Peter Eisentraut [Sun, 23 Sep 2001 13:32:24 +0000 (13:32 +0000)]
Don't refer to odbcinst.ini by absolute path.  SQLGetPrivateProfileString
handles this.

Don't install our own odbcinst.ini.  That's the driver manager's business.

23 years agoMake lpad/rpad/translate multibyte aware. Also add Copright notice etc.
Tatsuo Ishii [Sun, 23 Sep 2001 11:02:01 +0000 (11:02 +0000)]
Make lpad/rpad/translate multibyte aware. Also add Copright notice etc.

23 years agoAdd pg_database_encoding_max_length() function.
Tatsuo Ishii [Sun, 23 Sep 2001 10:59:45 +0000 (10:59 +0000)]
Add pg_database_encoding_max_length() function.

23 years agoAdd mention of MySQL compatibility for LIMIT in HISTORY.
Bruce Momjian [Sun, 23 Sep 2001 04:41:47 +0000 (04:41 +0000)]
Add mention of MySQL compatibility for LIMIT in HISTORY.

23 years agoUpdate regress tests for new LIMIT x,y behavior.
Bruce Momjian [Sun, 23 Sep 2001 04:28:18 +0000 (04:28 +0000)]
Update regress tests for new LIMIT x,y behavior.

23 years agoNo, file not needed.
Bruce Momjian [Sun, 23 Sep 2001 04:17:53 +0000 (04:17 +0000)]
No, file not needed.

23 years agoAdd new file.
Bruce Momjian [Sun, 23 Sep 2001 04:17:20 +0000 (04:17 +0000)]
Add new file.

23 years agoplease apply attached patch to current CVS.
Bruce Momjian [Sun, 23 Sep 2001 04:16:16 +0000 (04:16 +0000)]
please apply attached patch to current CVS.

Changes:

 1. Added support for boolean queries (indexable operator @@, looks like
       a @@ '1|(2&3)'
 2. Some code cleanup and optimization

        Regards,
                Oleg

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 23 Sep 2001 04:15:13 +0000 (04:15 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 23 Sep 2001 04:13:29 +0000 (04:13 +0000)]
Update TODO list.

23 years agoBig thanks to Solar Designer who pointed out a bug in bcrypt
Bruce Momjian [Sun, 23 Sep 2001 04:12:44 +0000 (04:12 +0000)]
Big thanks to Solar Designer who pointed out a bug in bcrypt
salt generation code.  He also urged using better random source
and making possible to choose using bcrypt and xdes rounds more
easily.  So, here's patch:

* For all salt generation, use Solar Designer's own code.  This
  is mostly due fact that his code is more fit for get_random_bytes()
  style interface.
* New function: gen_salt(type, rounds).  This lets specify iteration
  count for algorithm.
* random.c: px_get_random_bytes() function.
  Supported randomness soure: /dev/urandom, OpenSSL PRNG, libc random()
  Default: /dev/urandom.
* Draft description of C API for pgcrypto functions.

New files: API, crypt-gensalt.c, random.c

Marko Kreen

23 years agoThe attached patch is my first run-through of the JDBC test suite. A
Bruce Momjian [Sun, 23 Sep 2001 04:11:14 +0000 (04:11 +0000)]
The attached patch is my first run-through of the JDBC test suite. A
summary of changes:

 . removal of the tablename property from build.xml

 . addition of a dropTable method in JDBC2Tests and cleanups of many
methods in the same

 . all tests now use non-deprecated assertXYZ methods instead of the
deprecated assert method

 . failure in TimestampTest (testSetTimestamp) fixed. The failure is
because testSetTimestamp was inserting a timestamp with hour 7 but
checkTimeTest was expecting a timestamp with hour 8. AFAICS, there are
no issues wrt daylight savings time and timestamps being pushed in and
pulled out (but more explicit tests should be added in the future)

 . failure in TimeTest (testGetTime) fixed. Times to be inserted were
interpreted in the localtime zone but checking was done with the
assumption that the insertion was done in GMT.

 . formatting changes in a few of the source files (because I found
it convenient to have consistent formatting while working on them). The
formatting is consistent with the new format for java source files in
PostgreSQL.

Liam Stewart

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 23 Sep 2001 04:08:38 +0000 (04:08 +0000)]
Update TODO list.

23 years agoAdd mention of log file rotation with mention of syslog SIGHUP capability.
Bruce Momjian [Sun, 23 Sep 2001 04:06:24 +0000 (04:06 +0000)]
Add mention of log file rotation with mention of syslog SIGHUP capability.

23 years agoMention LIMIT change in HISTORY file.
Bruce Momjian [Sun, 23 Sep 2001 03:41:49 +0000 (03:41 +0000)]
Mention LIMIT change in HISTORY file.

23 years agoImplement TODO item:
Bruce Momjian [Sun, 23 Sep 2001 03:39:01 +0000 (03:39 +0000)]
Implement TODO item:

* Change LIMIT val,val to offset,limit to match MySQL

Documentation updates too.

23 years agoUpdate anoncvs info, per Marc.
Bruce Momjian [Sun, 23 Sep 2001 01:16:35 +0000 (01:16 +0000)]
Update anoncvs info, per Marc.

23 years agoUpdate FAQ.
Bruce Momjian [Sun, 23 Sep 2001 00:07:50 +0000 (00:07 +0000)]
Update FAQ.

23 years agoProvide some initial support for building the ODBC driver for
Peter Eisentraut [Sat, 22 Sep 2001 22:54:33 +0000 (22:54 +0000)]
Provide some initial support for building the ODBC driver for
an already installed iODBC or unixODBC driver manager.  In particular,
use the include files provided by the driver manager over our own,
and use the odbcinst library of the driver manager rather than gpps.c.

Migrate portability sections common to several files into psqlodbc.h.

23 years agoFix 6.4.* release dates, pointed out by Tom Lane.
Bruce Momjian [Sat, 22 Sep 2001 21:36:38 +0000 (21:36 +0000)]
Fix 6.4.* release dates, pointed out by Tom Lane.

23 years agoFix expected files.
Tatsuo Ishii [Sat, 22 Sep 2001 08:47:20 +0000 (08:47 +0000)]
Fix expected files.

23 years agoRemove test drivers
Tatsuo Ishii [Sat, 22 Sep 2001 08:44:49 +0000 (08:44 +0000)]
Remove test drivers
Also fix comment in conv.c.

23 years agosbasename $0 is now basename "$0"
Bruce Momjian [Sat, 22 Sep 2001 04:28:12 +0000 (04:28 +0000)]
sbasename $0 is now basename "$0"

23 years agoWAL file numbers: Sequenatial -> ever-increasing
Bruce Momjian [Sat, 22 Sep 2001 03:59:17 +0000 (03:59 +0000)]
WAL file numbers:  Sequenatial -> ever-increasing

23 years agoMake trim/ltrim/rtrim/btrim multibyte aware
Tatsuo Ishii [Sat, 22 Sep 2001 03:30:39 +0000 (03:30 +0000)]
Make trim/ltrim/rtrim/btrim multibyte aware
(previous commit was missing commit messages)

23 years ago*** empty log message ***
Tatsuo Ishii [Sat, 22 Sep 2001 03:26:30 +0000 (03:26 +0000)]
*** empty log message ***

23 years agoSuggest running ldconfig after installation if available.
Peter Eisentraut [Fri, 21 Sep 2001 23:20:02 +0000 (23:20 +0000)]
Suggest running ldconfig after installation if available.

23 years agoFor consistency with the rest of PostgreSQL, rename BLOBs to large objects
Peter Eisentraut [Fri, 21 Sep 2001 21:58:30 +0000 (21:58 +0000)]
For consistency with the rest of PostgreSQL, rename BLOBs to large objects
in messages and documentation.

23 years agoAdd 'reload' option to pg_ctl to send SIGHUP to the postmaster.
Tom Lane [Fri, 21 Sep 2001 21:10:56 +0000 (21:10 +0000)]
Add 'reload' option to pg_ctl to send SIGHUP to the postmaster.

23 years agoCode review for MD5 authorization patch. Clean up some breakage
Tom Lane [Fri, 21 Sep 2001 20:31:49 +0000 (20:31 +0000)]
Code review for MD5 authorization patch.  Clean up some breakage
(salts were always zero!?), add much missing documentation.

23 years agoReadd test/regress/README file, this time with a well-defined and simple
Peter Eisentraut [Fri, 21 Sep 2001 18:37:05 +0000 (18:37 +0000)]
Readd test/regress/README file, this time with a well-defined and simple
rule to remake it when necessary.

23 years agoAdd an overall timeout on the client authentication cycle, so that
Tom Lane [Fri, 21 Sep 2001 17:06:12 +0000 (17:06 +0000)]
Add an overall timeout on the client authentication cycle, so that
a hung client or lost connection can't indefinitely block a postmaster
child (not to mention the possibility of deliberate DoS attacks).
Timeout is controlled by new authentication_timeout GUC variable,
which I set to 60 seconds by default ... does that seem reasonable?

23 years agoFix type_maximum_size() to give the right answer in MULTIBYTE cases.
Tom Lane [Fri, 21 Sep 2001 15:27:38 +0000 (15:27 +0000)]
Fix type_maximum_size() to give the right answer in MULTIBYTE cases.
Avoid use of prototype-less function pointers in MB code.

23 years ago1) Avoid an overflow of connection string for Access(Microsoft Jet).
Hiroshi Inoue [Fri, 21 Sep 2001 07:51:52 +0000 (07:51 +0000)]
1) Avoid an overflow of connection string for Access(Microsoft Jet).
2) Change to retry lower version in case of "Unsupported frontend
procotol".

23 years agoProduce slightly saner-looking EXPLAIN output for a Result node.
Tom Lane [Fri, 21 Sep 2001 04:06:04 +0000 (04:06 +0000)]
Produce slightly saner-looking EXPLAIN output for a Result node.

23 years agoGive VACUUM its own GUC parameter for memory usage, rather than
Tom Lane [Fri, 21 Sep 2001 03:32:36 +0000 (03:32 +0000)]
Give VACUUM its own GUC parameter for memory usage, rather than
piggybacking on SortMem.  Add documentation for some recently-added
GUC parameters that had so far escaped it.

23 years agoQuick-and-dirty fix for recursive plpgsql functions, per bug report from
Tom Lane [Fri, 21 Sep 2001 00:11:31 +0000 (00:11 +0000)]
Quick-and-dirty fix for recursive plpgsql functions, per bug report from
Frank Miles 7-Sep-01.  This is really just sticking a finger in the dike.
Frank's case works now, but we still couldn't support a recursive function
returning a set.  Really need to restructure querytrees and execution
state so that the querytree is *read only*.  We've run into this over and
over and over again ... it has to happen sometime soon.

23 years agoSuppress gcc warning.
Tom Lane [Thu, 20 Sep 2001 23:31:08 +0000 (23:31 +0000)]
Suppress gcc warning.

23 years agoRemove some dead code and obsolete, misleading comments.
Tom Lane [Thu, 20 Sep 2001 21:07:57 +0000 (21:07 +0000)]
Remove some dead code and obsolete, misleading comments.

23 years agoDocument the sign() function.
Peter Eisentraut [Thu, 20 Sep 2001 16:41:26 +0000 (16:41 +0000)]
Document the sign() function.

23 years agoProvide tunable knob for x = NULL -> x IS NULL transformation, default to off.
Peter Eisentraut [Thu, 20 Sep 2001 14:20:28 +0000 (14:20 +0000)]
Provide tunable knob for x = NULL -> x IS NULL transformation, default to off.

23 years agoRemove old file.
Peter Eisentraut [Wed, 19 Sep 2001 21:28:55 +0000 (21:28 +0000)]
Remove old file.

23 years agoChange the version. We are moving towards the next release.
D'Arcy J.M. Cain [Wed, 19 Sep 2001 18:58:47 +0000 (18:58 +0000)]
Change the version.  We are moving towards the next release.

Fixed a nasty bug that messed up negative money amounts.

23 years agoAvoid unnecessary strcasecmp -- replace by strcmp. Fixes reported bug
Peter Eisentraut [Wed, 19 Sep 2001 15:19:12 +0000 (15:19 +0000)]
Avoid unnecessary strcasecmp -- replace by strcmp.  Fixes reported bug
that made setting serializable isolation level impossible in Turkish
locale.

23 years ago- Synced preproc.y with gram.y.
Michael Meskes [Wed, 19 Sep 2001 14:09:32 +0000 (14:09 +0000)]
- Synced preproc.y with gram.y.
- Synced pgc.l with scan.l.
- Synced keyword.c.
- Include the remaining patches by Christof Petig <christof.petig@wtal.de>.

23 years agoReplace useless strcasecmp's by strcmp's.
Peter Eisentraut [Wed, 19 Sep 2001 09:48:42 +0000 (09:48 +0000)]
Replace useless strcasecmp's by strcmp's.

23 years agoGenerate index.html as the root file name of the documentation set,
Peter Eisentraut [Tue, 18 Sep 2001 12:08:27 +0000 (12:08 +0000)]
Generate index.html as the root file name of the documentation set,
rather than making index.html a symlink to the autogenerated name.
Fixes fatal problems with tar programs that don't handle symlinks
very well (MacOS X).

(The names user.html, admin.html, etc. are still available as make
targets, but they aren't packaged anymore.)

Use the manifest file that the stylesheets generate as the file list
for packaging.  Put graphics in the right place while building, not
while packaging, so you can actually look at them after building.

23 years agoEXPLAIN ANALYZE feature to measure and show actual runtimes and tuple
Tom Lane [Tue, 18 Sep 2001 01:59:07 +0000 (01:59 +0000)]
EXPLAIN ANALYZE feature to measure and show actual runtimes and tuple
counts alongside the planner's estimates.  By Martijn van Oosterhout,
with some further work by Tom Lane.

23 years agoUnify the zip rules and variables.
Peter Eisentraut [Mon, 17 Sep 2001 23:00:27 +0000 (23:00 +0000)]
Unify the zip rules and variables.

23 years agoFix bogus failure-return value from lo_create, per report from Gavin
Tom Lane [Mon, 17 Sep 2001 20:05:47 +0000 (20:05 +0000)]
Fix bogus failure-return value from lo_create, per report from Gavin
Sherry.  Also clean up leakage of open files and LOs in failure exits
from lo_import and lo_export.

23 years agoAttached is a patch that fixes ResultSetMetaData.isNullable() in
Bruce Momjian [Mon, 17 Sep 2001 15:56:11 +0000 (15:56 +0000)]
Attached is a patch that fixes ResultSetMetaData.isNullable() in
the JDBC driver.

This method is currently unimplemented and always returns
ResultSetMetaData.columnNullable. This is obviously incorrect
when a column is defined with NOT NULL or PRIMARY KEY. And we
have to think of check constraints, views, functions etc.

The patch simply changes the return value to
ResultSetMetaData.columnNullableUnknown. This is until someone
comes up with a real implementation of course.

On Fri, 14 Sep 2001 17:53:50 +0200, Tomisaw Kity?ski wrote:
>Hello there,
>
>could someone tell me, please, do I have any chance to get
>proper implementation of above method in JDBC (1.1+) soon?
>
>Current "return 1" works fine on most tables, however it seems
>to be a little bit incorrect with some of them ;)

Ren? Pijlman

23 years agoI'm attaching a patch which fixes the corruption in strings caused
Bruce Momjian [Mon, 17 Sep 2001 15:54:50 +0000 (15:54 +0000)]
I'm attaching a patch which fixes the corruption in strings caused
by escape processing in the SQL statement. I've tested this for a
while now and it appears to work well. Previously string data
with {d was getting corrupt as the {d was being stripped regardless
of whether it was an escape code or not.

I also added checking for time and timestamp escape processing strings
as per 11.3 in the specification. The patch is against the latest
CVS.

Thomas O'Dowd

23 years agoChange vacuumlo to take 'lo' type into account.
Hiroshi Inoue [Mon, 17 Sep 2001 02:30:54 +0000 (02:30 +0000)]
Change vacuumlo to take 'lo' type into account.

23 years agoChange FixupBlobXrefs() to take 'lo' type into account.
Hiroshi Inoue [Mon, 17 Sep 2001 02:07:51 +0000 (02:07 +0000)]
Change FixupBlobXrefs() to take 'lo' type into account.

23 years agoSimplify and clean up FigureColname; make it work without coredumping
Tom Lane [Mon, 17 Sep 2001 01:06:36 +0000 (01:06 +0000)]
Simplify and clean up FigureColname; make it work without coredumping
for TypeCast case.

23 years agoUse portable putenv(), not unportable setenv().
Tom Lane [Mon, 17 Sep 2001 00:42:42 +0000 (00:42 +0000)]
Use portable putenv(), not unportable setenv().

23 years agoSuppress compiler warning.
Tom Lane [Mon, 17 Sep 2001 00:29:10 +0000 (00:29 +0000)]
Suppress compiler warning.

23 years agoUpdate documentation's info about location of our CVSROOT.
Tom Lane [Sun, 16 Sep 2001 23:52:29 +0000 (23:52 +0000)]
Update documentation's info about location of our CVSROOT.

23 years agoRussian translation from Serguei Mokhov
Peter Eisentraut [Sun, 16 Sep 2001 23:10:16 +0000 (23:10 +0000)]
Russian translation from Serguei Mokhov

23 years agoUpdate from Serguei Mokhov
Peter Eisentraut [Sun, 16 Sep 2001 23:06:33 +0000 (23:06 +0000)]
Update from Serguei Mokhov

23 years agoUpdate intro in face of TOAST.
Peter Eisentraut [Sun, 16 Sep 2001 22:53:52 +0000 (22:53 +0000)]
Update intro in face of TOAST.

23 years agoInstall dynamically loadable modules into a private subdirectory
Peter Eisentraut [Sun, 16 Sep 2001 16:11:11 +0000 (16:11 +0000)]
Install dynamically loadable modules into a private subdirectory
under libdir, for a cleaner separation in the installation layout
and compatibility with binary packaging standards.  Point backend's
default search location there.  The contrib modules are also
installed in the said location, giving them the benefit of the
default search path as well.  No changes in user interface
nevertheless.

23 years agoUpdate highlights. More items to add.
Bruce Momjian [Sun, 16 Sep 2001 04:26:22 +0000 (04:26 +0000)]
Update highlights.  More items to add.

23 years agoPoint to HISTORY File for current info.
Bruce Momjian [Sun, 16 Sep 2001 04:18:57 +0000 (04:18 +0000)]
Point to HISTORY File for current info.

23 years agoAdd list of 7.2 changes in preparation for beta.
Bruce Momjian [Sun, 16 Sep 2001 04:12:35 +0000 (04:12 +0000)]
Add list of 7.2 changes in preparation for beta.

23 years agoupdates
Peter Eisentraut [Sat, 15 Sep 2001 19:56:59 +0000 (19:56 +0000)]
updates

23 years agoMarkup examples as examples.
Peter Eisentraut [Sat, 15 Sep 2001 16:08:59 +0000 (16:08 +0000)]
Markup examples as examples.

23 years agoMarkup examples as examples. Fix formatting of examples.
Peter Eisentraut [Sat, 15 Sep 2001 00:48:59 +0000 (00:48 +0000)]
Markup examples as examples.  Fix formatting of examples.

23 years agoSome formatting changes, add CSS stylesheet. Include indexes in
Peter Eisentraut [Fri, 14 Sep 2001 20:37:55 +0000 (20:37 +0000)]
Some formatting changes, add CSS stylesheet.  Include indexes in
documentation tarball builds.

23 years ago> Here's a revised patch. Changes:
Bruce Momjian [Fri, 14 Sep 2001 17:46:40 +0000 (17:46 +0000)]
> Here's a revised patch. Changes:
>
> 1. Now outputs '\\' instead of '\134' when using encode(bytea, 'escape')
> Note that I ended up leaving \0 as \000 so that there are no ambiguities
> when decoding something like, for example, \0123.
>
> 2. Fixed bug in byteain which allowed input values which were not valid
> octals (e.g. \789), to be parsed as if they were octals.
>
> Joe
>

Here's rev 2 of the bytea string support patch. Changes:

1. Added missing declaration for MatchBytea function
2. Added PQescapeBytea to fe-exec.c
3. Applies cleanly on cvs tip from this afternoon

I'm hoping that someone can review/approve/apply this before beta starts, so
I guess I'd vote (not that it counts for much) to delay beta a few days :-)

Joe Conway

23 years agoAllow '1' in jdbc2 boolean test.
Bruce Momjian [Fri, 14 Sep 2001 17:40:41 +0000 (17:40 +0000)]
Allow '1' in jdbc2 boolean test.

23 years agoRemove --enable-unicode-conversion
Tatsuo Ishii [Fri, 14 Sep 2001 10:36:52 +0000 (10:36 +0000)]
Remove --enable-unicode-conversion
unicode-conversion is always on if --enable-multibyte is specified

   Tatsuo Ishii

23 years agoAdd semicolon
Tatsuo Ishii [Fri, 14 Sep 2001 10:28:09 +0000 (10:28 +0000)]
Add semicolon

23 years agoFix typo.
Tatsuo Ishii [Fri, 14 Sep 2001 10:24:35 +0000 (10:24 +0000)]
Fix typo.

23 years agoFix exmaple
Tatsuo Ishii [Fri, 14 Sep 2001 09:49:00 +0000 (09:49 +0000)]
Fix exmaple

    Tatsuo Ishii

23 years agoFix show message
Tatsuo Ishii [Fri, 14 Sep 2001 08:47:00 +0000 (08:47 +0000)]
Fix show message

    Tatsuo Ishii

23 years agoAdd semicolon
Tatsuo Ishii [Fri, 14 Sep 2001 08:24:29 +0000 (08:24 +0000)]
Add semicolon

    Tatsuo Ishii

23 years agoExample rules do not work.
Tatsuo Ishii [Fri, 14 Sep 2001 08:19:55 +0000 (08:19 +0000)]
Example rules do not work.

CREATE RULE bad_rule_combination_1 AS
    ON SELECT TO emp
    DO INSTEAD
        SELECT * FROM toyemp;

CREATE RULE bad_rule_combination_2 AS
    ON SELECT TO toyemp
    DO INSTEAD
        SELECT * FROM emp;

Tatsuo Ishii

23 years agoAdd semicolon
Tatsuo Ishii [Fri, 14 Sep 2001 08:05:55 +0000 (08:05 +0000)]
Add semicolon

23 years agoAdd semicolon
Tatsuo Ishii [Fri, 14 Sep 2001 07:59:31 +0000 (07:59 +0000)]
Add semicolon

    Tatsuo Ishii

23 years agoChange an *if condition*.
Hiroshi Inoue [Fri, 14 Sep 2001 06:46:45 +0000 (06:46 +0000)]
Change an *if condition*.

Hiroshi Inoue

23 years ago1) Improve the implementation of *Disallow Premature* for
Hiroshi Inoue [Fri, 14 Sep 2001 06:30:37 +0000 (06:30 +0000)]
1) Improve the implementation of *Disallow Premature* for
   older versions of servers.
2) Implement SQLProcedures.

Hiroshi Inoue

23 years agoUpdate TODO list.
Bruce Momjian [Fri, 14 Sep 2001 03:21:42 +0000 (03:21 +0000)]
Update TODO list.

23 years agoFix a coversation error with pre 6.4 versions.
Hiroshi Inoue [Fri, 14 Sep 2001 01:50:33 +0000 (01:50 +0000)]
Fix a coversation error with pre 6.4 versions.

Hiroshi Inoue

23 years agoUpdate TODO list.
Bruce Momjian [Fri, 14 Sep 2001 01:23:14 +0000 (01:23 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Fri, 14 Sep 2001 00:41:06 +0000 (00:41 +0000)]
Update TODO list.

23 years agoAdd missing paren to ODBC compiles.
Bruce Momjian [Thu, 13 Sep 2001 22:39:29 +0000 (22:39 +0000)]
Add missing paren to ODBC compiles.

23 years agoUpdate compatibility information.
Peter Eisentraut [Thu, 13 Sep 2001 19:40:34 +0000 (19:40 +0000)]
Update compatibility information.

23 years agoUpdate compatibility information.
Peter Eisentraut [Thu, 13 Sep 2001 19:05:29 +0000 (19:05 +0000)]
Update compatibility information.

23 years agoUpdate compatibility information.
Peter Eisentraut [Thu, 13 Sep 2001 18:17:44 +0000 (18:17 +0000)]
Update compatibility information.

23 years agoDidn't want that jdbc patch in there yet.
Bruce Momjian [Thu, 13 Sep 2001 17:01:31 +0000 (17:01 +0000)]
Didn't want that jdbc patch in there yet.

23 years ago> I found a problem with PQescapeString (I think). Since it escapes
Bruce Momjian [Thu, 13 Sep 2001 17:00:34 +0000 (17:00 +0000)]
> I found a problem with PQescapeString (I think). Since it escapes
> null bytes to be literally '\0', the following can happen:
> 1. User inputs string value as "<null byte>##" where ## are digits in the
> range of 0 to 7.
> 2. PQescapeString converts this to "\0##"
> 3. Escaped string is used in a context that causes "\0##" to be evaluated as
> an octal escape sequence.

I agree that this is a problem, though it is not possible to do
anything harmful with it.  In addition, it only occurs if there are
any NUL characters in its input, which is very unlikely if you are
using C strings.

The patch below addresses the issue by removing escaping of \0
characters entirely.

> If the goal is to "safely" encode null bytes, and preserve the rest of the
> string as it was entered, I think the null bytes should be escaped as \\000
> (note that if you simply use \000 the same string truncation problem
> occurs).

We can't do that, this would require 4n + 1 bytes of storage for the
result, breaking the interface.

Florian Weimer

23 years agoReplace ASCII-quotes with proper markup.
Peter Eisentraut [Thu, 13 Sep 2001 15:55:24 +0000 (15:55 +0000)]
Replace ASCII-quotes with proper markup.

23 years ago1) Not export ODBC 3.0 functions.
Hiroshi Inoue [Thu, 13 Sep 2001 00:27:11 +0000 (00:27 +0000)]
1) Not export ODBC 3.0 functions.
2) (Maybe) fix a bug reported by Mika Muntila.

23 years ago'chown -R postgres /usr/local/pgsql' is wrong. Need to do some more typing.
Peter Eisentraut [Wed, 12 Sep 2001 20:57:28 +0000 (20:57 +0000)]
'chown -R postgres /usr/local/pgsql' is wrong.  Need to do some more typing.

23 years agoAttached patch is correction for 'doc/jdbc.sgml' of PostgreSQL 7.1.3.
Bruce Momjian [Wed, 12 Sep 2001 15:55:00 +0000 (15:55 +0000)]
Attached patch is correction for 'doc/jdbc.sgml' of PostgreSQL 7.1.3.

Correction content:
  * I revised a mistake of type (copy and paste).
  * I revised multiplicity of description.

Ryouichi Matsuda

23 years agoFollowings are proposed fixes to jdbc.sgml(line numbers are for 7.1.3
Bruce Momjian [Wed, 12 Sep 2001 15:49:10 +0000 (15:49 +0000)]
Followings are proposed fixes to jdbc.sgml(line numbers are for 7.1.3
doc).

Hiroyuki Yatabe

23 years agomax_locks_per_transaction seems to be a more consistent name than
Peter Eisentraut [Wed, 12 Sep 2001 14:06:37 +0000 (14:06 +0000)]
max_locks_per_transaction seems to be a more consistent name than
max_locks_per_xact.

23 years agoCheck both -lrt and -lposix4 for fdatasync, to cover all Solaris versions.
Peter Eisentraut [Wed, 12 Sep 2001 12:14:41 +0000 (12:14 +0000)]
Check both -lrt and -lposix4 for fdatasync, to cover all Solaris versions.

23 years agoIt is not fixed and I doubt that it is working fine in current CVS. The
Bruce Momjian [Wed, 12 Sep 2001 04:01:57 +0000 (04:01 +0000)]
It is not fixed and I doubt that it is working fine in current CVS. The
bugfix is in the attached patch. Please apply it. Thanks.

 Output must be:

test=# SELECT to_char(485, 'RN');
     to_char
-----------------
         CDLXXXV
(1 row)

test=# SELECT to_char(485, 'FMRN');
 to_char
---------
 CDLXXXV
(1 row)

test=# SELECT to_char(1000, 'RN');
     to_char
-----------------
               M
(1 row)

test=# SELECT to_char(7.2, '"Welcome to"9.9 "release! :-)"');
           to_char
-----------------------------
 Welcome to 7.2 release! :-)
(1 row)

Karel Zak