Bruce Momjian [Sun, 17 Aug 2003 04:33:02 +0000 (04:33 +0000)]
The attached patch adds some clarification to the documentation of the
binary format read/created by COPY BINARY. It also mentions the
contrib/binarycopy module.
Bruce Momjian [Sun, 17 Aug 2003 03:17:22 +0000 (03:17 +0000)]
(I always forget what the magic numbers 0 through 2 means for the
"syslog" option.)
By the way: The "virtual_host" parameter is a bad name for that
particular option, I think. "Virtual host" signals that PostgreSQL will
behave differently according to which IP address it's contacted (like
Apache's virtual host support which makes the web-server serve different
sites according to different criteria). A better word for the options
would be "tcpip_listen_addr" or something like that.
Barry Lind [Fri, 15 Aug 2003 18:45:11 +0000 (18:45 +0000)]
Updated the blob regression test to actually use the getBlob/getClob methods
and test them, in addition to testing the underlying LargeObject API methods.
Tom Lane [Fri, 15 Aug 2003 00:22:26 +0000 (00:22 +0000)]
Rewrite array_cmp to not depend on deconstruct_array. Should be a little
faster, but more importantly does not leak memory. Still needs more work
though, per my recent note to pgsql-hackers.
Tom Lane [Wed, 13 Aug 2003 18:56:21 +0000 (18:56 +0000)]
libpq failed to cope with COPY FROM STDIN if the command was issued
via extended query protocol, because it sends Sync right after Execute
without realizing that the command to be executed is COPY. There seems
to be no reasonable way for it to realize that, either, so the best fix
seems to be to make the backend ignore Sync during copy-in mode. Bit of
a wart on the protocol, but little alternative. Also, libpq must send
another Sync after terminating the COPY, if the command was issued via
Execute.
Tom Lane [Wed, 13 Aug 2003 16:29:03 +0000 (16:29 +0000)]
Add PQexecPrepared() and PQsendQueryPrepared() functions, to allow
libpq users to perform Bind/Execute of previously prepared statements.
Per yesterday's discussion, this offers enough performance improvement
to justify bending the 'no new features during beta' rule.
Tom Lane [Wed, 13 Aug 2003 16:16:23 +0000 (16:16 +0000)]
Move MemoryContextCheck() call from bottom of PostgresMain loop to
just before CommitTransactionCommand(). This is a more sensible place
to put it since commit discards a lot of contexts, and we'd not find
out about stomps affecting only transaction-local contexts.
Bruce Momjian [Wed, 13 Aug 2003 03:12:04 +0000 (03:12 +0000)]
I'm quite fond of doing VPATH builds, i.e. building outside the source
tree. This also catches lots of little Makefile bugs, so here's a small
patch for one of them (replacing an explicit reference to thread.c with
a reference to it as the first prerequsite of the rule makes make look
for it in the place where it was found (the source tree) rather than in
the build tree. (using GNU make 3.79.1)
Tom Lane [Tue, 12 Aug 2003 22:42:01 +0000 (22:42 +0000)]
Include 'IPv4', 'IPv6', or 'Unix' in socket-creation failure messages,
in hopes of soothing fears of those with partial IPv6 support. Still an
open question whether we should report EAFNOSUPPORT errors at all,
though.
Tom Lane [Tue, 12 Aug 2003 18:52:38 +0000 (18:52 +0000)]
Marginal hacks to move some processing out of the per-client-message
processing loop; avoids extra overhead when using parse/bind/execute
messages instead of single Query message.
Tom Lane [Tue, 12 Aug 2003 18:23:21 +0000 (18:23 +0000)]
Change some frequently-reached elog(DEBUG...) calls to ereport(DEBUG...)
for speed reasons. (ereport falls out much more quickly when no output
is needed than elog does.)
Tom Lane [Mon, 11 Aug 2003 23:04:50 +0000 (23:04 +0000)]
Rewriter and planner should use only resno, not resname, to identify
target columns in INSERT and UPDATE targetlists. Don't rely on resname
to be accurate in ruleutils, either. This fixes bug reported by
Donald Fraser, in which renaming a column referenced in a rule did not
work very well.
Barry Lind [Mon, 11 Aug 2003 20:54:55 +0000 (20:54 +0000)]
Applied patch from Kim Ho to fix a regression against a 7.4 server. The result
of transaction isolation level changed from uppercase to lower case between 7.3 and 7.4. In testing, a regression was also fixed in this area when talking to
a 7.2 server due to changes in how notice messages are processed in the current
code.
Tom Lane [Mon, 11 Aug 2003 20:46:47 +0000 (20:46 +0000)]
Code cleanup inspired by recent resname bug report (doesn't fix the bug
yet, though). Avoid using nth() to fetch tlist entries; provide a
common routine get_tle_by_resno() to search a tlist for a particular
resno. This replaces a couple uses of nth() and a dozen hand-coded
search loops. Also, replace a few uses of nth(length-1, list) with
llast().
Tom Lane [Sun, 10 Aug 2003 19:48:08 +0000 (19:48 +0000)]
Repair potential deadlock created by recent changes to recycle btree
index pages: when _bt_getbuf asks the FSM for a free index page, it is
possible (and, in some cases, even moderately likely) that the answer
will be the same page that _bt_split is trying to split. _bt_getbuf
already knew that the returned page might not be free, but it wasn't
prepared for the possibility that even trying to lock the page could
be problematic. Fix by doing a conditional rather than unconditional
grab of the page lock.
Bruce Momjian [Fri, 8 Aug 2003 17:13:02 +0000 (17:13 +0000)]
Update as done:
< * Allow easy display of usernames in a group
> * -Allow easy display of usernames in a group
88,89d87
< * -Delay resolution of array expression type so assignment coercion
< can be performed on empty array expressions (Joe)
94c92,94
< o Support construction of array result values in expressions
> o -Support construction of array result values in expressions (Joe)
> o Delay resolution of array expression type so assignment coercion
> can be performed on empty array expressions (Joe) 148c148
< * Allow LIKE indexing optimization for non-ASCII locales
> * -Allow LIKE indexing optimization for non-ASCII locales using special index 173c173
< * Return proper effected tuple count from complex commands [return]
> * -Return proper effected tuple count from complex commands [return] 236c236
< o Allow SHOW of non-modifiable variables, like pg_controldata
> o -Allow SHOW of some non-modifiable variables, like pg_controldata 257a258
> o Add capability to create and call PROCEDURES 272c273
<
> * Allow psql \du to show groups, and add \dg for groups 424c425
< * Improve Subplan list handling
> * -Improve Subplan list handling
Bruce Momjian [Fri, 8 Aug 2003 16:34:40 +0000 (16:34 +0000)]
Done:
< o Allow array declarations and other data types in PL/PgSQL DECLARE
> o -Allow array declarations and other data types in PL/PgSQL DECLARE 254c254
< o Allow PL/PgSQL to support array element assignment
> o -Allow PL/PgSQL to support array element assignment (Joe)
Bruce Momjian [Fri, 8 Aug 2003 02:00:20 +0000 (02:00 +0000)]
Update done items:
< * Allow elog() to return error codes, module name, file name, line
< number, not just messages (Peter E)
< * Add error codes (Peter E)
< * Make error messages more consistent [error]
> * -Allow elog() to return error codes, module name, file name, line
> number, not just messages (Tom)
> * -Add error codes (Tom)
> * -Make error messages more consistent
40c40
< * Add GUC log_statement_and_duration to print statement and >= min duration
> * -Add GUC log_statement_and_duration to print statement and >= min duration
84c84
< * Allow current datestyle to restrict dates; prevent month/day swapping
> * -Allow current datestyle to restrict dates; prevent month/day swapping
86c86
< * Prevent month/day swapping of ISO dates to make invalid dates valid
> * -Prevent month/day swapping of ISO dates to make invalid dates valid
88c88
< * Delay resolution of array expression type so assignment coercion
> * -Delay resolution of array expression type so assignment coercion
93c93
< o Allow arrays to be ORDER'ed
> o -Allow arrays to be ORDER'ed 116c116
< * Remove Cyrillic recode support
> * -Remove Cyrillic recode support 144c144
< * Certain indexes will not shrink, e.g. indexes on ever-increasing
> * -Certain indexes will not shrink, e.g. indexes on ever-increasing 185c185
< * Have SELECT '13 minutes'::interval display zero seconds in ISO datestyle
> * -Have SELECT '13 minutes'::interval display zero seconds in ISO datestyle 196c196
< o -Add ALTER TABLE tab SET WITHOUT OIDS (Rod)
> o --Add ALTER TABLE tab SET WITHOUT OIDS (Rod) 221c221
< stored in the backend
> stored in the backend (Gavin) 235c235
< o Allow EXPLAIN EXECUTE to see prepared plans
> o -Allow EXPLAIN EXECUTE to see prepared plans 241d240
< o Add untrusted version of plpython 265c264
< * Allow psql to show transaction status if backend protocol changes made
> * -Allow psql to show transaction status if backend protocol changes made
272,273c271,272
< * Modify pg_get_triggerdef() to take a boolean to pretty-print,
< and use that as part of pg_dump along with psql
> * -Modify pg_get_triggerdef() to take a boolean to pretty-print,
> and use that as part of pg_dump along with psql 292c291
< o Add SQLSTATE
> o -Add SQLSTATE 296c295
< o Implement SQLDA (do we really need this?)
> o -Implement SQLDA 364d362
< * Allow binding query args over FE/BE protocol 378c376,377
< * Provide automatic running of vacuum in the background (Tom) [vacuum]
> * Provide automatic running of vacuum in the background in backend
> rather than in /contrib [vacuum] 427c426
< * Allow Subplans to use efficient joins(hash, merge) with upper variable
> * -Allow Subplans to use efficient joins(hash, merge) with upper variable 429c428
< * Allow merge and hash joins on expressions not just simple variables (Tom)
> * -Allow merge and hash joins on expressions not just simple variables (Tom) 474c473
< * Remove memory/file descriptor freeing befor elog(ERROR) (Bruce)
> * Remove memory/file descriptor freeing before ereport(ERROR) (Bruce)
489,490c488,489
< o Show transaction status in psql
< o Allow binding of query parameters, support for prepared queries
> o -Show transaction status in psql
> o -Allow binding of query parameters, support for prepared queries
492,494c491,493
< o Remove hard-coded limits on user/db/password names
< o Remove unused elements of startup packet (unused, tty, passlength)
< o Fix COPY/fastpath protocol?
> o -Remove hard-coded limits on user/db/password names
> o -Remove unused elements of startup packet (unused, tty, passlength)
> o -Fix COPY/fastpath protocol
496,497c495
< o Replication support?
< o Error codes
> o -Error codes 500d497
< o ecpg improvements? 503c500
< o Report server version number, database encoding, client encoding
> o -Report server version number, database encoding, client encoding
Tom Lane [Fri, 8 Aug 2003 00:10:31 +0000 (00:10 +0000)]
Fix floating-point timestamp comparisons to not go nuts if NaN is
encountered; per bug report from Christian van der Leeden 8/7/03.
Also, adjust larger/smaller routines (MAX/MIN) to share code with
comparisons for timestamp, interval, timetz.
Tom Lane [Thu, 7 Aug 2003 21:11:58 +0000 (21:11 +0000)]
Don't assume that struct option is available just because we can find a
getopt_long(). This is more or less the same problem as we saw earlier
with getaddrinfo() and struct addrinfo, and for the same reason: random
user-added libraries might contain the subroutine, but there's no
guarantee we will find the matching header files.
Tom Lane [Thu, 7 Aug 2003 19:20:24 +0000 (19:20 +0000)]
create_unique_plan() should not discard existing output columns of the
subplan it starts with, as they may be needed at upper join levels.
See comments added to code for the non-obvious reason why. Per bug report
from Robert Creager.
Tom Lane [Thu, 7 Aug 2003 16:45:21 +0000 (16:45 +0000)]
Don't use HAVE_STRUCT_ADDRINFO as a guide to whether netdb.h defines
macros like AI_NUMERICHOST; instead, test the macros individually.
Should fix recent reports of trouble on AIX and Unixware.