]> granicus.if.org Git - postgresql/log
postgresql
21 years agoDone in previous release:
Bruce Momjian [Wed, 2 Apr 2003 22:44:50 +0000 (22:44 +0000)]
Done in previous release:

< * Add --port flag to regression tests

21 years agoFix buffer overrun in to_ascii(), per report from Guido Notari.
Tom Lane [Wed, 2 Apr 2003 21:07:59 +0000 (21:07 +0000)]
Fix buffer overrun in to_ascii(), per report from Guido Notari.

21 years agoFix convSockAddr6to4(): eliminate bogus assumptions about byte ordering,
Tom Lane [Wed, 2 Apr 2003 20:00:21 +0000 (20:00 +0000)]
Fix convSockAddr6to4(): eliminate bogus assumptions about byte ordering,
remove useless SockAddr_ntop() call.  Per report from Andreas Pflug.

21 years agoFix platform-dependent failure introduced by recent to_char changes
Tom Lane [Wed, 2 Apr 2003 02:33:52 +0000 (02:33 +0000)]
Fix platform-dependent failure introduced by recent to_char changes
(ye good olde uninitialized-local-variable).

21 years agoThis is a derived file and should never have been added to CVS.
Tom Lane [Wed, 2 Apr 2003 00:58:08 +0000 (00:58 +0000)]
This is a derived file and should never have been added to CVS.

21 years agoWhack getaddrinfo() patch around until it works, more or less, on
Tom Lane [Wed, 2 Apr 2003 00:49:28 +0000 (00:49 +0000)]
Whack getaddrinfo() patch around until it works, more or less, on
machines without IPv6.  Or at least it works on HPUX 10.20 ...

21 years agoAdd missing semicolon.
Tom Lane [Tue, 1 Apr 2003 23:42:55 +0000 (23:42 +0000)]
Add missing semicolon.

21 years agoMore patches for informix compatibility.
Michael Meskes [Tue, 1 Apr 2003 14:37:25 +0000 (14:37 +0000)]
More patches for informix compatibility.

21 years agoUpdate Russian FAQ, from Viktor Vislobokov
Bruce Momjian [Tue, 1 Apr 2003 03:10:44 +0000 (03:10 +0000)]
Update Russian FAQ, from Viktor Vislobokov

21 years agoPlease, apply patch for contrib/ltree to current CVS and 7.3.2
Bruce Momjian [Mon, 31 Mar 2003 20:53:45 +0000 (20:53 +0000)]
Please, apply patch for contrib/ltree to current CVS and 7.3.2

CHANGES

Mar 28, 2003
    Added finctions index(ltree,ltree,offset), text2ltree(text),
                    ltree2text(text)

Teodor Sigaev

21 years agoSkip START WITH in sequence definition when it's the default value --
Bruce Momjian [Mon, 31 Mar 2003 20:48:45 +0000 (20:48 +0000)]
Skip START WITH in sequence definition when it's the default value --
and hasn't been called yet.

Fixes bug where it wasn't supplied (due to being NULL).

Rod Taylor

21 years agoThe following patch cleans up the deferred trigger mechanism. There is
Bruce Momjian [Mon, 31 Mar 2003 20:47:51 +0000 (20:47 +0000)]
The following patch cleans up the deferred trigger mechanism.  There is
an unneeded memory context and some variables that are not used anymore.
It's pretty trivial and the regression tests pass fine.  There's no
change in functionality, only deletion of unused code.  I left an empty
function because maybe I'll need it for nested transactions.

Alvaro Herrera

21 years agoTestConfiguration returns int, not bool. This mistake is relatively
Tom Lane [Mon, 31 Mar 2003 20:32:29 +0000 (20:32 +0000)]
TestConfiguration returns int, not bool.  This mistake is relatively
harmless on signed-char machines but would lead to core dump in the
deadlock detection code if char is unsigned.  Amazingly, this bug has
been here since 7.1 and yet wasn't reported till now.  Thanks to Robert
Bruccoleri for providing the opportunity to track it down.

21 years agoFix broken markup.
Tom Lane [Sun, 30 Mar 2003 21:48:37 +0000 (21:48 +0000)]
Fix broken markup.

21 years agoProperly document default value of log_min_error_statement in postgresql.conf.
Bruce Momjian [Sun, 30 Mar 2003 21:38:02 +0000 (21:38 +0000)]
Properly document default value of log_min_error_statement in postgresql.conf.

21 years agoForgot two new files and one that was moved.
Michael Meskes [Sun, 30 Mar 2003 13:26:09 +0000 (13:26 +0000)]
Forgot two new files and one that was moved.

21 years agoMoved Informix stuff to its own compat library. Interval datetype is now fully functi...
Michael Meskes [Sun, 30 Mar 2003 11:48:19 +0000 (11:48 +0000)]
Moved Informix stuff to its own compat library. Interval datetype is now fully functional.

21 years agoSimplify the socket handling code by supplying a replacement getaddrinfo()
Peter Eisentraut [Sat, 29 Mar 2003 11:31:52 +0000 (11:31 +0000)]
Simplify the socket handling code by supplying a replacement getaddrinfo()
function if the OS doesn't provide one.

21 years ago[ Backpatch to 7.3.X.]
Bruce Momjian [Sat, 29 Mar 2003 05:00:15 +0000 (05:00 +0000)]
[ Backpatch to 7.3.X.]

SSL_read/write can error needing ERROR_WANT_READ or ERROR_WANT_WRITE.

21 years ago[ Backpatch to 7.3.X.]
Bruce Momjian [Sat, 29 Mar 2003 03:56:44 +0000 (03:56 +0000)]
[ Backpatch to 7.3.X.]

typing error in src/backend/libpq/be-secure.c ???

Long Description
In src/backend/libpq/be-secure.c: secure_write
on SSL_ERROR_WANT_WRITE call secure_read instead
secure_write again. May be is this a typing error?

Sergey N. Yatskevich (syatskevich@n21lab.gosniias.msk.ru)

21 years agoAdd code to apply some simple sanity checks to the header fields of a
Tom Lane [Fri, 28 Mar 2003 20:17:13 +0000 (20:17 +0000)]
Add code to apply some simple sanity checks to the header fields of a
page when it's read in, per pghackers discussion around 17-Feb.  Add a
GUC variable zero_damaged_pages that causes the response to be a WARNING
followed by zeroing the page, rather than the normal ERROR; this is per
Hiroshi's suggestion that there needs to be a way to get at the data
in the rest of the table.

21 years agoRegression output didn't get updated to match recent commit.
Tom Lane [Fri, 28 Mar 2003 17:24:28 +0000 (17:24 +0000)]
Regression output didn't get updated to match recent commit.

21 years agoFix NULL casting warning, pointed out by Joe Conway
Bruce Momjian [Fri, 28 Mar 2003 16:34:50 +0000 (16:34 +0000)]
Fix NULL casting warning, pointed out by Joe Conway

21 years agoFix number of columns in SGML, from Weiping He
Bruce Momjian [Fri, 28 Mar 2003 00:23:29 +0000 (00:23 +0000)]
Fix number of columns in SGML, from Weiping He

21 years agoFix bogus coding of SET DEFAULT ri triggers ... or at least make it less
Tom Lane [Thu, 27 Mar 2003 19:25:40 +0000 (19:25 +0000)]
Fix bogus coding of SET DEFAULT ri triggers ... or at least make it less
bogus than it was.  Per bug report from Adrian Pop.

21 years agoAdd <stdlib> to add calloc() prototype.
Bruce Momjian [Thu, 27 Mar 2003 17:25:34 +0000 (17:25 +0000)]
Add <stdlib> to add calloc() prototype.

21 years agoFix syntax error in to_char fixes --- was defining variable in main code
Bruce Momjian [Thu, 27 Mar 2003 17:10:55 +0000 (17:10 +0000)]
Fix syntax error in to_char fixes --- was defining variable in main code
block, ala C++.

21 years agoIt may not be obvious to you, but the plpython regression tests
Bruce Momjian [Thu, 27 Mar 2003 16:58:21 +0000 (16:58 +0000)]
It may not be obvious to you, but the plpython regression tests
include output that vary depending on the python build one is
running. Basically, the order of keys in a dictionary is
non-deterministic, and that part of the test fails for me regularly.

I rewrote the test to work around this problem, and include a patch
file with that change and the change to the expected otuput as well.

Mike Meyer

21 years agoNew \d format:
Bruce Momjian [Thu, 27 Mar 2003 16:57:39 +0000 (16:57 +0000)]
New \d format:

Example:

test=# \d test
     Table "public.test"
 Column |  Type   | Modifiers
--------+---------+-----------
 a      | integer | not null
Indexes:
    "test_pkey" PRIMARY KEY btree (a)
Check Constraints:
    "$2" CHECK (a > 1)
Foreign Key Constraints:
    "$1" FOREIGN KEY (a) REFERENCES parent(b)
Rules:
    myrule AS ON INSERT TO test DO INSTEAD NOTHING
Triggers:
    "asdf asdf" AFTER INSERT OR DELETE ON test FOR EACH STATEMENT EXECUTE
PROCEDURE update_pg_pwd_and_pg_group(),
    mytrigger AFTER INSERT OR DELETE ON test FOR EACH ROW EXECUTE PROCEDURE
update_pg_pwd_and_pg_group()

I have minimised the double quoting of identifiers as much as I could
easily, and I will submit another patch when I have time to work on it that
will use a 'fmtId' function to determine it exactly.

I think it's a significant improvement in legibility...

Obviously the table example above is slightly degenerate in that not many
tables in production have heaps of (non-constraint) triggers and rules.

Christopher Kings-Lynne

21 years agoAdd new file.
Bruce Momjian [Thu, 27 Mar 2003 16:55:11 +0000 (16:55 +0000)]
Add new file.

21 years agoAdd new files.
Bruce Momjian [Thu, 27 Mar 2003 16:53:15 +0000 (16:53 +0000)]
Add new files.

21 years agoThis patch implements holdable cursors, following the proposal
Bruce Momjian [Thu, 27 Mar 2003 16:51:29 +0000 (16:51 +0000)]
This patch implements holdable cursors, following the proposal
(materialization into a tuple store) discussed on pgsql-hackers earlier.
I've updated the documentation and the regression tests.

Notes on the implementation:

- I needed to change the tuple store API slightly -- it assumes that it
won't be used to hold data across transaction boundaries, so the temp
files that it uses for on-disk storage are automatically reclaimed at
end-of-transaction. I added a flag to tuplestore_begin_heap() to control
this behavior. Is changing the tuple store API in this fashion OK?

- in order to store executor results in a tuple store, I added a new
CommandDest. This works well for the most part, with one exception: the
current DestFunction API doesn't provide enough information to allow the
Executor to store results into an arbitrary tuple store (where the
particular tuple store to use is chosen by the call site of
ExecutorRun). To workaround this, I've temporarily hacked up a solution
that works, but is not ideal: since the receiveTuple DestFunction is
passed the portal name, we can use that to lookup the Portal data
structure for the cursor and then use that to get at the tuple store the
Portal is using. This unnecessarily ties the Portal code with the
tupleReceiver code, but it works...

The proper fix for this is probably to change the DestFunction API --
Tom suggested passing the full QueryDesc to the receiveTuple function.
In that case, callers of ExecutorRun could "subclass" QueryDesc to add
any additional fields that their particular CommandDest needed to get
access to. This approach would work, but I'd like to think about it for
a little bit longer before deciding which route to go. In the mean time,
the code works fine, so I don't think a fix is urgent.

- (semi-related) I added a NO SCROLL keyword to DECLARE CURSOR, and
adjusted the behavior of SCROLL in accordance with the discussion on
-hackers.

- (unrelated) Cleaned up some SGML markup in sql.sgml, copy.sgml

Neil Conway

21 years agoRemove email file.
Bruce Momjian [Thu, 27 Mar 2003 16:47:58 +0000 (16:47 +0000)]
Remove email file.

21 years agoDone:
Bruce Momjian [Thu, 27 Mar 2003 16:47:10 +0000 (16:47 +0000)]
Done:

>  o -Allow cursors outside transactions

21 years ago* Make pg_get_triggerdef documentation consistent with other pg_get_
Bruce Momjian [Thu, 27 Mar 2003 16:45:51 +0000 (16:45 +0000)]
* Make pg_get_triggerdef documentation consistent with other pg_get_
functions
* Document pg_conversion_is_visible() which was created in one of my
previous patches and didn't get documented for some reason

Christopher Kings-Lynne

21 years agoAttached are two patches for psql's tab-completion.c.
Bruce Momjian [Thu, 27 Mar 2003 16:45:01 +0000 (16:45 +0000)]
Attached are two patches for psql's tab-completion.c.

The first cleans up a couple of minor errors and ommissions
and adds tab completion support to more slash commands, e.g.
\dv.

The second is an attempt to add tab completion for schemas
and fully qualified relation names (e.g. public.mytable ).
I think this covers the TODO-item:
"Allow psql to do table completion for SELECT * FROM schema_part and table
completion for SELECT * FROM schema_name."

This happens via union selects querying:
 - relation_name in current search path;
 - schema_name;
 - schema.relation_name
matching the current input string.

E.g:
  SELECT p[TAB]
will produce a list of all appropriate relation names in the current search
path which begin with 'p', and also all schema names which begin with 'p';
  \d pub[TAB]
will produce any relation names in the current search path and also
any schema names beginning with 'pub';
  \d public.[TAB]
will produce a list of all relations in the schema 'public';
  \d public.my[TAB]
produces all relation names beginning with 'my' in schema 'public'.

It seems to work for me; comments, suggestions, particularly regarding
the coding and queries, are very welcome.

Note that tables, indexes, views and sequences relations in the
'pg_catalog' namespace are excluded even though they are in
the current search path. I found not doing this produced annoying behaviour
when expanding names beginning with 'p'. People who work with system
tables a lot may not like this though; I can look for another solution
if necessary.

Ian Barwick

21 years agoAdds
Bruce Momjian [Thu, 27 Mar 2003 16:43:07 +0000 (16:43 +0000)]
Adds

ALTER TABLE foo CLUSTER ON bar;

In pg_dumps, to preserve cluster settings.

Christopher Kings-Lynne

21 years agoReorder language creation to restore constistently.
Bruce Momjian [Thu, 27 Mar 2003 16:39:17 +0000 (16:39 +0000)]
Reorder language creation to restore constistently.

Laurent FAILLIE

21 years agoto_char fixes, Karel Zak
Bruce Momjian [Thu, 27 Mar 2003 16:35:31 +0000 (16:35 +0000)]
to_char fixes, Karel Zak

21 years agoGetTupleForTrigger must use outer transaction's command counter for time
Tom Lane [Thu, 27 Mar 2003 14:33:11 +0000 (14:33 +0000)]
GetTupleForTrigger must use outer transaction's command counter for time
qual checking, not GetCurrentCommandId.  Per test case from Steve Wolfe.

21 years agoMore changes to pgtypeslib and set optimization to -O1.
Michael Meskes [Thu, 27 Mar 2003 14:29:17 +0000 (14:29 +0000)]
More changes to pgtypeslib and set optimization to -O1.

21 years agoMerge documentation into one book. (Build with "make html".) Replace
Peter Eisentraut [Tue, 25 Mar 2003 16:15:44 +0000 (16:15 +0000)]
Merge documentation into one book.  (Build with "make html".)  Replace
vague cross-references with real links.

21 years agoI mean, bison 1.875. :-)
Bruce Momjian [Tue, 25 Mar 2003 15:11:53 +0000 (15:11 +0000)]
I mean, bison 1.875.  :-)

21 years agoMention bison 1.85 is now required.
Bruce Momjian [Tue, 25 Mar 2003 14:58:15 +0000 (14:58 +0000)]
Mention bison 1.85 is now required.

21 years agoplpgsql can assign to subscripted variables now, e.g.
Tom Lane [Tue, 25 Mar 2003 03:16:41 +0000 (03:16 +0000)]
plpgsql can assign to subscripted variables now, e.g.
x[42] := whatever;
The facility is pretty primitive because it doesn't do array slicing and
it has the same semantics as array update in SQL (array must already
be non-null, etc).  But it's a start.

21 years agopatch to notify listeners on error from Csaba Nagy
Dave Cramer [Tue, 25 Mar 2003 02:46:23 +0000 (02:46 +0000)]
patch to notify listeners on error from Csaba Nagy

21 years agoUse PQfreemem() consistently, and document its use for Notify.
Bruce Momjian [Tue, 25 Mar 2003 02:44:36 +0000 (02:44 +0000)]
Use PQfreemem() consistently, and document its use for Notify.

Keep PQfreeNotify() around for binary compatibility.

21 years agoadded DISTINCT to the query to get cross reference. This is required when two columns...
Dave Cramer [Tue, 25 Mar 2003 02:28:45 +0000 (02:28 +0000)]
added DISTINCT to the query to get cross reference. This is required when two columns in a table are both foreign keys to another table. From Peter Royal proyal@pace2020.com

21 years agofixed problem where information from previous updates was leaking into subsequent...
Dave Cramer [Tue, 25 Mar 2003 02:24:07 +0000 (02:24 +0000)]
fixed problem where information from previous updates was leaking into subsequent updates patch from Shawn Green, slightly modified

21 years agoFactor out duplicate code for computing values of PLpgSQL_datum items.
Tom Lane [Tue, 25 Mar 2003 00:34:24 +0000 (00:34 +0000)]
Factor out duplicate code for computing values of PLpgSQL_datum items.
This is to help localize the changes needed for adding a new kind of
PLpgSQL_datum (like, say, an array element...)

21 years agoIgnore SIGXFSZ (if platform has it), so that ulimit violations work like
Tom Lane [Mon, 24 Mar 2003 22:40:14 +0000 (22:40 +0000)]
Ignore SIGXFSZ (if platform has it), so that ulimit violations work like
disk-full conditions instead of provoking a backend crash.  Per suggestion
from Frederic Surleau.

21 years agoModify keys_are_unique optimization to release buffer pins before it
Tom Lane [Mon, 24 Mar 2003 21:42:33 +0000 (21:42 +0000)]
Modify keys_are_unique optimization to release buffer pins before it
returns NULL.  This avoids out-of-buffers failures during many-way
indexscans, as in Shraibman's complaint of 21-Mar.

21 years agoAdd comment to postgresql.conf:
Bruce Momjian [Mon, 24 Mar 2003 20:39:10 +0000 (20:39 +0000)]
Add comment to postgresql.conf:

#search_path = '$user,public'   # schema names

21 years agoPrevent multiple queries in a single string into a single transaction
Bruce Momjian [Mon, 24 Mar 2003 18:33:52 +0000 (18:33 +0000)]
Prevent multiple queries in a single string into a single transaction
when autocommit is off, and document grouping when autocommit is on.

21 years agoAnother big editing pass for consistent content and presentation.
Peter Eisentraut [Mon, 24 Mar 2003 14:32:51 +0000 (14:32 +0000)]
Another big editing pass for consistent content and presentation.

21 years agoNew build number
Barry Lind [Mon, 24 Mar 2003 04:01:02 +0000 (04:01 +0000)]
New build number

21 years agoFix bug in reading acls (didn't treat null acl as meaning the table owner had
Barry Lind [Mon, 24 Mar 2003 03:48:32 +0000 (03:48 +0000)]
Fix bug in reading acls (didn't treat null acl as meaning the table owner had
full privs), also updated the regression test for this case.

 Modified Files:
  jdbc/org/postgresql/jdbc1/AbstractJdbc1DatabaseMetaData.java
  jdbc/org/postgresql/test/jdbc2/DatabaseMetaDataTest.java

21 years agoAdjust amrescan code so that it's allowed to call index_rescan with a
Tom Lane [Sun, 23 Mar 2003 23:01:03 +0000 (23:01 +0000)]
Adjust amrescan code so that it's allowed to call index_rescan with a
NULL key pointer, indicating that the existing scan key should be reused.
This behavior isn't used yet but will be needed for my planned fix to
the keys_are_unique code.

21 years agoRename README in autovacuum code to match Makefile.
Bruce Momjian [Sun, 23 Mar 2003 20:16:06 +0000 (20:16 +0000)]
Rename README in autovacuum code to match Makefile.

21 years agoInstead of storing pg_statistic stavalues entries as text strings, store
Tom Lane [Sun, 23 Mar 2003 05:14:37 +0000 (05:14 +0000)]
Instead of storing pg_statistic stavalues entries as text strings, store
them as arrays of the internal datatype.  This requires treating the
stavalues columns as 'anyarray' rather than 'text[]', which is not 100%
kosher but seems to work fine for the purposes we need for pg_statistic.
Perhaps in the future 'anyarray' will be allowed more generally.

21 years agoTweak selectivity and related routines to cope with domains. Per report
Tom Lane [Sun, 23 Mar 2003 01:49:02 +0000 (01:49 +0000)]
Tweak selectivity and related routines to cope with domains.  Per report
from Andreas Pflug.

21 years agoRemove ecpg #warning with approval from Michael.
Bruce Momjian [Sat, 22 Mar 2003 19:48:41 +0000 (19:48 +0000)]
Remove ecpg #warning with approval from Michael.

21 years agoAdd missing #include for calloc in ecpg.
Bruce Momjian [Sat, 22 Mar 2003 17:12:23 +0000 (17:12 +0000)]
Add missing #include for calloc in ecpg.

21 years agoDepartment of second thoughts: probably shouldn't use nth() to get the
Tom Lane [Sat, 22 Mar 2003 17:11:25 +0000 (17:11 +0000)]
Department of second thoughts: probably shouldn't use nth() to get the
appropriate targetlist entry out of the subquery.  Use an explicit search
like we do everywhere else.

21 years agoFix comment-only query to return Null result set, rather than nothing.
Bruce Momjian [Sat, 22 Mar 2003 04:23:34 +0000 (04:23 +0000)]
Fix comment-only query to return Null result set, rather than nothing.

Cleans up blank query handling to be more consistent.

21 years agoAdd PQfreemem() call for Win32.
Bruce Momjian [Sat, 22 Mar 2003 03:29:06 +0000 (03:29 +0000)]
Add PQfreemem() call for Win32.

21 years agoBack out to_char fixes until regression tests are fixed.
Bruce Momjian [Sat, 22 Mar 2003 02:12:24 +0000 (02:12 +0000)]
Back out to_char fixes until regression tests are fixed.

21 years agoAdjust subquery qual pushdown rules to be more forgiving: if a qual
Tom Lane [Sat, 22 Mar 2003 01:49:38 +0000 (01:49 +0000)]
Adjust subquery qual pushdown rules to be more forgiving: if a qual
refers to a non-DISTINCT output column of a DISTINCT ON subquery, or
if it refers to a function-returning-set, we cannot push it down.
But the old implementation refused to push down *any* quals if the
subquery had any such 'dangerous' outputs.  Now we just look at the
output columns actually referenced by each qual expression.  More code
than before, but probably no slower since we don't make unnecessary checks.

21 years agoFix compile warning.
Tom Lane [Fri, 21 Mar 2003 23:18:52 +0000 (23:18 +0000)]
Fix compile warning.

21 years agoAdd hostmask() function:
Bruce Momjian [Fri, 21 Mar 2003 21:54:29 +0000 (21:54 +0000)]
Add hostmask() function:

+       <entry><function>hostmask</function>(<type>inet</type>)</entry>
+       <entry><type>inet</type></entry>
+       <entry>construct hostmask for network</entry>
+       <entry><literal>hostmask('192.168.23.20/30')</literal></entry>
+       <entry><literal>0.0.0.3</literal></entry>

Greg Wickham

21 years agoMake "win" a separate port from "cygwin". This means you can now
Peter Eisentraut [Fri, 21 Mar 2003 17:18:34 +0000 (17:18 +0000)]
Make "win" a separate port from "cygwin".  This means you can now
configure under native Windows (MinGW that is), but you won't get very far
compiling yet.  The dynaloader files are from Jan Wieck's patch set.

21 years agoRemove mention of transactions for insensitive cursors.
Bruce Momjian [Fri, 21 Mar 2003 17:11:46 +0000 (17:11 +0000)]
Remove mention of transactions for insensitive cursors.

21 years agoFix ALTER TABLE / CLUSTER ON breakage.
Bruce Momjian [Fri, 21 Mar 2003 15:43:02 +0000 (15:43 +0000)]
Fix ALTER TABLE / CLUSTER ON breakage.

Alvaro Herrera1

21 years agoDo not free prepares statements at the end of a transaction.
Michael Meskes [Fri, 21 Mar 2003 15:31:04 +0000 (15:31 +0000)]
Do not free prepares statements at the end of a transaction.

21 years agoChanges to the parser to accept new datatypes.
Michael Meskes [Fri, 21 Mar 2003 14:17:47 +0000 (14:17 +0000)]
Changes to the parser to accept new datatypes.

21 years agoAllow error query to start transaction in autocommit off mode.
Bruce Momjian [Fri, 21 Mar 2003 04:33:15 +0000 (04:33 +0000)]
Allow error query to start transaction in autocommit off mode.

21 years agoFix sequence casting.
Bruce Momjian [Fri, 21 Mar 2003 03:55:21 +0000 (03:55 +0000)]
Fix sequence casting.

21 years agoDone:
Bruce Momjian [Fri, 21 Mar 2003 03:47:38 +0000 (03:47 +0000)]
Done:

> * -Change NUMERIC data type to use base 10,000 internally

21 years agoImprove psql comment coding.
Bruce Momjian [Fri, 21 Mar 2003 03:28:29 +0000 (03:28 +0000)]
Improve psql comment coding.

21 years agoReimplement NUMERIC datatype using base-10000 arithmetic; also improve
Tom Lane [Fri, 21 Mar 2003 01:58:05 +0000 (01:58 +0000)]
Reimplement NUMERIC datatype using base-10000 arithmetic; also improve
some of the algorithms for higher functions.  I see about a factor of ten
speedup on the 'numeric' regression test, but it's unlikely that that test
is representative of real-world applications.
initdb forced due to change of on-disk representation for NUMERIC.

21 years agoFix multiline C comments in psql.
Bruce Momjian [Thu, 20 Mar 2003 22:08:50 +0000 (22:08 +0000)]
Fix multiline C comments in psql.

21 years agoUpdate for new serial functionality.
Bruce Momjian [Thu, 20 Mar 2003 21:00:49 +0000 (21:00 +0000)]
Update for new serial functionality.

21 years agoThis is not the only place in the system catalogs where NULL is
Bruce Momjian [Thu, 20 Mar 2003 20:05:32 +0000 (20:05 +0000)]
This is not the only place in the system catalogs where NULL is
effectively used to mean a default value that could also be spelled
out explicitly.  (ACLs behave that way, and useconfig/datconfig
do too IIRC.)

It's a bit of a hack, but it saves table space and backend code ---
without this convention the default would have to be inserted "manually"
since we have no mechanism to supply defaults when C code is forming a
new catalog tuple.

I'm inclined to leave the code alone.  But Alvaro is right that it'd be
good to point out the 'infinity' option in the CREATE USER and ALTER
USER man pages.  (Doc patch please?)

Alvaro Herrera

21 years agoThe documentation for SELECT is incorrect in a sense: the syntax for a
Bruce Momjian [Thu, 20 Mar 2003 19:00:01 +0000 (19:00 +0000)]
The documentation for SELECT is incorrect in a sense: the syntax for a
join is defined as:

from_item [ NATURAL ] join_type from_item
    [ ON join_condition | USING ( join_column_list ) ]

However, if the join_type is an INNER or OUTER join, an ON, USING, or
NATURAL clause *must* be specified (it's not optional, as that segment
of the docs suggest).

I'm not exactly sure what the best way to fix this is, so I've attached
a patch adding a FIXME comment to the relevant section of the SGML. If
anyone has any ideas on the proper way to outline join syntax, please
speak up.

Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC

21 years agoplease apply attached patch to current CVS.
Bruce Momjian [Thu, 20 Mar 2003 18:59:18 +0000 (18:59 +0000)]
please apply attached patch to current CVS.
btree_gist now supports int2 !
Thanks Janko Richter for contribution.

21 years agoThis patch creates a function named pg_get_triggerdef that takes the oid of
Bruce Momjian [Thu, 20 Mar 2003 18:58:02 +0000 (18:58 +0000)]
This patch creates a function named pg_get_triggerdef that takes the oid of
a trigger as its parameter.  It is basically copied from the pg_dump
code.

Christopher Kings-Lynne

21 years agoNow that the CLUSTER ALL machinery is in place, the clusterdb script can
Bruce Momjian [Thu, 20 Mar 2003 18:53:18 +0000 (18:53 +0000)]
Now that the CLUSTER ALL machinery is in place, the clusterdb script can
be simplified (I'd thought that it can even be removed).  This patch
does that.

Alvaro Herrera

21 years agoAdd ALTER TABLE <tablename> CLUSTER ON <indexname>
Bruce Momjian [Thu, 20 Mar 2003 18:52:48 +0000 (18:52 +0000)]
Add ALTER TABLE <tablename> CLUSTER ON <indexname>

Alvaro Herrera

21 years agoIt would also be handy if users could see their own pg_stat_activity
Bruce Momjian [Thu, 20 Mar 2003 18:51:16 +0000 (18:51 +0000)]
It would also be handy if users could see their own pg_stat_activity
queries while the rest remain blank.

Kevin Brown

21 years agoI have updated my pg_autovacuum program (formerly pg_avd, the name
Bruce Momjian [Thu, 20 Mar 2003 18:14:46 +0000 (18:14 +0000)]
I have updated my pg_autovacuum program (formerly pg_avd, the name
changed as per discussion on the patches list).

This version should be a good bit better.  It addresses all the issues
pointed out by Neil Conway. Vacuum and Analyze are now handled
separately.  It now monitors for xid wraparound.  The number of database
connections and queries has been significantly reduced compared the
previous version.  I have moved it from bin to contrib.  More detail on
the changes are in the TODO file.

I have not tested the xid wraparound code as I have to let my AthlonXP
1600 run select 1 in a tight loop for approx. two days in order to
perform the required 500,000,000 xacts.

Matthew T. O'Connor

21 years agoFix typo in SGML tags.
Bruce Momjian [Thu, 20 Mar 2003 17:37:46 +0000 (17:37 +0000)]
Fix typo in SGML tags.

21 years agoRe-add type conversion examples.
Bruce Momjian [Thu, 20 Mar 2003 16:17:32 +0000 (16:17 +0000)]
Re-add type conversion examples.

21 years agoStarted adding date and timestamp.
Michael Meskes [Thu, 20 Mar 2003 15:56:50 +0000 (15:56 +0000)]
Started adding date and timestamp.

21 years agoRemove extern for cancelConn, from Karel Zak
Bruce Momjian [Thu, 20 Mar 2003 15:44:17 +0000 (15:44 +0000)]
Remove extern for cancelConn, from Karel Zak

21 years agoRemove compile warning.
Bruce Momjian [Thu, 20 Mar 2003 15:39:53 +0000 (15:39 +0000)]
Remove compile warning.

21 years agoAt present, dates are put into a dump in the format specified by the
Bruce Momjian [Thu, 20 Mar 2003 07:05:21 +0000 (07:05 +0000)]
At present, dates are put into a dump in the format specified by the
default datestyle.  This is not portable between installations.

This patch sets DATESTYLE to ISO at the start of a pg_dump, so that the
dates written into the dump will be restorable onto any database,
regardless of how its default datestyle is set.

Oliver Elphick

21 years agoTodo items:
Bruce Momjian [Thu, 20 Mar 2003 07:02:11 +0000 (07:02 +0000)]
Todo items:
Add ALTER SEQUENCE to modify min/max/increment/cache/cycle values

Also updated create sequence docs to mention NO MINVALUE, & NO MAXVALUE.

New Files:
doc/src/sgml/ref/alter_sequence.sgml
src/test/regress/expected/sequence.out
src/test/regress/sql/sequence.sql

ALTER SEQUENCE is NOT transactional.  It behaves similarly to setval().
It matches the proposed SQL200N spec, as well as Oracle in most ways --
Oracle lacks RESTART WITH for some strange reason.

--
Rod Taylor <rbt@rbt.ca>

21 years agoDone:
Bruce Momjian [Thu, 20 Mar 2003 07:00:57 +0000 (07:00 +0000)]
Done:

>  o -Add ALTER SEQUENCE to modify min/max/increment/cache/cycle values

21 years agoAttached is an update to contrib/tablefunc. It implements a new hashed
Bruce Momjian [Thu, 20 Mar 2003 06:46:30 +0000 (06:46 +0000)]
Attached is an update to contrib/tablefunc. It implements a new hashed
version of crosstab. This fixes a major deficiency in real-world use of
the original version. Easiest to undestand with an illustration:

Data:
-------------------------------------------------------------------
select * from cth;
  id | rowid |        rowdt        |   attribute    |      val
----+-------+---------------------+----------------+---------------
   1 | test1 | 2003-03-01 00:00:00 | temperature    | 42
   2 | test1 | 2003-03-01 00:00:00 | test_result    | PASS
   3 | test1 | 2003-03-01 00:00:00 | volts          | 2.6987
   4 | test2 | 2003-03-02 00:00:00 | temperature    | 53
   5 | test2 | 2003-03-02 00:00:00 | test_result    | FAIL
   6 | test2 | 2003-03-02 00:00:00 | test_startdate | 01 March 2003
   7 | test2 | 2003-03-02 00:00:00 | volts          | 3.1234
(7 rows)

Original crosstab:
-------------------------------------------------------------------
SELECT * FROM crosstab(
   'SELECT rowid, attribute, val FROM cth ORDER BY 1,2',4)
AS c(rowid text, temperature text, test_result text, test_startdate
text, volts text);
  rowid | temperature | test_result | test_startdate | volts
-------+-------------+-------------+----------------+--------
  test1 | 42          | PASS        | 2.6987         |
  test2 | 53          | FAIL        | 01 March 2003  | 3.1234
(2 rows)

Hashed crosstab:
-------------------------------------------------------------------
SELECT * FROM crosstab(
   'SELECT rowid, attribute, val FROM cth ORDER BY 1',
   'SELECT DISTINCT attribute FROM cth ORDER BY 1')
AS c(rowid text, temperature int4, test_result text, test_startdate
timestamp, volts float8);
  rowid | temperature | test_result |   test_startdate    | volts
-------+-------------+-------------+---------------------+--------
  test1 |          42 | PASS        |                     | 2.6987
  test2 |          53 | FAIL        | 2003-03-01 00:00:00 | 3.1234
(2 rows)

Notice that the original crosstab slides data over to the left in the
result tuple when it encounters missing data. In order to work around
this you have to be make your source sql do all sorts of contortions
(cartesian join of distinct rowid with distinct attribute; left join
that back to the real source data). The new version avoids this by
building a hash table using a second distinct attribute query.

The new version also allows for "extra" columns (see the README) and
allows the result columns to be coerced into differing datatypes if they
are suitable (as shown above).

In testing a "real-world" data set (69 distinct rowid's, 27 distinct
categories/attributes, multiple missing data points) I saw about a
5-fold improvement in execution time (from about 2200 ms old, to 440 ms
new).

I left the original version intact because: 1) BC, 2) it is probably
slightly faster if you know that you have no missing attributes.

README and regression test adjustments included. If there are no
objections, please apply.

Joe Conway

21 years agoI'm continuing to work on cleaning up code in psql. As things appear
Bruce Momjian [Thu, 20 Mar 2003 06:43:35 +0000 (06:43 +0000)]
I'm continuing to work on cleaning up code in psql.  As things appear
now, my changes seem to work.  Some possible minor bugs got squished
on the way but I can't be sure without more feedback from people who
really put the code to the test.

The new patch mostly simplifies variable handling and reduces code
duplication.  Changes in the command parser eliminate some redundant
variables (boolean state + depth counter), replaces some
"else if" constructs with switches, and so on.  It is meant to be
applied together with my previous patch, although I hope they don't
conflict; I went back to the CVS version for this one.

One more thing I thought should perhaps be changed: an IGNOREEOF
value of n will ignore only n-1 EOFs.  I didn't want to touch this
for fear of breaking existing applications, but it does seem a tad
illogical.

Jeroen T. Vermeulen