Tom Lane [Sat, 30 Dec 2000 02:34:56 +0000 (02:34 +0000)]
Clean up spinlock assembly code slightly (just cosmetic improvements)
for Alpha gcc case. For Alpha non-gcc case, replace use of
__INTERLOCKED_TESTBITSS_QUAD builtin with __LOCK_LONG_RETRY and
__UNLOCK_LONG. The former does not execute an MB instruction and
therefore was guaranteed not to work on multiprocessor machines.
The LOCK_LONG builtins produce code that is the same in all essential
details as the gcc assembler code.
Tom Lane [Sat, 30 Dec 2000 01:20:55 +0000 (01:20 +0000)]
Paranoia about possible values of errno after a shmget/semget failure.
In theory we should always get EEXIST if there's a key collision, but
if the kernel code tests error conditions in a weird order, perhaps
EACCES or EIDRM could occur too.
Tom Lane [Fri, 29 Dec 2000 21:31:21 +0000 (21:31 +0000)]
Fix failure in CreateCheckPoint on some Alpha boxes --- it's not OK to
assume that TAS() will always succeed the first time, even if the lock
is known to be free. Also, make sure that code will eventually time out
and report a stuck spinlock, rather than looping forever. Small cleanups
in s_lock.h, too.
Tom Lane [Fri, 29 Dec 2000 20:39:09 +0000 (20:39 +0000)]
stamp-h needs to be made by config.status, not elsewhere, per recipe in
Autoconf manual. In particular, touching it before creating config.status
is guaranteed to lose.
Bruce Momjian [Thu, 28 Dec 2000 23:56:46 +0000 (23:56 +0000)]
Attached are patches for two fixes to reduce memory usage by the JDBC
drivers.
The first fix fixes the PreparedStatement object to not allocate
unnecessary objects when converting native types to Stings. The old
code used the following format:
(new Integer(x)).toString()
whereas this can more efficiently be occompilshed by:
Integer.toString(x);
avoiding the unnecessary object creation.
The second fix is to release some resources on the close() of a
ResultSet. Currently the close() method on ResultSet is a noop. The
purpose of the close() method is to release resources when the ResultSet
is no longer needed. The fix is to free the tuples cached by the
ResultSet when it is closed (by clearing out the Vector object that
stores the tuples). This is important for my application, as I have a
cache of Statement objects that I reuse. Since the Statement object
maintains a reference to the ResultSet and the ResultSet kept references
to the old tuples, my cache was holding on to a lot of memory.
Peter Eisentraut [Thu, 28 Dec 2000 17:34:52 +0000 (17:34 +0000)]
Qualify %.in rule to avoid triggering on configure.in, repair unportable
attempt to install more than one file per 'install' invocation, clean up
some other oddities.
Tom Lane [Wed, 27 Dec 2000 23:59:14 +0000 (23:59 +0000)]
Fix portability problems recently exposed by regression tests on Alphas.
1. Distinguish cases where a Datum representing a tuple datatype is an OID
from cases where it is a pointer to TupleTableSlot, and make sure we use
the right typlen in each case.
2. Make fetchatt() and related code support 8-byte by-value datatypes on
machines where Datum is 8 bytes. Centralize knowledge of the available
by-value datatype sizes in two macros in tupmacs.h, so that this will be
easier if we ever have to do it again.
Bruce Momjian [Wed, 27 Dec 2000 05:44:57 +0000 (05:44 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.309
diff -r1.309 TODO
3c3
< Last updated: Sat Dec 23 09:48:47 EST 2000
---
> Last updated: Wed Dec 27 00:44:44 EST 2000
60a61,70
> * Add replication of distributed databases [replication]
> o automatic fallover
> o load balancing
> o master/slave replication
> o multi-master replication
> o partition data across servers
> o sample implementation in contrib/rserv
> o queries across databases or servers (two-phase commit)
> * Point-in-time data recovery using backup and write-ahead log
> * Allow row re-use without vacuum(Vadim) 112c122
< * Make file in/out interface to TOAST columns
---
> * Make binary/file in/out interface for TOAST columns 122a133
> * Automatically create rules on views so they are updateable, per SQL92 173c184
< * allow psql \d to show temporary table schema
---
> * allow psql \d to show temporary table structure 195d205
< * Add replication of distributed databases [replication] 200a211,212
> * Incremental backups
> * Allow SQL92 schemas 205d216
< * Allow row re-use without vacuum(Vadim) 229a241
> * Allow Java server-side programming 278a291
> * Automatically keep clustering on a table 329d341
< \9b5~
Bruce Momjian [Sat, 23 Dec 2000 14:48:57 +0000 (14:48 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.308
diff -r1.308 TODO
3c3
< Last updated: Thu Dec 21 14:45:00 EST 2000
---
> Last updated: Sat Dec 23 09:48:47 EST 2000
72a73
> * Remove unused files during database vacuum or postmaster startup
Tom Lane [Sat, 23 Dec 2000 04:05:31 +0000 (04:05 +0000)]
Replace overly-cute coding with code that (a) has defined behavior
according to the ANSI C spec, (b) gets the boundary conditions right,
and (c) is about a third as long and three times more intelligible.
Tom Lane [Fri, 22 Dec 2000 23:12:07 +0000 (23:12 +0000)]
Small cleanup of temp-table handling. Disallow creation of a non-temp
table that inherits from a temp table. Make sure the right things happen
if one creates a temp table, creates another temp that inherits from it,
then renames the first one. (Previously, system would end up trying to
delete the temp tables in the wrong order.)
Peter Eisentraut [Fri, 22 Dec 2000 19:31:56 +0000 (19:31 +0000)]
Avoid using the terms 'installation', 'site', or 'instance' when referring
to the thing you get from running initdb. That's called a database cluster
(per SQL).
Tom Lane [Fri, 22 Dec 2000 18:35:09 +0000 (18:35 +0000)]
Add 'ONLY' to queries generated by RI triggers, so as to preserve pre-7.1
semantics of RI operations. Eventually we ought to look at making RI
work properly across inheritance trees, but not for 7.1 ...
Tom Lane [Fri, 22 Dec 2000 18:00:24 +0000 (18:00 +0000)]
Change default output formatting for CIDR to be unabbreviated, per
recommendation from Paul Vixie. Add a new abbrev() function to produce
abbreviated format as text. No forced initdb, but new function is not
available unless you do an initdb or add the pg_proc row manually.
If pghost == "" and pgport == "" then PQsetdbLogin() fails with a
error message:
Is the postmaster running locally
and accepting connections on Unix socket '/tmp/.s.PGSQL.0'?
I see many applications such as PHP fails due to this behavior.
Now if pgport == "", then it is assumed to be a DEF_PGPORT_STR. This
is the same behavior as the version prior 7.1.
Bruce Momjian [Fri, 22 Dec 2000 03:08:52 +0000 (03:08 +0000)]
In looking at the 7.1beta1 code for JDBC, I noticed that support was
added to support character set encodings. However I noticed that the
encoding that is used isn't obtained from the DB. Since Java uses
unicode UCS2 internally the character set encoding is used to translate
strings from/to the DB encoding. So it seems logical that the code
would get the encoding from the DB instead of the current method of
requiring the user pass it as a parameter.
Attached is a patch that gets the DB encoding from the DB in the same
manner as is done in libpq/fe-connect.c. The patch is created off of
the latest CVS sources (Connection.java version 1.10).
Tom Lane [Fri, 22 Dec 2000 00:51:54 +0000 (00:51 +0000)]
Revise lock manager to support "session level" locks as well as "transaction
level" locks. A session lock is not released at transaction commit (but it
is released on transaction abort, to ensure recovery after an elog(ERROR)).
In VACUUM, use a session lock to protect the master table while vacuuming a
TOAST table, so that the TOAST table can be done in an independent
transaction.
I also took this opportunity to do some cleanup and renaming in the lock
code. The previously noted bug in ProcLockWakeup, that it couldn't wake up
any waiters beyond the first non-wakeable waiter, is now fixed. Also found
a previously unknown bug of the same kind (failure to scan all members of
a lock queue in some cases) in DeadLockCheck. This might have led to failure
to detect a deadlock condition, resulting in indefinite waits, but it's
difficult to characterize the conditions required to trigger a failure.
Bruce Momjian [Thu, 21 Dec 2000 19:45:07 +0000 (19:45 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.307
diff -r1.307 TODO
3c3
< Last updated: Thu Dec 21 14:12:04 EST 2000
---
> Last updated: Thu Dec 21 14:45:00 EST 2000
32d31
< * Prevent BETWEEN from using duplicate nodes
Bruce Momjian [Thu, 21 Dec 2000 19:12:11 +0000 (19:12 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.306
diff -r1.306 TODO
3c3
< Last updated: Fri Dec 15 17:44:57 EST 2000
---
> Last updated: Thu Dec 21 14:12:04 EST 2000
31a32
> * Prevent BETWEEN from using duplicate nodes
Tom Lane [Thu, 21 Dec 2000 17:36:15 +0000 (17:36 +0000)]
Fix longstanding bug with VIEW using BETWEEN: OffsetVarNodes would get
applied to the duplicated subtree twice. Probably someday we should
fix the parser not to generate multiple links to the same subtree,
but for now a quick copyObject() is the path of least resistance.
Tom Lane [Wed, 20 Dec 2000 21:51:52 +0000 (21:51 +0000)]
Prevent freshly-started backend from ignoring SIGUSR1, per race condition
observed by Inoue. Also, don't call ProcRemove() from postmaster if we
have detected a backend crash --- too risky if shared memory is corrupted.
It's not needed anyway, considering we are going to reinitialize shared
memory and semaphores as soon as the last child is dead.
Bruce Momjian [Wed, 20 Dec 2000 03:19:24 +0000 (03:19 +0000)]
On Sunday 17 December 2000 15:07, Bruce Momjian wrote:
> We need additions to alter_table.sgml for the new OWNER option mention
> in the features list.
Bruce Momjian [Mon, 18 Dec 2000 18:45:05 +0000 (18:45 +0000)]
>> Here is a patch for the beos port (All regression tests are OK).
>> xlog.c : special case for beos to avoid 'link' which does not work yet
>> beos/sem.c : implementation of new sem_ctl call (GETPID) and a new
>sem_op
>> flag (IPCNOWAIT)
>> dynloader/beos.c : add a verification of symbol validity (seem that
the
>> loader sometime return OK with an invalid symbol)
>> postmaster.c : add beos forking support for the new checkpoint
process
>> postgres.c : remove beos special case for getrusage
>> beos.h : Correction of a bas definition of AF_UNIX, misc defnitions
>>
>>
>> thanks
>>
>>
>> cyril
Tom Lane [Mon, 18 Dec 2000 17:33:42 +0000 (17:33 +0000)]
Ensure that 'errno' is saved and restored by all signal handlers that
might change it. Experimentation shows that the signal handler call
mechanism does not save/restore errno for you, at least not on Linux
or HPUX, so this is definitely a real risk.
Tom Lane [Mon, 18 Dec 2000 06:50:51 +0000 (06:50 +0000)]
Make sure make_rels_by_clause_joins doesn't return multiple references
to same joinrel. Although make_rels_by_joins doesn't mind, GEQO has
an Assert that doesn't like this.
Tom Lane [Mon, 18 Dec 2000 00:44:50 +0000 (00:44 +0000)]
Clean up backend-exit-time cleanup behavior. Use on_shmem_exit callbacks
to ensure that we have released buffer refcounts and so forth, rather than
putting ad-hoc operations before (some of the calls to) proc_exit. Add
commentary to discourage future hackers from repeating that mistake.
Tom Lane [Sun, 17 Dec 2000 04:32:29 +0000 (04:32 +0000)]
Tweak select_common_type() to deal with possibility of multiple preferred
types in a category --- it was taking the last preferred type among the
inputs, rather than the first one as intended.
Peter Eisentraut [Sat, 16 Dec 2000 13:03:25 +0000 (13:03 +0000)]
Fix rules to re-generate config.h. The examples in the Autoconf manual are
flawed because the timestamp file is already updated when Makefile.global
is remade, and the rule for config.h never gets run.
Tom Lane [Sat, 16 Dec 2000 02:29:36 +0000 (02:29 +0000)]
Restructure performance tips into a single chapter ('populating a
database' was way too small to make a chapter). Add a section about
using JOIN syntax to direct the planner.
Bruce Momjian [Fri, 15 Dec 2000 22:45:09 +0000 (22:45 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.305
diff -r1.305 TODO
3c3
< Last updated: Fri Dec 15 17:40:22 EST 2000
---
> Last updated: Fri Dec 15 17:44:57 EST 2000 312a313
> * Mark is Mark Hollomon <mhh@mindspring.com> 326a328
> \9b5~
Bruce Momjian [Fri, 15 Dec 2000 22:40:34 +0000 (22:40 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.304
diff -r1.304 TODO
3c3
< Last updated: Fri Dec 15 17:03:28 EST 2000
---
> Last updated: Fri Dec 15 17:40:22 EST 2000 318a319
> * Ross is Ross J. Reedstrom <reedstrm@wallace.ece.rice.edu>
Bruce Momjian [Fri, 15 Dec 2000 22:03:40 +0000 (22:03 +0000)]
Update TODO list.
Index: TODO
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/doc/TODO,v
retrieving revision 1.303
diff -r1.303 TODO
3c3
< Last updated: Fri Dec 15 16:46:43 EST 2000
---
> Last updated: Fri Dec 15 17:03:28 EST 2000 319a320
> * Stephan is Stephan Szabo <sszabo@megazone23.bigpanda.com>