misleadingly-named WriteBuffer routine, and instead require routines that
change buffer pages to call MarkBufferDirty (which does exactly what it says).
We also require that they do so before calling XLogInsert; this takes care of
the synchronization requirement documented in SyncOneBuffer.  Note that
because bufmgr takes the buffer content lock (in shared mode) while writing
out any buffer, it doesn't matter whether MarkBufferDirty is executed before
the buffer content change is complete, so long as the content change is
completed before releasing exclusive lock on the buffer.  So it's OK to set
the dirtybit before we fill in the LSN.
This eliminates the former kluge of needing to set the dirtybit in LockBuffer.
Aside from making the code more transparent, we can also add some new
debugging assertions, in particular that the caller of MarkBufferDirty must
hold the buffer content lock, not merely a pin.

torn-page problems.  This introduces some issues of its own, mainly
that there are now some critical sections of unreasonably broad scope,
but it's a step forward anyway.  Further cleanup will require some
code refactoring that I'd prefer to get Oleg and Teodor involved in.

startup or recovery process.  Since such a process isn't a real backend,
pgstat.c gets confused.  This accounts for recent reports of strange
"invalid server process ID -1" log messages during crash recovery.
There isn't any point in attempting to make the report, since we'll discard
stats in such scenarios anyhow.

This commit doesn't make much functional change, but it does eliminate some
duplicated code --- for instance, PageIsNew tests are now done inside
XLogReadBuffer rather than by each caller.
The GIST xlog code still needs a lot of love, but I'll worry about that
separately.

have symlinks (ie, Windows).  Although it'll never be called on to do anything
useful during normal operation on such a platform, it's still needed to
re-create dropped directories during WAL replay.

failures even when the hardware and OS did nothing wrong.  Per recent analysis
of a problem report from Alex Bahdushka.

For the moment I've just diked out the test of the parameter, rather than
removing the GUC infrastructure and documentation, in case we conclude that
there's something salvageable there.  There seems no chance of it being
resurrected in the 8.1 branch though.

passed extend = true whenever we are reading a page we intend to reinitialize
completely, even if we think the page "should exist".  This is because it
might indeed not exist, if the relation got truncated sometime after the
current xlog record was made and before the crash we're trying to recover
from.  These two thinkos appear to explain both of the old bug reports
discussed here:
http://archives.postgresql.org/pgsql-hackers/2005-05/msg01369.php

tuples as needed "to keep VACUUM from complaining", but actually there is
a more compelling reason to do it: failure to do so violates MVCC semantics.
This is because a pre-existing serializable transaction might try to use
the index after we finish (re)building it, and it might fail to find tuples
it should be able to see.  We got this mostly right, but not in the case
of partial indexes: the code mistakenly discarded recently-dead tuples for
partial indexes.  Fix that, and adjust the comments.

when an error occurs during xlog replay.  Also, replace the former risky
'write into a fixed-size buffer with no overflow detection' API for XLOG
record description routines; use an expansible StringInfo instead.  (The
latter accounts for most of the patch bulk.)

Qingqing Zhou

command or expression, rather than one copy for each textual occurrence as
it did before.  This might result in some small performance improvement,
but the compelling reason to do it is that not doing so can result in
unexpected grouping failures because the main SQL parser won't see different
parameter numbers as equivalent.  Add a regression test for the failure case.
Per report from Robert Davidson.

for various mistakes involving INSERT and UPDATE target columns.

the logic it contained to switch to insertion sort for near-sorted input was
in fact a big loss, because it could fairly easily be fooled into applying
insertion sort to large subfiles that weren't all that well ordered.  Remove
that, and instead add a simple check for already-perfectly-sorted input, as
per suggestion from Dann Corbit.  This adds at worst O(N*lgN) overhead, and
usually far less, while sometimes allowing a subfile sort to finish in O(N)
time.  Preliminary testing says this is an improvement over the basic
Bentley & McIlroy code for many nonrandom inputs, and it costs almost
nothing when the input is random.

"void" rather than "int".

both in code and in the messages emitted to the user.

> 1) Fix the problems with the \s command.
> When the saveHistory is executed by the \s command we must not do the
> conversion \n -> \x01  (per
> http://archives.postgresql.org/pgsql-hackers/2006-03/msg00317.php )
>
> 2) Fix the handling of Ctrl+C
>
> Now when you do
> wsdb=# select 'your long query here '
> wsdb-#
> and press afterwards the CtrlC the line "select 'your long query here
'"
> will be in the history
>
> (partly per
> http://archives.postgresql.org/pgsql-hackers/2006-03/msg00297.php )
>
> 3) Fix the handling of commands with not closed brackets, quotes,
double
> quotes. (now those commands are not splitted in parts...)
>
> 4) Fix the behaviour when SINGLELINE mode is used. (before it was
almost
> broken ;(

Sergey E. Koposov

non-NULL: palloc() ereports on OOM, so we can safely assume it returns a
valid pointer.

test, per Dave Page and buildfarm.  Perhaps we will need a join_2 instead,
but for the moment assume that this test tracks the other diffs.

byte-swapping on the port number which causes the call to fail on Intel
Macs.

This patch uses htons() instead of htonl() and fixes this bug.

Ashley Clark

2005-05-13.  When we find that a new inner tuple can't possibly match any
outer tuple (because it contains a NULL), we can't immediately skip the
tuple when we are in NEXTINNER state.  Doing so can lead to emitting
multiple copies of the tuple in FillInner mode, because we may rescan the
tuple after returning to a previous marked tuple.  Instead, proceed to
NEXTOUTER state the same as we used to do.  After we've found that there's
no need to return to the marked position, we can go to SKIPINNER_ADVANCE
state instead of SKIP_TEST when the inner tuple is unmatchable; this
preserves the performance improvement.  Per bug report from Bruce.
I also made a couple of cosmetic code rearrangements and added a regression
test for the problem.

worked and someone asked about them, so we might as well document them.

make the LDAP code's error messages look like they were written by someone
who had heard of our style guidelines.

The original coding stored the raw parser output (ColumnDef and TypeName
nodes) which was ugly, bulky, and wrong because it failed to create any
dependency on the referenced datatype --- and in fact would not track type
renamings and suchlike.  Instead store a list of column type OIDs in the
RTE.

Also fix up general failure of recordDependencyOnExpr to do anything sane
about recording dependencies on datatypes.  While there are many cases where
there will be an indirect dependency (eg if an operator returns a datatype,
the dependency on the operator is enough), we do have to record the datatype
as a separate dependency in examples like CoerceToDomain.

initdb forced because of change of stored rules.

just doesn't work with Bison 2.0 ... fix it ...

during parse analysis, not only errors detected in the flex/bison stages.
This is per my earlier proposal.  This commit includes all the basic
infrastructure, but locations are only tracked and reported for errors
involving column references, function calls, and operators.  More could
be done later but this seems like a good set to start with.  I've also
moved the ReportSyntaxErrorPosition logic out of psql and into libpq,
which should make it available to more people --- even within psql this
is an improvement because warnings weren't handled by ReportSyntaxErrorPosition.

Most of the changes add the mandatory USING clause to DROP OPERATOR
CLASS statements.  DROP TYPE is now DROP TYPE CASCADE; without
CASCADE a DROP TYPE fails due to the circular dependency on the
type's I/O functions.  The DROP FUNCTION statements for the I/O
functions have been removed, as DROP TYPE CASCADE removes them
automatically. Patch from Michael Fuhr.

include file in the same area,.

derived from Jan's.

similar constants if they were not previously defined. All these
constants must be defined by limits.h according to C89, so we can
safely assume they are present.

case where we run low on array slots before we run low on memory is much
more probable than I had thought, and so it's important to treat each
tape fairly in that case.  To fix this, track per-tape slot allocations
just like we track per-tape space allocation.  Also, in the FINALMERGE
code path avoid scanning all the input tapes when we really only need to
read from one.  This should fix poor behavior with very large work_mem
as exhibited by Stefan Kaltenbrunner.
I didn't do anything about putting an upper bound on the number of tapes,
but maybe we should still consider that.

with not responding to query cancel during the last stage of btree index
creation.

var_samp(), stddev_pop(), and stddev_samp(). var_samp() and stddev_samp()
are just renamings of the historical Postgres aggregates variance() and
stddev() -- the latter names have been kept for backward compatibility.
This patch includes updates for the documentation and regression tests.
The catversion has been bumped.

NB: SQL2003 requires that DISTINCT not be specified for any of these
aggregates. Per discussion on -patches, I have NOT implemented this
restriction: if the user asks for stddev(DISTINCT x), presumably they
know what they are doing.