Since we taught btree to handle ScalarArrayOpExpr quals natively (commit
9e8da0f7), the planner has always included
ScalarArrayOpExpr quals in index conditions if possible.  However, if the
qual is for a non-first index column, this could result in an inferior plan
because we can no longer take advantage of index ordering (cf. commit
807a40c5).  It can be better to omit the
ScalarArrayOpExpr qual from the index condition and let it be done as a
filter, so that the output doesn't need to get sorted.  Indeed, this is
true for the query introduced as a test case by the latter commit.

To fix, restructure get_index_paths and build_index_paths so that we
consider paths both with and without ScalarArrayOpExpr quals in non-first
index columns.  Redesign the API of build_index_paths so that it reports
what it found, saving useless second or third calls.

Report and patch by Andrew Gierth (though rather heavily modified by me).
Back-patch to 9.2 where this code was introduced, since the issue can
result in significant performance regressions compared to plans produced
by 9.1 and earlier.

Perl 5.12 ships with a somewhat broken version of Test::Simple, so skip
the tests if that is found.

The relevant fix is

    0.98  Wed, 23 Feb 2011 14:38:02 +1100
        Bug Fixes
        * subtest() should not fail if $? is non-zero. (Aaron Crane)

The prove program included in Perl 5.8 does not support the --ext
option, so don't use that and use wildcards on the command line instead.

Note that the tests will still all be skipped, because, for instance,
the version of Test::More is too old, but at least the regular
mechanisms for handling that will apply, instead of failing to call
prove altogether.

Discussion: 20140925133459.GB9633@alap3.anarazel.de
Author: Oskari Saarenmaa

Windows has one a locale whose name contains a non-ASCII character:
"Norwegian (Bokmål)" (that's an 'a' with a ring on top). That causes
trouble; when passing it setlocale(), it's not clear what encoding the
argument should be in. Another problem is that the locale name is stored in
pg_database catalog table, and the encoding used there depends on what
server encoding happens to be in use when the database is created. For
example, if you issue the CREATE DATABASE when connected to a UTF-8
database, the locale name is stored in pg_database in UTF-8. As long as all
locale names are pure ASCII, that's not a problem.

To work around that, map the troublesome locale name to a pure-ASCII alias
of the same locale, "norwegian-bokmal".

Now, this doesn't change the existing values that are already in
pg_database and in postgresql.conf. Old clusters will need to be fixed
manually. Instructions for that need to be put in the release notes.

This fixes bug #11431 reported by Alon Siman-Tov. Backpatch to 9.2;
backpatching further would require more work than seems worth it.

Needed at least on Windows.

I updated the docs and usage blurp, but forgot to commit the code changes
required.

Spotted by Michael Paquier.

Spotted by Tom Lane.

Apparently, this is a very common mistake for users to make; it is
better to have it fail reasonably rather than throw potentially a large
number of errors.  Since we have a magic string at the start of the
file, we can detect the case easily and there's no other possible useful
behavior anyway.

Author: Craig Ringer

This file was documenting an older version of patch 0ac5ad51; update
it to match what was really committed

Author: Florian Pflug

We have a project policy that I/O functions must not be volatile, as per
commit aab353a6, but we weren't doing
anything to enforce that.  In most usage the marking of the function
doesn't matter as long as its behavior is sane --- but I/O casts can
expose the marking as user-visible behavior, as per today's complaint
from Joe Van Dyk about contrib/ltree.

This test as such will only protect us against future errors in built-in
data types.  To catch the same error in contrib or third-party types,
perhaps we should make CREATE TYPE complain?  But that's a separate
issue from enforcing the policy for built-in types.

Don't crash if an ispell dictionary definition contains flags but not
any compound affixes.  (This isn't a security issue since only superusers
can install affix files, but still it's a bad thing.)

Also, be more careful about detecting whether an affix-file FLAG command
is old-format (ispell) or new-format (myspell/hunspell).  And change the
error message about mixed old-format and new-format commands into something
intelligible.

Per bug #11770 from Emre Hasegeli.  Back-patch to all supported branches.

Testing reveals that the memory allocation we do at transaction start
has small but measurable overhead on simple transactions.  To cut
down on that overhead, defer some of that work to the point when
AFTER triggers are first used, thus avoiding it altogether if they
never are.

Patch by me.  Review by Andres Freund.

This commit simply removes the second argument of PSQLexec that was
set to the same value everywhere. Comments and code blocks related
to this parameter are removed.

Noticed by Heikki Linnakangas, reviewed by Michael Paquier

Previously, this was not exposed outside of miscinit.c.  It is needed
for the pending pg_background patch, and will also be needed for
parallelism.  Without it, there's no way for a background worker to
re-create the exact authentication environment that was present in the
process that started it, which could lead to security exposures.

Previously the archive recovery always created .ready file for
the last WAL file of the old timeline at the end of recovery even when
it's restored from the archive and has .done file. That is, there was
the case where the WAL file had both .ready and .done files.
This caused the already-archived WAL file to be archived again.

This commit prevents the archive recovery from creating .ready file
for the last WAL file if it has .done file, in order to prevent it from
being archived again.

This bug was added when cascading replication feature was introduced,
i.e., the commit 52861058.
So, back-patch to 9.2, where cascading replication was added.

Reviewed by Michael Paquier

In a couple of code paths, pg_class_aclcheck is called in succession
with multiple different modes set.  This patch combines those modes to
have a single call of this function and reduce a bit process overhead
for permission checking.

Author: Michael Paquier <michael@otacoo.com>
Reviewed-by: Fabrízio de Royes Mello <fabriziomello@gmail.com>

The EOF-detection logic in pqReadData was a bit confused about who should
set up the error message in case the kernel gives us read-ready-but-no-data
rather than ECONNRESET or some other explicit error condition.  Since the
whole point of this situation is that the lower-level functions don't know
there's anything wrong, pqReadData itself must set up the message.  But
keep the assumption that if an errno was reported, a message was set up at
lower levels.

Per bug #11712 from Marko Tiikkaja.  It's been like this for a very long
time, so back-patch to all supported branches.

Declare static variable as static and external as extern.

The _bt_tuplecompare() function mentioned in comment hasn't existed for a
long time.

Peter Geoghegan

When commit 846e91e0 switched the linker
driver from dlltool/dllwrap to gcc, it became possible for linking to
choose shared libgcc.  Backends having loaded a module dynamically
linked to libgcc can exit abnormally, which the postmaster treats like a
crash.  Resume use of static libgcc exclusively, like 9.3 and earlier.
Back-patch to 9.4.

This improves consistency with the MSVC build.  On buildfarm member
narwhal, since commit 846e91e0,
shfolder.dll:SHGetFolderPath() crashes when dblink calls it by way of
pqGetHomeDirectory().  Back-patch to 9.4, where that commit first
appeared.  How it caused this regression remains a mystery.  This is a
partial revert of commit 889f0381, which
adopted shfolder.dll for Windows NT 4.0 compatibility.  PostgreSQL 8.2
dropped support for that operating system.

The last three updates to the sequence regression test have all forgotten
to touch the alternate expected-output file.  Sigh.

Michael Paquier

Author: Herwin Weststrate <herwin@quarantainenet.nl>
Reviewed-by: Ali Akbar <the.apaan@gmail.com>

Building the documentation with XSLT does not check the DTD, like a
DSSSL build would.  One can often get away with having invalid XML, but
the stylesheets might then create incorrect output, as they are not
designed to handle that.  Therefore, check the validity of the XML
against the DTD, using xmllint, during the build.

Add xmllint detection to configure, and add some documentation.

xmllint comes with libxml2, which is already in use, but it might be in
a separate package, such as libxml2-utils on Debian.
Reviewed-by: Fabien COELHO <coelho@cri.ensmp.fr>

The old text was written in ancient times when RPM packages could be
shared more or less freely across a plethora of RPM-based Linux
distributions.  This isn't really the case anymore, so just make this
information more concrete for the Red Hat family.

based on patch from Dag-Erling Smørgrav <des@des.no>

The duplication originated in cdd46c76, where restartpoints were
introduced.

In LogCheckpointStart's case the duplication actually lead to the
compiler's format string checking not to be effective because the
format string wasn't constant.

Arguably these messages shouldn't be elog(), but ereport() style
messages. That'd even allow to translate the messages... But as
there's more mistakes of that kind in surrounding code, it seems
better to change that separately.

Commit 7dbb6069 added a new CHECKPOINT_FLUSH_ALL flag. As that
commit needed to be backpatched I didn't change the numeric values of
the existing flags as that could lead to nastly problems if any
external code issued checkpoints. That's not a concern on master, so
renumber them there.

Also add a comment about CHECKPOINT_FLUSH_ALL above
CreateCheckPoint().

CREATE DATABASE and ALTER DATABASE .. SET TABLESPACE copy the source
database directory on the filesystem level. To ensure the on disk
state is consistent they block out users of the affected database and
force a checkpoint to flush out all data to disk. Unfortunately, up to
now, that checkpoint didn't flush out dirty buffers from unlogged
relations.

That bug means there could be leftover dirty buffers in either the
template database, or the database in its old location. Leading to
problems when accessing relations in an inconsistent state; and to
possible problems during shutdown in the SET TABLESPACE case because
buffers belonging files that don't exist anymore are flushed.

This was reported in bug #10675 by Maxim Boguk.

Fix by Pavan Deolasee, modified somewhat by me. Reviewed by MauMau and
Fujii Masao.

Backpatch to 9.1 where unlogged tables were introduced.

Apparently, computers are now a bit faster than when this was first
added, so we need to make room for a digit or two in the ops/sec format.

While we're at it, adjust some of the other output for a more consistent
line length.

json_agg and json_object_agg and their associated transition functions
should have been marked as stable rather than immutable, as they call IO
functions indirectly. Changing this probably isn't going to make much
difference, as you can't use an aggregate function in an index
expression, but we should be correct nevertheless.

json_object, on the other hand, should be marked immutable rather than
stable, as it does not call IO functions.

As discussed on -hackers, this change is being made without bumping the
catalog version, as we don't want to do that at this stage of the  cycle,
and  the changes are very unlikely to affect anyone.

If an inline-able SQL function taking a composite argument is used in a
LATERAL subselect, and the composite argument is a lateral reference,
the planner could fail with "variable not found in subplan target list",
as seen in bug #11703 from Karl Bartel.  (The outer function call used in
the bug report and in the committed regression test is not really necessary
to provoke the bug --- you can get it if you manually expand the outer
function into "LATERAL (SELECT inner_function(outer_relation))", too.)

The cause of this is that we generate the reltargetlist for the referenced
relation before doing eval_const_expressions() on the lateral sub-select's
expressions (cf find_lateral_references()), so what's scheduled to be
emitted by the referenced relation is a whole-row Var, not the simplified
single-column Var produced by optimizing the function's FieldSelect on the
whole-row Var.  Then setrefs.c fails to match up that lateral reference to
what's available from the outer scan.

Preserving the FieldSelect optimization in such cases would require either
major planner restructuring (to recursively do expression simplification
on sub-selects much earlier) or some amazingly ugly kluge to change the
reltargetlist of a possibly-already-planned relation.  It seems better
just to skip the optimization when the Var is from an upper query level;
the case is not so common that it's likely anyone will notice a few
wasted cycles.

AFAICT this problem only occurs for uplevel LATERAL references, so
back-patch to 9.3 where LATERAL was added.

Cosmetic oversight in commit 32984d8f.

Marko Tiikkaja

David Rowley

Etsuro Fujita

Revert the output of the individual backslash commands that change print
settings back to the 9.3 way (not showing the command name in
parentheses).  Implement \pset without arguments separately, showing all
settings with values in a table form.