The previous text was a bit misleading, as well as unnecessarily vague
about what information would be discarded.  Per gripe from Craig Skinner.

The old code entered a new hash table entry first, then scanned
pg_class to determine what value to fill in, and then populated the
entry.  This fails to work properly if a cache invalidation happens
as a result of opening pg_class.  Repair.

Along the way, get rid of the idea of blowing away the entire hash
table as a method of processing invalidations.  Instead, just delete
all the entries one by one.  This is probably not quite as cheap but
it's simpler, and shouldn't happen often.

Andres Freund

It's a trivial amount of RAM held until the end of the regression
test run; but it's probably worth fixing to silence future warnings
from code analyzers.

This was the only memory leak pointed out by clang's static code
analysis tool.

The root page is filled with as many items as fit, and the rest are inserted
using normal insertions. However, I fumbled the variable names, and the code
actually memcpy'd all the items on the page, overflowing the buffer. While
at it, rename the variable to make the distinction more clear.

Reported by Teodor Sigaev. This bug was introduced by my recent
refactorings, so no backpatching required.

This avoids an unused variable warning on Windows when building without
asserts

From: David Rowley <dgrowleyml@gmail.com>

Avoid the use of **, which was only introduced in Git version 1.8.2.

We can't search for the isolationtester binary until after we've set
up the environment, because otherwise when find_other_exec() tries
to invoke it with the -V option, it might fail for inability to
locate a working libpq.  So postpone that step.

Andres Freund

Reported by Thom Brown.

Per report from Colin 't Hart

This removes the remaining pieces of the IRIX port that was removed by
ea91a6be.

Simple oversight in commit 1cb108ef ---
recursively examining a subquery output column is only sane if the
original Var refers to a single output column.  Found by Kevin Grittner.

The pretty-printing logic in ruleutils.c operates by inserting a newline
and some indentation whitespace into strings that are already valid SQL.
This naturally results in leaving some trailing whitespace before the
newline in many cases; which can be annoying when processing the output
with other tools, as complained of by Joe Abbate.  We can fix that in
a pretty localized fashion by deleting any trailing whitespace before
we append a pretty-printing newline.  In addition, we have to modify the
code inserted by commit 2f582f76 so that
we also delete trailing whitespace when transposing items from temporary
buffers into the main result string, when a temporary item starts with a
newline.

This results in rather voluminous changes to the regression test results,
but it's easily verified that they are only removal of trailing whitespace.

Back-patch to 9.3, because the aforementioned commit resulted in many
more cases of trailing whitespace than had occurred in earlier branches.

Although the SQL spec forbids duplicate table aliases, historically
we've allowed queries like
    SELECT ... FROM tab1 x CROSS JOIN (tab2 x CROSS JOIN tab3 y) z
on the grounds that the aliased join (z) hides the aliases within it,
therefore there is no conflict between the two RTEs named "x".  The
LATERAL patch broke this, on the misguided basis that "x" could be
ambiguous if tab3 were a LATERAL subquery.  To avoid breaking existing
queries, it's better to allow this situation and complain only if
tab3 actually does contain an ambiguous reference.  We need only remove
the check that was throwing an error, because the column lookup code
is already prepared to handle ambiguous references.  Per bug #8444.

This is a similar fix as c6ec8793aa59d1842082e14b4b4aae7d4bd883fd
9.2. This should never happen in 9.3 and newer since the special case
cannot happen there, but this patch synchronizes up the code so there
is no confusion on why they're different. An empty block is as harmless
in 9.3 as it was in 9.2, and can safely be ignored.

Set per file type attributes in .gitattributes to fine-tune whitespace
checks.  With the associated cleanups, the tree is now clean for git

Commit 9b4d52f2 failed to notice
that pg_regress_ecpg needed updating.

This patch was independently submitted by both David Rowley
and Andres Freund.

If a page is deleted, and reused for something else, just as a search is
following a rightlink to it from its left sibling, the search would continue
scanning whatever the new contents of the page are. That could lead to
incorrect query results, or even something more curious if the page is
reused for a different kind of a page.

To fix, modify the search algorithm to lock the next page before releasing
the previous one, and refrain from deleting pages from the leftmost branch
of the tree.

Add a new Concurrency section to the README, explaining why this works.
There is a lot more one could say about concurrency in GIN, but that's for
another patch.

Backpatch to all supported versions.

Inspired by, but different from, a patch from Ivan Lezhnjov IV

This makes it possible to, for example, use the isolation tester to
test a contrib module.

Andres Freund

Pending patches for logical replication will use this to determine
which columns of a tuple ought to be considered as its candidate key.

Andres Freund, with minor, mostly cosmetic adjustments by me

This change prevents us from doing inappropriate subquery flattening in
cases such as dangerous functions hidden inside a sub-SELECT in the
targetlist of another sub-SELECT.  That could result in unexpected behavior
due to multiple evaluations of a volatile function, as in a recent
complaint from Etienne Dube.  It's been questionable from the very
beginning whether these functions should look into subqueries (as noted in
their comments), and this case seems to provide proof that they should.

Because the new code only descends into SubLinks, not SubPlans or
InitPlans, the change only affects the planner's behavior during
prepjointree processing and not later on --- for example, you can still get
it to use a volatile function in an indexqual if you wrap the function in
(SELECT ...).  That's a historical behavior, for sure, but it's reasonable
given that the executor's evaluation rules for subplans don't depend on
whether there are volatile functions inside them.  In any case, we need to
constrain the behavioral change as narrowly as we can to make this
reasonable to back-patch.

contain_volatile_functions() is best applied to the output of
expression_planner(), not its input, so that insertion of function
default arguments and constant-folding have been done.  (See comments
at CheckMutability, for instance.)  It's perhaps unlikely that anyone
will notice a difference in practice, but still we should do it properly.

In passing, change variable type from Node* to Expr* to reduce the net
number of casts needed.

Noted while perusing uses of contain_volatile_functions().

This avoids warnings from more-anal-than-average compilers, and might
prevent hidden syntax problems in the future.

Andres Freund

ExecBuildSlotValueDescription() printed "null" for each dropped column in
a row being complained of by ExecConstraints().  This has some sanity in
terms of the underlying implementation, but is of course pretty surprising
to users.  To fix, we must pass the target relation's descriptor to
ExecBuildSlotValueDescription(), because the slot descriptor it had been
using doesn't get labeled with attisdropped markers.

Per bug #8408 from Maxim Boguk.  Back-patch to 9.2 where the feature of
printing row values in NOT NULL and CHECK constraint violation messages
was introduced.

Michael Paquier and Tom Lane

Before jamming a desired targetlist into a plan node, one really ought to
make sure the plan node can handle projections, and insert a buffering
Result plan node if not.  planagg.c forgot to do this, which is a hangover
from the days when it only dealt with IndexScan plan types.  MergeAppend
doesn't project though, not to mention that it gets unhappy if you remove
its possibly-resjunk sort columns.  The code accidentally failed to fail
for cases in which the min/max argument was a simple Var, because the new
targetlist would be equivalent to the original "flat" tlist anyway.
For any more complex case, it's been broken since 9.1 where we introduced
the ability to optimize min/max using MergeAppend, as reported by Raphael
Bauduin.  Fix by duplicating the logic from grouping_planner that decides
whether we need a Result node.

In 9.2 and 9.1, this requires back-porting the tlist_same_exprs() function
introduced in commit 4387cf95, else we'd
uselessly add a Result node in cases that worked before.  It's rather
tempting to back-patch that whole commit so that we can avoid extra Result
nodes in mainline cases too; but I'll refrain, since that code hasn't
really seen all that much field testing yet.

Broken by my refactoring.

Per buildfarm results.  It looks like the older the Windows version, the
more errno codes it hasn't got ...

glibc, at least, is capable of returning "???" instead of anything useful
if it doesn't like the setting of LC_CTYPE.  If this happens, or in the
previously-known case of strerror() returning an empty string, try to
print the C macro name for the error code ("EACCES" etc).  Only if we
don't have the error code in our compiled-in list of popular error codes
(which covers most though not quite all of what's called out in the POSIX
spec) will we fall back to printing a numeric error code.  This should
simplify debugging.

Note that this functionality is currently only provided for %m in backend
ereport/elog messages.  That may be sufficient, since we don't fool with the
locale environment in frontend clients, but it's foreseeable that we might
want similar code in libpq for instance.

There was some talk of back-patching this, but let's see how the buildfarm
likes it first.  It seems likely that at least some of the POSIX-defined
error code symbols don't exist on all platforms.  I don't want to clutter
the entire list with #ifdefs, but we may need more than are here now.

MauMau, edited by me

These things didn't work because the planner omitted to do the necessary
preprocessing of a WindowFunc's argument list.  Add the few dozen lines
of code needed to handle that.

Although this sounds like a feature addition, it's really a bug fix because
the default-argument case was likely to crash previously, due to lack of
checking of the number of supplied arguments in the built-in window
functions.  It's not a security issue because there's no way for a
non-superuser to create a window function definition with defaults that
refers to a built-in C function, but nonetheless people might be annoyed
that it crashes rather than producing a useful error message.  So
back-patch as far as the patch applies easily, which turns out to be 9.2.
I'll put a band-aid in earlier versions as a separate patch.

(Note that these features still don't work for aggregates, and fixing that
case will be harder since we represent aggregate arg lists as target lists
not bare expression lists.  There's no crash risk though because CREATE
AGGREGATE doesn't accept defaults, and we reject named-argument notation
when parsing an aggregate call.)

Early close became apparent when invalidation messages were
processed in a new location under CLOBBER_CACHE_ALWAYS builds, due
to additional locking.

Back-patch to 9.3

Not sure how I missed these in previous commit.

Merge the isEnoughSpace and placeToPage functions in the b-tree interface
into one function that tries to put a tuple on page, and returns false if
it doesn't fit.

Move createPostingTree function to gindatapage.c, and change its contract
so that it can be passed more items than fit on the root page. It's in a
better position than the callers to know how many items fit.

Move ginMergeItemPointers out of gindatapage.c, into a separate file.

These changes make no difference now, but reduce the footprint of Alexander
Korotkov's upcoming patch to pack item pointers more tightly.

For rather inscrutable reasons, SQL:2008 disallows copying-and-modifying a
window definition that has any explicit framing clause.  The error message
we gave for this only made sense if the referencing window definition
itself contains an explicit framing clause, which it might well not.
Moreover, in the context of an OVER clause it's not exactly obvious that
"OVER (windowname)" implies copy-and-modify while "OVER windowname" does
not.  This has led to multiple complaints, eg bug #5199 from Iliya
Krapchatov.  Change to a hopefully more intelligible error message, and
in the case where we have just "OVER (windowname)", add a HINT suggesting
that omitting the parentheses will fix it.  Also improve the related
documentation.  Back-patch to all supported branches.

The previous commit was intended to make psql show the full path name when
doing a \s (history save), but it was very badly implemented and would show
confusing if not outright wrong information in many situations; for
instance if the path name given to \s is absolute, or if \cd commands
involving relative paths have been issued.  Consensus seems to be that
we don't especially need this functionality in \s, and certainly not in \s
alone.  So revert rather than trying to fix it up.  Per gripe from
Ian Barwick.

Although the bogus behavior exists in all supported versions, I'm not
back-patching, because the work created for translators (by change of
a translatable message) would probably outweigh the value of what is
after all a mostly-cosmetic change.

The relation should not be accessible to any other process, but it
should be locked for consistency.  Since this is not known to
cause any bug, it will not be back-patch, at least for now.

Per report from Andres Freund

Constant quals aren't handled the same way they used to be.  Also,
add mention of a couple more major steps in grouping_planner.
Per complaint a couple months back from Etsuro Fujita.

Per bug report from Tomonari Katsumata.

Back-patch to 9.3.