Add section about function table reformatting.

I can't see any way to make this table fit in PDF column width
without either a fundamental redesign or abbreviating EXCLUSIVE.
So I did the latter.

It'd be nicer if the abbreviating didn't leak into the HTML output
as well; but the hackery required to make the output different
seems like more trouble than it's really worth.

Discussion: https://postgr.es/m/6916.1589146280@sss.pgh.pa.us

Even after the tweaking I did in commit 5545b69a, some of the
longer keywords mentioned in the SQL standard don't fit the
available space in PDF output.

I experimented with various solutions like putting such keywords
on their own table lines, but everything looked ugly or confusing
or both; worse, the weirdness also appeared in the HTML version,
which (normally) doesn't need it.

The best answer seems to be to insert &zwsp; into long keywords
so that they can be broken into two lines when, and only when,
needed.  It doesn't look too awful if the break happens after
an underscore --- and fortunately, all the problematic keywords
have underscores.

Discussion: https://postgr.es/m/6916.1589146280@sss.pgh.pa.us

Use xreflabel attributes instead of endterm attributes to control the
appearance of links to subsections of SQL command reference pages.
This is simpler, it matches what we do elsewhere (e.g. for GUC variables),
and it doesn't draw "Unresolved ID reference" warnings from the PDF
toolchain.

Fix some places where the text was absolutely dependent on an <xref>
rendering exactly so, by using a <link> around the required text
instead.  At least one of those spots had already been turned into
bad grammar by subsequent changes, and the whole idea is just too
fragile for my taste.  <xref> does NOT have fixed output, don't write
as if it does.

Consistently include a page-level link in cross-man-page references,
because otherwise they are useless/nonsensical in man-page output.
Likewise, be consistent about mentioning "below" or "above" in same-page
references; we were doing that in about 90% of the cases, but now it's
100%.

Also get rid of another nonfunctional-in-PDF idea, of making
cross-references to functions by sticking ID tags on <row> constructs.
We can put the IDs on <indexterm>s instead --- which is probably not any
more sensible in abstract terms, but it works where the other doesn't.
(There is talk of attaching cross-reference IDs to most or all of
the docs' function descriptions, but for now I just fixed the two
that exist.)

Discussion: https://postgr.es/m/14480.1589154358@sss.pgh.pa.us

Restructure the function that locates the root of the to-be-deleted
subtree during nbtree page deletion.  Handle the conditions that make
page deletion unsafe in a slightly more uniform way, and acknowledge the
fact that the behavior with incomplete splits on internal pages is
different (as pointed out in the nbtree README as of commit 35bc0ec7).
Also invent new terminology that avoids ambiguity around which pages are
about to be deleted.  Consistently use the term "to-be-deleted subtree",
not the ambiguous term "branch".

We were calling the subtree parent page the "top parent page", but that
was quite misleading.  The top parent page usually refers to a page
unlinked from its siblings and marked deleted (during the second stage
of page deletion).  There was one kind of top parent page that we merely
removed a downlink from, and another kind of top parent page that we
actually marked deleted.  Eliminate the ambiguity by inventing a new
term ("subtree parent page") that refers to the former kind of page
only.

The one in xlogreader.h was pointed out by Antonin Houska; I (Álvaro) noticed the
others by grepping.

Author: Antonin Houska <ah@cybertec.at>
Discussion: https://postgr.es/m/28250.1589186654@antos

Source-Git-URL: https://git.postgresql.org/git/pgtranslation/messages.git
Source-Git-Hash: 80d8f54b3c5533ec036404bd3c3b24ff4825d037

This patch eliminates a few more "exceed the available area" warnings
whose causes aren't particularly connected to anything else.

The only one really worthy of comment is that I increased the space
allowed for an <orderedlist>'s numbers, because the default of 1em
doesn't quite work for more than one digit.  The rest are one-off
insertions of &zwsp; and suchlike tweaks, in places where they
shouldn't do any damage to the material.  (In particular, although
I split some long identifiers with zwsp's, there are other nearby
occurrences of each one; so those changes shouldn't hurt greppability
of the document sources.)

The last caller of this routine was removed in b4166911, and as a wise
man said one day, dead code tends to silently break.

Per discussion between Fujii Masao, Peter Geoghegan, Vignesh C and me.

Reported-by: Peter Geoghegan
Discussion: https://postgr.es/m/CAH2-Wz=sg5H8-vG4d5UmAofdcRMpeTDt2K-NUWp4GSfhenRGAQ@mail.gmail.com

I made up a very crude hack to compare the docs with reality (as
embodied in the system catalogs) ... and indeed they don't match
everywhere.  Missing oid columns, wrong data types, wrong "references"
links, columns listed in the wrong order.  None of this seems quite
important enough to back-patch.

Somehow we'd never noticed this oversight, even though it means
that such basic columns as pg_proc.proargtypes were not being
validated by the oidjoins test.  Correct the query and update
the test script with the newly-found dependencies.

Incremental sort always processes at least one full group group before
switching to prefix groups, so it's enough to check just the number of
full groups. There was no risk of division by zero due to the extra
condition, but it made the code harder to understand.

Reported-by: Ranier Vilela
Discussion: https://postgr.es/m/CAEudQAp+7qoS92-4V1vLChpdY3vEkLCbf+gye6P-4cirE-0z0A@mail.gmail.com

ExecReScanIncrementalSort was resetting bounded=false, which means the
optimization would be disabled on all rescans. This happens because
ExecSetTupleBound is called before the rescan, not after it.

Author: James Coleman
Reviewed-by: Tomas Vondra
Discussion: https://postgr.es/m/20200414065336.GI1492@paquier.xyz

The executor flags were not handled entirely correctly, although the
bugs were mostly harmless and it was mostly comment inaccuracy. We don't
need to strip any of the flags for child nodes.

Incremental sort does not support backward scans of mark/restore, so
MARK/BACKWARDS flags should not be possible. So we simply ensure this
using an assert, and we don't bother removing them when initializing
the child node.

With REWIND it's a bit less clear - incremental sort does not support
REWIND, but there is no way to signal this - it's legal to just ignore
the flag. We however continue passing the flag to child nodes, because
they might be useful to leverage that.

Reported-by: Michael Paquier
Author: James Coleman
Reviewed-by: Tomas Vondra
Discussion: https://postgr.es/m/20200414065336.GI1492@paquier.xyz

We seem to have forgotten to do this in the v12 cycle, so add it as
a task in the RELEASE_CHANGES list, in hopes we won't forget again.

While here, fix findoidjoins.c so that it actually works in the
new dispensation where OID is a regular column, and change it to only
consider system relations (this avoids being fooled by the OID column
in the brintest test table).

Also tweak the largeobject test so that the somewhat-recently-added
manual creation of a LO with an OID in the system range doesn't
fool findoidjoins.c.  For the moment I just made that use an unused
OID, but we might have to find a more robust solution someday.

Almost all error messages already include file name where relevant, but
this one had been overlooked.  Repair.

Backpatch to 9.5.

Author: Euler Taveira <euler.taveira@2ndquadrant.com>
Discussion: https://postgr.es/m/CAH503wA_VOrcKL_43p9atRejCDYmOZ8MzfK9S6TJrQqBqNeAXA@mail.gmail.comReviewed-by: Álvaro Herrera <alvherre@alvh.no-ip.org>

Code review for 0dc8ead4, prompted by a bug closed by 91c40548.

XLogReader's system for opening and closing segments had gotten too
complicated, with callbacks being passed at both the XLogReaderAllocate
level (read_page) as well as at the WALRead level (segment_open).  This
was confusing and hard to follow, so restructure things so that these
callbacks are passed together at XLogReaderAllocate time, and add
another callback to the set (segment_close) to make it a coherent whole.
Also, ensure XLogReaderState is an argument to all the callbacks, so
that they can grab at the ->private data if necessary.

Document the whole arrangement more clearly.

Author: Álvaro Herrera <alvherre@alvh.no-ip.org>
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Discussion: https://postgr.es/m/20200422175754.GA19858@alvherre.pgsql

*in* database, *in* cluster, *on* server; and some related fixes

There were a few different ways to line-wrap the error messages.  Make
them all the same, and use placeholders for the actual program names,
to save translation work.

Several combinations of generated columns and inheritance in CREATE
TABLE were not handled correctly.  Specifically:

- Disallow a child column specifying a generation expression if the
  parent column is a generated column.  The child column definition
  must be unadorned and the parent column's generation expression will
  be copied.

- Prohibit a child column of a generated parent column specifying
  default values or identity.

- Allow a child column of a not-generated parent column specifying
  itself as a generated column.  This previously did not work, but it
  was possible to arrive at the state via other means (involving ALTER
  TABLE), so it seems sensible to support it.

Add tests for each case.  Also add documentation about the rules
involving generated columns and inheritance.

Discussion:
    https://www.postgresql.org/message-id/flat/15830.1575468847%40sss.pgh.pa.us
    https://www.postgresql.org/message-id/flat/2678bad1-048f-519a-ef24-b12962f41807%40enterprisedb.com
    https://www.postgresql.org/message-id/flat/CAJvUf_u4h0DxkCMCeEKAWCuzGUTnDP-G5iVmSwxLQSXn0_FWNQ%40mail.gmail.com

When creating a new index, the attstorage setting of the table column
is copied to regular (non-expression) index columns.  But a later
ALTER TABLE ... SET STORAGE is not propagated to indexes, thus
creating an inconsistent and undumpable state.

Discussion: https://www.postgresql.org/message-id/flat/9765d72b-37c0-06f5-e349-2a580aafd989%402ndquadrant.com

Commit 6e654546 avoids locking bufmgr partitions to make pg_buffercache
less disruptive on production systems but forgot to update the docs.

Reported-by: Sawada Masahiko
Author: Sawada Masahiko
Reviewed-by: Amit Kapila
Backpatch-through: 10
Discussion: https://postgr.es/m/CA+fd4k6sD8oeP1qJbFAor=rCpYckU9DsywHiYx3x5Hz5Z8Ua_w@mail.gmail.com

TimelineHistoryRead and TimelineHistoryWrite wait events are reported
during waiting for a read and write of a timeline history file, respectively.
However, previously, TimelineHistoryRead wait event was not reported
while readTimeLineHistory() was reading a timeline history file. Also
TimelineHistoryWrite was not reported while writeTimeLineHistory() was
writing one line with the details of the timeline split, at the end.
This commit fixes these issues.

Back-patch to v10 where wait events for a timeline history file was added.

Author: Masahiro Ikeda
Reviewed-by: Michael Paquier, Fujii Masao
Discussion: https://postgr.es/m/d11b0c910b63684424e06772eb844ab5@oss.nttdata.com

Factor out code common to _bt_lock_branch_parent() and _bt_pagedel()
into a new utility function.  This new function is used to check that
the left sibling of a deletion target page does not have the
INCOMPLETE_SPLIT page flag set.  If it is set then deletion is unsafe;
there won't be a usable pivot tuple (with a downlink) in the parent page
that points to the deletion target page.  The page deletion algorithm is
not prepared to deal with that.  Also restructure an existing, related
utility function that checks if the right sibling of the target page has
the ISHALFDEAD page flag set.

This organization highlights the symmetry between the two cases.  The
goal is to make the design of page deletion clearer.  Both functions
involve a sibling page with a flag that indicates that there was an
interrupted operation (a page split or a page deletion) that resulted in
a page pointed to by sibling pages, but not pointed to in the parent.
And, both functions indicate if page deletion is unsafe due to the
absence of a particular downlink in the parent page.

checkcondition_str() failed to report multiple matches for a prefix
pattern correctly: it would dutifully merge the match positions, but
then after exiting that loop, if the last prefix-matching word had
had no suitable positions, it would report there were no matches.
The upshot would be failing to recognize a match that the query
should match.

It looks like you need all of these conditions to see the bug:
* a phrase search (else we don't ask for match position details)
* a prefix search item (else we don't get to this code)
* a weight restriction (else checkclass_str won't fail)

Noted while investigating a problem report from Pavel Borisov,
though this is distinct from the issue he was on about.

Back-patch to 9.6 where phrase search was added.

This converts the contrib documentation to the new style, and mops up
a couple of function tables that were outside chapter 9 in the main
docs.

A few contrib modules choose not to present their functions in the
standard tabular format.  There might be room to rethink those decisions
now that the standard format is more friendly to verbose descriptions.
But I have not undertaken to do that here; I just converted existing
tables.

Reported-by: Amit Langote

This merges three partition publication items into two.

Reported-by: Andrey M. Borodin

Discussion: https://postgr.es/m/D49B37B1-E2B9-4F67-8C6C-5CFD4015E8C5@yandex-team.ru

Reported-by: Fabien COELHO

Discussion: https://postgr.es/m/alpine.DEB.2.22.394.2005051811320.2183756@pseudo

In the wake of commit f2159931, we don't need to set table columns'
align specs retail.  Undo a few such settings I'd added in commit
5545b69a.  (The column width adjustments stay, though.)

We were acquiring object locks then deleting objects one by one, instead
of acquiring all object locks first, ignoring those that did not exist,
and then deleting all objects together.   The latter is the correct
protocol to use, and what this commits changes to code to do.  Failing
to follow that leads to "cache lookup failed for relation XYZ" error
reports when DROP OWNED runs concurrently with other DDL -- for example,
a session termination that removes some temp tables.

Author: Álvaro Herrera
Reported-by: Mithun Chicklore Yogendra (Mithun CY)
Reviewed-by: Ahsan Hadi, Tom Lane
Discussion: https://postgr.es/m/CADq3xVZTbzK4ZLKq+dn_vB4QafXXbmMgDP3trY-GuLnib2Ai1w@mail.gmail.com

I concluded that we really just ought to force all tables in PDF output
to default to "left" alignment (instead of "justify"); that is what the
HTML toolchain does and that's what most people have been designing the
tables to look good with.  There are few if any places where "justify"
produces better-looking output, and there are many where it looks
horrible.  So change stylesheet-fo.xsl to make that true.

Also tweak column widths in a few more tables to make them look better
and avoid "exceed the available area" warnings.  This commit fixes
basically everything that can be fixed through that approach.  The
remaining tables that give warnings either are scheduled for redesign
as per recent discussions, or need a fundamental rethink because they
Just Don't Work in a narrow view.

Attempting to use an installation path of Python that includes spaces
caused the MSVC builds to fail.  This fixes the issue by using the same
quoting method as ad7595b8 for OpenSSL.

Author: Victor Wagner
Discussion: https://postgr.es/m/20200430150608.6dc6b8c4@antares.wagner.home
Backpatch-through: 9.5

Oversight in commit bc3087b6.