- Add more tests for CREATE TABLE AS with WITH NO DATA. The relation creation is
done at executor startup, however the main regression test suite is lacking
scenarios where no data is inserted which is something that can happen when
using EXECUTE or EXPLAIN with CREATE TABLE AS and WITH NO DATA. Some patches
are worked on to reshape the way CTAS relations are created, so this makes
sure that we do not miss some query patterns already supported. Reported-by:
Andreas Karlsson Author: Michael Paquier Reviewed-by: Andreas Karlsson
Discussion: https://postgr.es/m/20190206091817.GB14980@... https://git.postgresql.org/pg/commitdiff/537898bd81bd8bd3650846e0abde4298ff1373da
- Avoid creation of the free space map for small heap relations, take 2.
Previously, all heaps had FSMs. For very small tables, this means that the FSM
took up more space than the heap did. This is wasteful, so now we refrain from
creating the FSM for heaps with 4 pages or fewer. If the last known target
block has insufficient space, we still try to insert into some other page
before giving up and extending the relation, since doing otherwise leads to
table bloat. Testing showed that trying every page penalized performance
slightly, so we compromise and try every other page. This way, we visit at
most two pages. Any pages with wasted free space become visible at next
relation extension, so we still control table bloat. As a bonus, directly
attempting one or two pages can even be faster than consulting the FSM would
have been. Once the FSM is created for a heap we don't remove it even if
somebody deletes all the rows from the corresponding relation. We don't think
it is a useful optimization as it is quite likely that relation will again
grow to the same size. Author: John Naylor, Amit Kapila Reviewed-by: Amit
Kapila Tested-by: Mithun C Y Discussion:
- Move port-specific parts of with_temp_install to port makefile. Rather than
define ld_library_path_ver with a big nested $(if), just put the overriding
values in the makefiles for the relevant ports. Also add a variable for port
makefiles to append their own stuff to with_temp_install, and use it to set
LD_LIBRARY_PATH_RPATH=1 on FreeBSD which is needed to make LD_LIBRARY_PATH
override DT_RPATH if DT_RUNPATH is not set (which seems to depend in
unpredictable ways on the choice of compiler, at least on my system).
Backpatch for the benefit of anyone doing regression tests on FreeBSD. (For
other platforms there should be no functional change.)
- Fix dumping of matviews with indirect dependencies on primary keys. Commit
62215de29 turns out to have been not quite on-the-mark. When we are forced to
postpone dumping of a materialized view into the dump's post-data section
(because it depends on a unique index that isn't created till that section),
we may also have to postpone dumping other matviews that depend on said
matview. The previous fix didn't reliably work for such cases: it'd break the
dependency loops properly, producing a workable object ordering, but it didn't
necessarily mark all the matviews as "postponed_def". This led to harmless
bleating about "archive items not in correct section order", as reported by
Tom Cassidy in bug #15602. Less harmlessly, selective-restore options such as
--section might misbehave due to the matview dump objects not being properly
labeled. The right way to fix it is to consider that each pre-data dependency
we break amounts to moving the no-longer-dependent object into post-data, and
hence we should mark that object if it's a matview. Back-patch to all
supported versions, since the issue's been there since matviews were
- Doc: in each release branch, keep only that branch's own release notes.
Historically we've had each release branch include all prior branches' notes,
including minor-release changes, back to the beginning of the project. That's
basically an O(N^2) proposition, and it was starting to catch up with us: as
of HEAD the back-branch release notes alone accounted for nearly 30% of the
documentation. While there's certainly some value in easy access to
back-branch notes, this is getting out of hand. Hence, switch over to the
rule that each branch contains only its own release notes. So as to not make
older notes too hard to find, each branch will provide URLs for the
immediately preceding branches' release notes on the project website. There
might be value in providing aggregated notes across all branches somewhere on
the website, but that's a task for another day. Discussion:
- Remove unnecessary "inline" marker introduced in commit 4be058fe9. Some of our
older buildfarm members bleat about this coding, along the lines of
prepjointree.c:112: warning: 'get_result_relid' declared inline after being
called prepjointree.c:112: warning: previous declaration of 'get_result_relid'
was here Modern compilers will probably inline this function without being
prompted, so rather than move the function, let's just drop the marking.
- Propagate lateral-reference information to indirect descendant relations.
create_lateral_join_info() computes a bunch of information about lateral
references between base relations, and then attempts to propagate those
markings to appendrel children of the original base relations. But the
original coding neglected the possibility of indirect descendants
(grandchildren etc). During v11 development we noticed that this was wrong
for partitioned-table cases, but failed to realize that it was just as wrong
for any appendrel. While the case can't arise for appendrels derived from
traditional table inheritance (because we make a flat appendrel for that),
nested appendrels can arise from nested UNION ALL subqueries. Failure to mark
the lower-level relations as having lateral references leads to confusion in
add_paths_to_append_rel about whether unparameterized paths can be built.
It's not very clear whether that leads to any user-visible misbehavior; the
lack of field reports suggests that it may cause nothing worse than minor cost
misestimation. Still, it's a bug, and it leads to failures of Asserts that I
intend to add later. To fix, we need to propagate information from all
appendrel parents, not just those that are RELOPT_BASERELs. We can still do
it in one pass, if we rely on the append_rel_list to be ordered with ancestor
relationships before descendant ones; add assertions checking that. While
fixing this, we can make a small performance improvement by traversing the
append_rel_list just once instead of separately for each appendrel parent
relation. Noted while investigating bug #15613, though this patch does not
fix that (which is why I'm not committing the related Asserts yet).
Discussion: https://postgr.es/m/3951.1549403812@... https://git.postgresql.org/pg/commitdiff/bdd9a99aac3bb0eaa49b5db81b2bd9402617fa32
- Split create_foreignscan_path() into three functions. Up to now postgres_fdw
has been using create_foreignscan_path() to generate not only base-relation
paths, but also paths for foreign joins and foreign upperrels. This is wrong,
because create_foreignscan_path() calls get_baserel_parampathinfo() which will
only do the right thing for baserels. It accidentally fails to fail for
unparameterized paths, which are the only ones postgres_fdw (thought it) was
handling, but we really need different APIs for the baserel and join cases.
In HEAD, the best thing to do seems to be to split up the baserel, joinrel,
and upperrel cases into three functions so that they can have different APIs.
I haven't actually given create_foreign_join_path a different API in this
commit: we should spend a bit of time thinking about just what we want to do
there, since perhaps FDWs would want to do something different from the
build-up-a-join-pairwise approach that get_joinrel_parampathinfo expects. In
the meantime, since postgres_fdw isn't prepared to generate parameterized
joins anyway, just give it a defense against trying to plan joins with lateral
refs. In addition (and this is what triggered this whole mess) fix bug #15613
from Srinivasan S A, by teaching file_fdw and postgres_fdw that plain baserel
foreign paths still have outer refs if the relation has lateral_relids. Add
some assertions in relnode.c to catch future occurrences of the same error ---
in particular, to catch other FDWs doing that, but also as backstop against
core-code mistakes like the one fixed by commit bdd9a99aa. Bug #15613 also
needs to be fixed in the back branches, but the appropriate fix will look
quite a bit different there, since we don't want to assume that existing FDWs
get the word right away. Discussion:
- Call set_rel_pathlist_hook before generate_gather_paths, not after. The
previous ordering of these steps satisfied the nominal requirement that
set_rel_pathlist_hook could editorialize on the whole set of Paths constructed
for a base relation. In practice, though, trying to change the set of partial
paths was impossible. Adding one didn't work because (a) it was too late to
be included in Gather paths made by the core code, and (b) calling
add_partial_path after generate_gather_paths is unsafe, because it might try
to delete a path it thinks is dominated, but that is already embedded in some
Gather path(s). Nor could the hook safely remove partial paths, for the same
reason that they might already be embedded in Gathers. Better to call
extensions first, let them add partial paths as desired, and then gather. In
v11 and up, we already doubled down on that ordering by postponing gathering
even further for single-relation queries; so even if the hook wished to
editorialize on Gather path construction, it could not. Report and patch by
KaiGai Kohei. Back-patch to 9.6 where Gather paths were added. Discussion:
- Refactor the representation of indexable clauses in IndexPaths. In place of
three separate but interrelated lists (indexclauses, indexquals, and
indexqualcols), an IndexPath now has one list "indexclauses" of IndexClause
nodes. This holds basically the same information as before, but in a more
useful format: in particular, there is now a clear connection between an
indexclause (an original restriction clause from WHERE or JOIN/ON) and the
indexquals (directly usable index conditions) derived from it. We also change
the ground rules a bit by mandating that clause commutation, if needed, be
done up-front so that what is stored in the indexquals list is always directly
usable as an index condition. This gets rid of repeated re-determination of
which side of the clause is the indexkey during costing and plan generation,
as well as repeated lookups of the commutator operator. To minimize the added
up-front cost, the typical case of commuting a plain OpExpr is handled by a
new special-purpose function commute_restrictinfo(). For RowCompareExprs,
generating the new clause properly commuted to begin with is not really any
more complex than before, it's just different --- and we can save doing that
work twice, as the pretty-klugy original implementation did. Tracking the
connection between original and derived clauses lets us also track explicitly
whether the derived clauses are an exact or lossy translation of the original.
This provides a cheap solution to getting rid of unnecessary rechecks of
boolean index clauses, which previously seemed like it'd be more expensive
than it was worth. Another pleasant (IMO) side-effect is that EXPLAIN now
always shows index clauses with the indexkey on the left; this seems less
confusing. This commit leaves expand_indexqual_conditions() and some related
functions in a slightly messy state. I didn't bother to change them any more
than minimally necessary to work with the new data structure, because all that
code is going to be refactored out of existence in a follow-on patch.
Discussion: https://postgr.es/m/22182.1549124950@... https://git.postgresql.org/pg/commitdiff/1a8d5afb0dfc5d0dcc6eda0656a34cb1f0cf0bdf
- Create the infrastructure for planner support functions. Rename/repurpose
pg_proc.protransform as "prosupport". The idea is still that it names an
internal function that provides knowledge to the planner about the behavior of
the function it's attached to; but redesign the API specification so that it's
not limited to doing just one thing, but can support an extensible set of
requests. The original purpose of simplifying a function call is handled by
the first request type to be invented, SupportRequestSimplify. Adjust all the
existing transform functions to handle this API, and rename them fron
"xxx_transform" to "xxx_support" to reflect the potential generalization of
what they do. (Since we never previously provided any way for extensions to
add transform functions, this change doesn't create an API break for them.)
Also add DDL and pg_dump support for attaching a support function to a
user-defined function. Unfortunately, DDL access has to be restricted to
superusers, at least for now; but seeing that support functions will pretty
much have to be written in C, that limitation is just theoretical. (This
support is untested in this patch, but a follow-on patch will add cases that
exercise it.) Discussion: https://postgr.es/m/15193.1548028093@... https://git.postgresql.org/pg/commitdiff/1fb57af92069ee104c09e2016af9e0e620681be3
- Build out the planner support function infrastructure. Add support function
requests for estimating the selectivity, cost, and number of result rows (if a
SRF) of the target function. The lack of a way to estimate selectivity of a
boolean-returning function in WHERE has been a recognized deficiency of the
planner since Berkeley days. This commit finally fixes it. In addition,
non-constant estimates of cost and number of output rows are now possible. We
still fall back to looking at procost and prorows if the support function
doesn't service the request, of course. To make concrete use of the
possibility of estimating output rowcount for SRFs, this commit adds support
functions for array_unnest(anyarray) and the integer variants of
generate_series; the lack of plausible rowcount estimates for those, even when
it's obvious to a human, has been a repeated subject of complaints.
Obviously, much more could now be done in this line, but I'm mostly just
trying to get the infrastructure in place. Discussion:
- Repair unsafe/unportable snprintf usage in pg_restore. warn_or_exit_horribly()
was blithely passing a potentially-NULL string pointer to a %s format
specifier. That works (at least to the extent of not crashing) on some
platforms, but not all, and since we switched to our own snprintf.c it doesn't
work for us anywhere. Of the three string fields being handled this way here,
I think that only "owner" is supposed to be nullable ... but considering that
this is error-reporting code, it has very little business assuming anything,
so put in defenses for all three. Per a crash observed on buildfarm member
crake and then reproduced here. Because of the portability aspect, back-patch
to all supported versions.
- Solve cross-version-upgrade testing problem induced by 1fb57af92. Renaming
varchar_transform to varchar_support had a side effect I hadn't foreseen: the
core regression tests leave around a transform object that relies on that
function, so the name change breaks cross-version upgrade tests, because the
name used in the older branches doesn't match. Since the dependency on
varchar_transform was chosen with the aid of a dartboard anyway (it would
surely not work as a language transform support function), fix by just
choosing a different random builtin function with the right signature. Also
add some comments explaining why this isn't horribly unsafe. I chose to make
the same substitution in a couple of other copied-and-pasted test cases, for
consistency, though those aren't directly contributing to the testing problem.
Per buildfarm. Back-patch, else it doesn't fix the problem.
- Fix heap_getattr() handling of fast defaults. Previously heap_getattr()
returned NULL for attributes with a fast default value (c.f. 16828d5c0273), as
it had no handling whatsoever for that case. A previous fix, 7636e5c60f,
attempted to fix issues caused by this oversight, but just expanding OLD
tuples for triggers doesn't actually solve the underlying issue. One known
consequence of this bug is that the check for HOT updates can return the wrong
result, when a previously fast-default'ed column is set to NULL. Which in turn
means that an index over a column with fast default'ed columns might be
corrupt if the underlying column(s) allow NULLs. Fix by handling fast default
columns in heap_getattr(), remove now superfluous expansion in
GetTupleForTrigger(). Author: Andres Freund Discussion:
https://postgr.es/m/20190201162404.onngi77f26baem4g@... Backpatch: 11, where fast defaults were introduced
- Plug leak in BuildTupleHashTable by creating ExprContext in correct context.
In bf6c614a2f2c5 I added a expr context to evaluate the grouping expression.
Unfortunately the code I added initialized them while in the calling context,
rather the table context. Additionally, I used CreateExprContext() rather
than CreateStandaloneExprContext(), which creates the econtext in the estate's
query context. Fix that by using CreateStandaloneExprContext when in the
table's tablecxt. As we rely on the memory being freed by a memory context
reset that means that the econtext's shutdown callbacks aren't being called,
but that seems ok as the expressions are tightly controlled due to
ExecBuildGroupingEqual(). Bug: #15592 Reported-By: Dmitry Marakasov Author:
Andres Freund Discussion:
https://postgr.es/m/20190114222838.h6r3fuyxjxkykf6t@... Backpatch: 11, where I broke this in bf6c614a2f2c5
- Allow to reset execGrouping.c style tuple hashtables. This has the advantage
that the comparator expression, the table's slot, etc do not have to be
rebuilt. Additionally the simplehash.h hashtable within the tuple hashtable
now keeps its previous size and doesn't need to be reallocated. That both
reduces allocator overhead, and improves performance in cases where the input
estimation was off by a significant factor. To avoid an API/ABI break, the
new parameter is exposed via the new BuildTupleHashTableExt(), and
BuildTupleHashTable() now is a wrapper around the former, that continues to
allocate the table itself in the tablecxt. Using this fixes performance
issues discovered in the two bugs referenced. This commit however has not
converted the callers, that's done in a separate commit. Bug: #15592 #15486
Reported-By: Jakub Janeček, Dmitry Marakasov Author: Andres Freund Discussion:
https://postgr.es/m/15486-05850f065da42931@... https://postgr.es/m/20190114180423.ywhdg2iagzvh43we@... Backpatch: 11, this is a prerequisite for other fixes
- Avoid amcheck inline compression false positives. The previous tacit
assumption that index_form_tuple() hides differences in the TOAST state of its
input datums was wrong. Normalize input varlena datums by decompressing
compressed values, and forming a new index tuple for fingerprinting using
uncompressed inputs. The final normalized representation may actually be
compressed once again within index_form_tuple(), though that shouldn't matter.
When the original tuple is found to have no datums that are compressed inline,
fingerprint the original tuple directly. Normalization avoids false positive
reports of corruption in certain cases. For example, the executor can apply
toasting with some inline compression to an entire heap tuple because its
input has a single external TOAST pointer. Varlena datums for other
attributes that are not particularly good candidates for inline compression
can be compressed in the heap tuple in passing, without the representation of
the same values in index tuples ever receiving concomitant inline compression.
Add a test case to recreate the issue in a simpler though less realistic way:
by exploiting differences in pg_attribute.attstorage between heap and index
relations. This bug was discovered by me during testing of an upcoming set of
nbtree enhancements. It was also independently reported by Andreas Kunert, as
bug #15597. His test case was rather more realistic than the one I ended up
using. Bug: #15597 Discussion:
https://postgr.es/m/CAH2-WznrVd9ie+TTJ45nDT+v2nUt6YJwQrT9SebCdQKtAvfPZw@... Discussion: https://postgr.es/m/15597-294e5d3e7f01c407@... Backpatch: 11-, where heapallindexed verification was introduced.
Masahiko Sawada sent in another revision of a patch to add a
--disable-index-cleanup option to VACUUM and vacuumdb.
Tom Lane sent in a patch to enable extensions to supply
Haribabu Kommi sent in a patch to recommend that log_file_mode be 0640 for
group-readable directories to allow reading of log files by the members of the
Kuroda Hayato sent in another revision of a patch to add DECLARE STATEMENT to
Amul Sul sent in another revision of a patch to improve the partition-wise
Edmund Horner sent in another revision of a patch to add a selectivity estimate
for CTID system variables, support backward scans over restricted ranges in heap
access methods, and support range quals in TID scans.
Lætitia Avrot sent in another revision of a patch to implement log10 and
Thomas Munro sent in another revision of a patch to add a synchronous replay
mode for avoiding stale reads on hot standbys.
Surafel Temesgen sent in another revision of a patch to add --rows-per-insert to
David Rowley sent in another revision of a patch to add basic support for using
the POPCNT and SSE4.2s LZCNT opcodes.
Surafel Temesgen sent in another revision of a patch to implement FETCH FIRST
... WITH TIES.
Andrew Gierth sent in another revision of a patch to replace strotod with
strotof as infrastructure for the Ryu patch.
Pavel Stěhule sent in another revision of a patch to implement pragmas in
Michael Banck sent in another revision of a patch to enable verifying checksums
Peter Eisentraut sent in a patch to Use EXECUTE FUNCTION syntax for triggers
David Rowley sent in two revisions of a patch to tighten up a few overly lax
regexes in pg_dump's tap tests.
Haribabu Kommi sent in another revision of a patch to document the new pluggable
Álvaro Herrera sent in three more revisions of a patch to propagate REPLICA
IDENTITY to partitions.
David Rowley sent in another revision of a patch to forgo generating
single-subpath Append and MergeAppend paths.
Antonin Houska sent in another revision of a patch to push down aggregates.
David Rowley sent in another revision of a patch to fix a performance issue in
foreign-key-aware join estimation.
Jerry Jelinek sent in another revision of a patch to allow disable of WAL
Álvaro Herrera sent in another revision of a patch to call out unsuitable
Andrey V. Lepikhov sent in another revision of a patch to reduce the amount of
WAL generated by CREATE INDEX for GiST, GIN and SP-GiST.
Peter Eisentraut sent in a patch to fix the optimization of foreign-key on
Kyotaro HORIGUCHI sent in a patch to fix a bug which manifests as an internal
error while setting reloption on system catalogs.
David Rowley and Amit Kapila traded patches to document the fact that UPDATEs
which can move a tuple from one partition to another "just work."
John Naylor sent in another revision of a patch to improve the FSM regression
David Rowley sent in another revision of a patch to fix inadequate executor
locking of indexes.
Pavan Deolasee sent in a patch to add a separate table level option to control
Marius Timmer sent in another revision of a patch to implement
Peter Eisentraut sent in another revision of a patch to implement collations
with nondeterministic comparison.
Robert Haas sent in a patch to fix a failure of dsa_allocate().
Michaël Paquier sent in a patch to ensure that CTAS works in the WITH NO DATA
Andreas Karlsson sent in two revisions of a patch to add CINE support to CREATE
TABLE AS EXECUTE.
Arseny Sher sent in a patch to remove the assertion in reorderbuffer that cmax
Amit Khandekar sent in two revisions of a patch to pre-fetch the buffers keeping
a constant distance ahead of the buffer reads.
Amit Langote, Imai Yoshikazu, and Takayuki Tsunakawa traded patches to speed up
planning with partitions.
Kyotaro HORIGUCHI and Tomáš Vondra traded patches to remove catcache entries
that haven't been used for a certain time, track syscache usage, and add a
Peter Eisentraut sent in a patch to make more use of unconstify() by replacing
casts whose only purpose is to cast away const.
Michaël Paquier sent in another revision of a patch to add a max_wal_senders
Peter Eisentraut sent in another revision of a patch to implement REINDEX
John Naylor sent in two revisions of a patch to use Getopt::Long in catalog
Antonin Houska sent in another revision of a patch to fix some problems with
plan estimates in the PostgreSQL FDW.
Tomáš Vondra sent in three revisions of a patch to fix a performance issue in
Kyotaro HORIGUCHI sent in two revisions of a patch to explicitly mark some
attributes in the catalog as not needing a toast relation.
Noah Misch sent in another revision of a patch to synchronize with the upstream
Peter Eisentraut sent in a patch to set the fallback_application_name for a
walreceiver to cluster_name.
Pavel Stěhule sent in another revision of a patch to implement \dP (partitions)
Ryo Matsumura sent in another revision of a patch to add a 'bytea' type to ECPG.
Antonin Houska sent in a patch to fix a situation where an incorrect visibility
test function was being assigned to snapshot.
Sergey Cherkashin sent in another revision of a patch to add a client connection
check during the execution of the query.
Ashutosh Sharma sent in two revisions of a patch to make it possible to create a
view on a table without columns.
Brandur Leach sent in two revisions of a patch to create a SortSupport
implementation for inet/cidr.
Aleksey Kondratov sent in another revision of a patch to pg_rewind which makes
it possible to use restore_command from postgresql.conf or the command line.
Tom Lane sent in two revisions of a patch to remove findDependentObjects()'s
dependency on scan order.
Álvaro Herrera sent in two revisions of a patch to fix trigger dropping.
Haribabu Kommi sent in a patch to avoid counting parallel worker transactions
Hironobu SUZUKI sent in another revision of a patch to pgbench to add a
pseudo-random permutation function.