We've had multiple issues with the clause_is_computable_at logic that
I introduced in 2489d76c4: it's been known to accept more than one
clone of the same qual at the same plan node, and also to accept no
clones at all. It's looking impractical to get it 100% right on the
basis of the currently-stored information, so fix it by introducing a
new RestrictInfo field "incompatible_relids" that explicitly shows
which outer joins a given clone mustn't be pushed above.
In principle we could populate this field in every RestrictInfo, but
that would cost space and there doesn't presently seem to be a need
for it in general. Also, while deconstruct_distribute_oj_quals can
easily fill the field with the remaining members of the commutative
join set that it's considering, computing it in the general case
seems again pretty complicated. So for now, just fill it for
clone quals.
Along the way, fix a bug that may or may not be only latent:
equivclass.c was generating replacement clauses with is_pushed_down
and has_clone/is_clone markings that didn't match their
required_relids. This led me to conclude that leaving the clone flags
out of make_restrictinfo's purview wasn't such a great idea after all,
so add them.
Per report from Richard Guo.
Discussion: https://postgr.es/m/CAMbWs48EYi_9-pSd0ORes1kTmTeAjT4Q3gu49hJtYCbSn2JyeA@mail.gmail.com
In commit 9df8f903e I (tgl) switched join_is_removable() from
using the min relid sets of the join under consideration to
using its full syntactic relid sets. This was a mistake,
as it allowed join removal in cases where a reference to the
join output would survive in some syntactically-lower join
condition. Revert to the former coding.
Richard Guo
Discussion: https://postgr.es/m/CAMbWs4-EU9uBGSP7G-iTwLBhRQ=rnZKvFDhD+n+xhajokyPCKg@mail.gmail.com
After applying outer-join identity 3 in the forward direction,
it was possible for the planner to mistakenly apply a qual clause
from above the two outer joins at the now-lower join level.
This can give the wrong answer, since a value that would get nulled
by the now-upper join might not yet be null.
To fix, when we perform such a transformation, consider that the
now-lower join hasn't really completed the outer join it's nominally
responsible for and thus its relid set should not include that OJ's
relid (nor should its output Vars have that nullingrel bit set).
Instead we add those bits when the now-upper join is performed.
The existing rules for qual placement then suffice to prevent
higher qual clauses from dropping below the now-upper join.
There are a few complications from needing to consider transitive
closures in case multiple pushdowns have happened, but all in all
it's not a very complex patch.
This is all new logic (from 2489d76c4) so no need to back-patch.
The added test cases all have the same results as in v15.
Tom Lane and Richard Guo
Discussion: https://postgr.es/m/0b819232-4b50-f245-1c7d-c8c61bf41827@postgrespro.ru
It seems like the code that these checks are backstopping may have
a few bugs left in it. Use a test-and-elog so that the tests are
performed even in non-assert builds, and so that we get something
more informative than "server closed the connection" on failure.
Committed separately with the idea that eventually we'll revert
this. It might be awhile though.
Discussion: https://postgr.es/m/3014965.1684293045@sss.pgh.pa.us
I've had a bee in my bonnet for some time about getting rid of
RestrictInfo.is_pushed_down, because it's squishily defined and
requires not-inexpensive extra tests to use (cf RINFO_IS_PUSHED_DOWN).
In commit 2489d76c4, I tried to make remove_rel_from_query() not
depend on that macro; but the replacement test is buggy,
as exposed by a report from Rushabh Lathia and Robert Haas.
That change was pretty incidental to the main goal of 2489d76c4,
so let's just revert it for now. (Getting rid of is_pushed_down
is still far away, anyway.)
Discussion: https://postgr.es/m/CA+TgmoYco=hmg+iX1CW9Y1_CzNoSL81J03wUG-d2_3=rue+L2A@mail.gmail.com
Our policy since commit ab77a5a45 has been that a plan node having
any initplans is automatically not parallel-safe. (This could be
relaxed, but not today.) clean_up_removed_plan_level neglected
this, and could attach initplans to a parallel-safe child plan
node without clearing the plan's parallel-safe flag. That could
lead to "subplan was not initialized" errors at runtime, in case
an initplan referenced another one and only the referencing one
got transmitted to parallel workers.
The fix in clean_up_removed_plan_level is trivial enough.
materialize_finished_plan also moves initplans from one node
to another, but it's okay because it already copies the source
node's parallel_safe flag. The other place that does this kind
of thing is standard_planner's hack to inject a top-level Gather
when debug_parallel_query is active. But that's actually dead
code given that we're correctly enforcing the "initplans aren't
parallel safe" rule, so just replace it with an Assert that
there are no initplans.
Also improve some related comments.
Normally we'd add a regression test case for this sort of bug.
The mistake itself is already reached by existing tests, but there
is accidentally no visible problem. The only known test case that
creates an actual failure seems too indirect and fragile to justify
keeping it as a regression test (not least because it fails to fail
in v11, though the bug is clearly present there too).
Per report from Justin Pryzby. Back-patch to all supported branches.
Discussion: https://postgr.es/m/ZDVt6MaNWkRDO1LQ@telsasoft.com
We were transferring partPruneInfos from PlannerInfo into PlannerGlobal
wrong, essentially relying on all of them being transferred, and
adjusting their list indexes based on that. But apparently it's
possible that some of them are skipped, so that strategy leads to a
corrupted execution tree. Instead, adjust each Append/MergeAppend's
partpruneinfo index as we copy from one list to the other, which seems
safer anyway. This requires adjusting the RT offset of the RTE
referenced in each partPruneInfo ahead of actually adjusting the RTE
itself, which seems a bit too ad-hoc.
This problem was introduced by commit ec38694894. However, it may be
that we no longer require the change introduced there, so perhaps we
should revert both the present commit and that one.
Problem noticed by sqlsmith.
Author: Amit Langote <amitlangote09@gmail.com
Discussion: https://postgr.es/m/CA+HiwqG6tbc2oadsbyyy24b2AL295XHQgyLRWghmA7u_SL1K8A@mail.gmail.com
create_append_path() would only apply get_baserel_parampathinfo
when the path is for a partitioned table, but it's also potentially
useful for paths for UNION ALL appendrels. Specifically, that
supports building a Memoize path atop this one.
While we're in the vicinity, delete some dead code in
create_merge_append_plan(): there's no need for it to support
parameterized MergeAppend paths, and it doesn't look like that
is going to change anytime soon. It'll be easy enough to undo
this when/if it becomes useful.
Richard Guo
Discussion: https://postgr.es/m/CAMbWs4_ABSu4PWG2rE1q10tJugEXHWgru3U8dAgkoFvgrb6aEA@mail.gmail.com
This function has been semi-deprecated ever since we invented
bms_next_member(). Its habit of scribbling on the input bitmapset
isn't great, plus for sufficiently large bitmapsets it would take
O(N^2) time to complete a loop. Now we have the additional problem
that reducing the input to empty while leaving it still accessible
would violate a planned invariant. So let's just get rid of it,
after updating the few extant callers to use bms_next_member().
Patch by me; thanks to Nathan Bossart and Richard Guo for review.
Discussion: https://postgr.es/m/1159933.1677621588@sss.pgh.pa.us
It's possible, in admittedly-rather-contrived cases, for an eclass
to generate a derived "join" qual that constrains the post-outer-join
value(s) of some RHS variable(s) without mentioning the LHS at all.
While the mechanisms were set up to work for this, we fell foul of
the "get_common_eclass_indexes" filter installed by commit 3373c7155:
it could decide that such an eclass wasn't relevant to the join, so
that the required qual clause wouldn't get emitted there or anywhere
else.
To fix, apply get_common_eclass_indexes only at inner joins, where
its rule is still valid. At an outer join, fall back to examining all
eclasses that mention either input (or the OJ relid, though it should
be impossible for an eclass to mention that without mentioning either
input). Perhaps we can improve on that later, but the cost/benefit of
adding more complexity to skip some irrelevant eclasses is dubious.
To allow cheaply distinguishing outer from inner joins, pass the
ojrelid to generate_join_implied_equalities as a separate argument.
This also allows cleaning up some sloppiness that had crept into
the definition of its join_relids argument, and it allows accurate
calculation of nominal_join_relids for a child outer join. (The
latter oversight seems not to have been a live bug, but it certainly
could have caused problems in future.)
Also fix what might be a live bug in check_index_predicates: it was
being sloppy about what it passed to generate_join_implied_equalities.
Per report from Richard Guo.
Discussion: https://postgr.es/m/CAMbWs4-DsTBfOvXuw64GdFss2=M5cwtEhY=0DCS7t2gT7P6hSA@mail.gmail.com
initsplan.c figured that it could push Var-free qual clauses to
the top of the current JoinDomain, which is okay in the abstract.
But if the current domain is inside some outer join, and we later
commute an inside-the-domain outer join with one outside it,
we end up placing the pushed-up qual clause incorrectly.
In distribute_qual_to_rels, avoid this by using the syntactic scope
of the qual clause; with the exception that if we're in the top-level
join domain we can still use the full query relid set, ensuring the
resulting gating Result node goes to the top of the plan. (This is
approximately as smart as the pre-v16 code was. Perhaps we can do
better later, but it's not clear that such cases are worth a lot of
sweat.)
In process_implied_equality, we don't have a clear notion of syntactic
scope, but we do have the results of SpecialJoinInfo construction.
Thumb through those and remove any lower outer joins that might get
commuted to above the join domain. Again, we can make an exception
for the top-level join domain. It'd be possible to work harder here
(for example, by keeping outer joins that aren't shown as potentially
commutable), but I'm going to stop here for the moment. This issue
has convinced me that the current representation of join domains
probably needs further refinement, so I'm disinclined to write
inessential dependent logic just yet.
In passing, tighten the qualscope passed to process_implied_equality
by generate_base_implied_equalities_no_const; there's no need for
it to be larger than the rel we are currently considering.
Tom Lane and Richard Guo, per report from Tender Wang.
Discussion: https://postgr.es/m/CAHewXNk9eJ35ru5xATWioTV4+xZPHptjy9etdcNPjUfY9RQ+uQ@mail.gmail.com
This was not something that required consideration before MERGE
was invented; but MERGE builds a join tree that left-joins to the
result relation, meaning that remove_useless_joins will consider
removing it. That should generally be stopped by the query's use
of output variables from the result relation. However, if the
result relation is inherited (e.g. a partitioned table) then
we don't add any row identity variables to the query until
expand_inherited_rtentry, which happens after join removal.
This was exposed as of commit 3c569049b, which made it possible
to deduce that a partitioned table could contain at most one row
matching a join key, enabling removal of the not-yet-expanded
result relation. Ooops.
To fix, let's just teach join_is_removable that the query result
rel is never removable. It's a cheap enough test in any case,
and it'll save some cycles that we'd otherwise expend in proving
that it's not removable, even in the cases we got right.
Back-patch to v15 where MERGE was added. Although I think the
case cannot be reached in v15, this seems like cheap insurance.
Per investigation of a report from Alexander Lakhin.
Discussion: https://postgr.es/m/36bee393-b351-16ac-93b2-d46d83637e45@gmail.com
For no clearly good reason, make_modifytable assumed that it
could not reach its get-the-FDW-info-the-hard-way path in MERGE.
It's currently possible to demonstrate that assertion failing,
which seems to be due to an upstream planner bug; but there's no
good reason to do it like this at all. Let's apply the principle
of separation of concerns and make the MERGE check separately,
after getting or not getting the fdwroutine pointer.
Per report from Alexander Lakhin. No test case, since I think
the potential test condition will go away soon.
Discussion: https://postgr.es/m/36bee393-b351-16ac-93b2-d46d83637e45@gmail.com
force_parallel_mode is meant to be used to allow us to exercise the
parallel query infrastructure to ensure that it's working as we expect.
It seems some users think this GUC is for forcing the query planner into
picking a parallel plan regardless of the costs. A quick look at the
documentation would have made them realize that they were wrong, but the
GUC is likely too conveniently named which, evidently, seems to often
result in users expecting that it forces the planner into usefully
parallelizing queries.
Here we rename the GUC to something which casual users are less likely to
mistakenly think is what they need to make their query run more quickly.
For now, the old name can still be used. We'll revisit if the old name
mapping can be removed once the buildfarm configs are all updated.
Reviewed-by: John Naylor
Discussion: https://postgr.es/m/CAApHDvrsOi92_uA7PEaHZMH-S4Xv+MGhQWA+GrP8b1kjpS1HjQ@mail.gmail.com
In commit b78f6264e I opined that it was "too risky" to delete a
relation's RelOptInfo from the planner's data structures when we have
realized that we don't need to join to it; so instead we just marked
it as a dead relation. In hindsight that judgment seems flawed: any
subsequent access to such a dead relation is arguably a bug in
itself, so leaving the RelOptInfo present just helps to mask bugs.
Let's delete it instead, allowing removal of the whole notion of a
"dead relation". So far as the regression tests can find, this
requires no other code changes, except for one Assert in equivclass.c
that was very dubiously not complaining about access to a dead rel.
Discussion: https://postgr.es/m/229905.1676062220@sss.pgh.pa.us
analyzejoins.c took care to clean out removed relids from the
clause_relids and required_relids of RestrictInfos associated with
the doomed rel ... but it paid no attention to the fact that if such a
RestrictInfo contains an OR clause, there will be sub-RestrictInfos
containing similar fields.
I'm more than a bit surprised that this oversight hasn't caused
visible problems before. In any case, it's certainly broken now,
so add logic to clean out the sub-RestrictInfos recursively.
We might need to back-patch this someday.
Per bug #17786 from Robins Tharakan.
Discussion: https://postgr.es/m/17786-f1ea7fbdab97daec@postgresql.org
One of the add_nulling_relids calls in deconstruct_distribute_oj_quals
added an OJ relid to too few Vars, while the other added it to too
many. We should consider the syntactic structure not
min_left/righthand while deciding which Vars to decorate, and when
considering pushing up a lower outer join pursuant to transforming the
second form of OJ identity 3 to the first form, we only want to
decorate Vars coming from its LHS.
In a related bug, I realized that make_outerjoininfo was failing to
check a very basic property that's needed to apply OJ identity 3:
the syntactically-upper outer join clause can't refer to the lower
join's LHS. This didn't break the join order restriction logic,
but it led to setting bogus commute_xxx bits, possibly resulting
in bogus nullingrel markings in modified quals.
Richard Guo and Tom Lane
Discussion: https://postgr.es/m/CAMbWs497CmBruMx1SOjepWEz+T5NWa4scqbdE9v7ZzSXqH_gQw@mail.gmail.com
Discussion: https://postgr.es/m/CAEP4nAx9C5gXNBfEA0JBfz7B+5f1Bawt-RWQWyhev-wdps8BZA@mail.gmail.com
The portion of join_is_removable() that checks PlaceHolderVars
can be made a little more accurate and intelligible than it was.
The key point is that we can allow join removal even if a PHV
mentions the target rel in ph_eval_at, if that mention was only
added as a consequence of forcing the PHV up to a join level
that's at/above the outer join we're trying to get rid of.
We can check that by testing for the OJ's relid appearing in
ph_eval_at, indicating that it's supposed to be evaluated after
the outer join, plus the existing test that the contained
expression doesn't actually mention the target rel.
While here, add an explicit check that there'll be something left
in ph_eval_at after we remove the target rel and OJ relid. There
is an Assert later on about that, and I'm not too sure that the
case could happen for a PHV satisfying the other constraints,
but let's just check. (There was previously a bms_is_subset test
that meant to cover this risk, but it's broken now because it
doesn't account for the fact that we'll also remove the OJ relid.)
The real reason for revisiting this code though is that the
Assert I left behind in 8538519db turns out to be easily
reachable, because if a PHV of this sort appears in an upper-level
qual clause then that clause's clause_relids will include the
PHV's ph_eval_at relids. This is a mirage though: we have or soon
will remove these relids from the PHV's ph_eval_at, and therefore
they no longer belong in qual clauses' clause_relids either.
Remove that Assert in join_is_removable, and replace the similar
one in remove_rel_from_query with code to remove the deleted relids
from clause_relids.
Per bug #17773 from Robins Tharakan.
Discussion: https://postgr.es/m/17773-a592e6cedbc7bac5@postgresql.org
make_outerjoininfo was set up to update SpecialJoinInfo's
commute_below, commute_above_l, commute_above_r fields as soon as
it found a pair of outer joins that look like they can commute.
However, this decision could be negated later in the same loop due
to finding an intermediate outer join that prevents commutation.
That left us with commute_xxx fields that were contradictory to the
join order restrictions expressed in min_lefthand/min_righthand.
The latter fields would keep us from actually choosing a bad join
order; but the inconsistent commute_xxx fields could bollix details
such as the varnullingrels values created for intermediate join
relation targetlists, ending in an assertion failure in setrefs.c.
To fix, wait till the end of make_outerjoininfo where we have
accurate values for min_lefthand/min_righthand, and then insert
only relids not present in those sets into the commute_xxx fields.
Per SQLSmith testing by Robins Tharakan. Note that while Robins
bisected the failure to commit b448f1c8d, it's really the fault of
2489d76c4. The outerjoin_delayed logic removed in the later commit
was keeping us from deciding that troublesome join pairs commute,
at least in the specific example seen here.
Discussion: https://postgr.es/m/CAEP4nAyAORgE8K_RHSmvWbE9UaChhjbEL1RrDU3neePwwRUB=A@mail.gmail.com
deconstruct_distribute tweaks the outer join scope (ojscope)
it passes to distribute_qual_to_rels when considering an outer
join qual that's above potentially-commutable outer joins.
However, if the current join is *not* potentially commutable,
we shouldn't do that. The argument that distribute_qual_to_rels
will not do something wrong with the bogus ojscope falls flat
if we don't pass it non-null postponed_oj_qual_list. Moreover,
there's no need to play games in this case since we aren't going
to commute anything.
Per SQLSmith testing by Robins Tharakan.
Discussion: https://postgr.es/m/CAEP4nAw74k4b-=93gmfCNX3MOY3y4uPxqbk_MnCVEpdsqHJVsg@mail.gmail.com
If we have a RestrictInfo that mentions both the removal-candidate
relation and the outer join's relid, then that is a pushed-down
condition not a join condition, so it should be grounds for deciding
that we can't remove the outer join. In commit 2489d76c4, I'd blindly
included the OJ's relid into "joinrelids" as per the new standard
convention, but the checks of attr_needed and ph_needed should only
allow the join's input rels to be mentioned.
Having done that, the check for references in pushed-down quals
a few lines further down should be redundant. I left it in place
as an Assert, though.
While researching this I happened across a couple of comments that
worried about the effects of update_placeholder_eval_levels.
That's gone as of b448f1c8d, so we can remove some worry.
Per bug #17769 from Robins Tharakan. The submitted test case
triggers this more or less accidentally because we flatten out
a LATERAL sub-select after we've done join strength reduction;
if we did that in the other order, this problem would be masked
because the outer join would get simplified to an inner join.
To ensure that the committed test case will continue to test
what it means to even if we make that happen someday, use a
test clause involving COALESCE(), which will prevent us from
using it to do join strength reduction.
Patch by me, but thanks to Richard Guo for initial investigation.
Discussion: https://postgr.es/m/17769-e4f7a5c9d84a80a7@postgresql.org
After pulling up LATERAL subqueries, we may have qual clauses that
refer to relations outside their syntactic scope. Before doing any
such pullup, prepjointree.c checks to make sure that it wouldn't
create a semantically-invalid situation; but we leave it to
deconstruct_jointree() to actually move these quals up the join
tree to a place where they can be evaluated. In commit 2489d76c4,
I (tgl) refactored deconstruct_jointree() in a way that caused
assertion failures while moving such quals, because the new logic
failed to distinguish "this jointree node is a parent of the source
one" from "this jointree node is processed after the source
one in depth-first order".
Fix this, and at the same time reduce the overhead a bit, by
getting rid of the common PostponedQual list and instead making each
JoinTreeItem contain a list of quals that needed to be postponed to
its level. We can help distribute_qual_to_rels find the appropriate
JoinTreeItem efficiently by adding parent-item links to the
JoinTreeItem data structure. This ends up being the same number
of relid subset checks as the original (pre-bug) logic, but less
list manipulation is required during multi-level postponements.
Richard Guo and Tom Lane, per bug #17768 from Robins Tharakan.
Discussion: https://postgr.es/m/17768-5ac8730ece54478f@postgresql.org
Here remove some dead code from heapgettup() and heapgettup_pagemode()
which was trying to support NoMovementScanDirection scans. This code can
never be reached as standard_ExecutorRun() never calls ExecutePlan with
NoMovementScanDirection.
Additionally, plans which were scanning an unordered index would use
NoMovementScanDirection rather than ForwardScanDirection. There was no
real need for this, so here we adjust this so we use ForwardScanDirection
for unordered index scans. A comment in pathnodes.h claimed that
NoMovementScanDirection was used for PathKey reasons, but if that was
true, it no longer is, per code in build_index_paths().
This does change the non-text format of the EXPLAIN output so that
unordered index scans now have a "Forward" scan direction rather than
"NoMovement". The text format of EXPLAIN has not changed.
Author: Melanie Plageman
Reviewed-by: Tom Lane, David Rowley
Discussion: https://postgr.es/m/CAAKRu_bvkhka0CZQun28KTqhuUh5ZqY=_T8QEqZqOL02rpi2bw@mail.gmail.com
In commit 2489d76c4, I'd thought it'd be safe to assert that a
PlaceHolderVar appearing in a scan-level expression has empty
nullingrels. However this is not so, as when we determine that a
join relation is certainly empty we'll put its targetlist into a
Result-with-constant-false-qual node, and nothing is done to adjust
the nullingrels of the Vars or PHVs therein. (Arguably, a Result
used in this way isn't really a scan-level node, but it certainly
isn't an upper node either ...)
It's not clear this is worth any close analysis, so let's just
take out the faulty Assert.
Per report from Robins Tharakan. I added a test case based on
his example, just in case somebody tries to tighten this up.
Discussion: https://postgr.es/m/CAEP4nAz7Enq3+DEthGG7j27DpuwSRZnW0Nh6jtNh75yErQ_nbA@mail.gmail.com
EquivalenceClasses are now understood as applying within a "join
domain", which is a set of inner-joined relations (possibly underneath
an outer join). We no longer need to treat an EC from below an outer
join as a second-class citizen.
I have hopes of eventually being able to treat outer-join clauses via
EquivalenceClasses, by means of only applying deductions within the
EC's join domain. There are still problems in the way of that, though,
so for now the reconsider_outer_join_clause logic is still here.
I haven't been able to get rid of RestrictInfo.is_pushed_down either,
but I wonder if that could be recast using JoinDomains.
I had to hack one test case in postgres_fdw.sql to make it still test
what it was meant to, because postgres_fdw is inconsistent about
how it deals with quals containing non-shippable expressions; see
https://postgr.es/m/1691374.1671659838@sss.pgh.pa.us. That should
be improved, but I don't think it's within the scope of this patch
series.
Patch by me; thanks to Richard Guo for review.
Discussion: https://postgr.es/m/830269.1656693747@sss.pgh.pa.us
Remove RestrictInfo.nullable_relids, along with a good deal of
infrastructure that calculated it. One use-case for it was in
join_clause_is_movable_to, but we can now replace that usage with
a check to see if the clause's relids include any outer join
that can null the target relation. The other use-case was in
join_clause_is_movable_into, but that test can just be dropped
entirely now that the clause's relids include outer joins.
Furthermore, join_clause_is_movable_into should now be
accurate enough that it will accept anything returned by
generate_join_implied_equalities, so we can restore the Assert
that was diked out in commit 95f4e59c3.
Remove the outerjoin_delayed mechanism. We needed this before to
prevent quals from getting evaluated below outer joins that should
null some of their vars. Now that we consider varnullingrels while
placing quals, that's taken care of automatically, so throw the
whole thing away.
Teach remove_useless_result_rtes to also remove useless FromExprs.
Having done that, the delay_upper_joins flag serves no purpose any
more and we can remove it, largely reverting 11086f2f2.
Use constant TRUE for "dummy" clauses when throwing back outer joins.
This improves on a hack I introduced in commit 6a6522529. If we
have a left-join clause l.x = r.y, and a WHERE clause l.x = constant,
we generate r.y = constant and then don't really have a need for the
join clause. But we must throw the join clause back anyway after
marking it redundant, so that the join search heuristics won't think
this is a clauseless join and avoid it. That was a kluge introduced
under time pressure, and after looking at it I thought of a better
way: let's just introduce constant-TRUE "join clauses" instead,
and get rid of them at the end. This improves the generated plans for
such cases by not having to test a redundant join clause. We can also
get rid of the ugly hack used to mark such clauses as redundant for
selectivity estimation.
Patch by me; thanks to Richard Guo for review.
Discussion: https://postgr.es/m/830269.1656693747@sss.pgh.pa.us
Traditionally we used the same Var struct to represent the value
of a table column everywhere in parse and plan trees. This choice
predates our support for SQL outer joins, and it's really a pretty
bad idea with outer joins, because the Var's value can depend on
where it is in the tree: it might go to NULL above an outer join.
So expression nodes that are equal() per equalfuncs.c might not
represent the same value, which is a huge correctness hazard for
the planner.
To improve this, decorate Var nodes with a bitmapset showing
which outer joins (identified by RTE indexes) may have nulled
them at the point in the parse tree where the Var appears.
This allows us to trust that equal() Vars represent the same value.
A certain amount of klugery is still needed to cope with cases
where we re-order two outer joins, but it's possible to make it
work without sacrificing that core principle. PlaceHolderVars
receive similar decoration for the same reason.
In the planner, we include these outer join bitmapsets into the relids
that an expression is considered to depend on, and in consequence also
add outer-join relids to the relids of join RelOptInfos. This allows
us to correctly perceive whether an expression can be calculated above
or below a particular outer join.
This change affects FDWs that want to plan foreign joins. They *must*
follow suit when labeling foreign joins in order to match with the
core planner, but for many purposes (if postgres_fdw is any guide)
they'd prefer to consider only base relations within the join.
To support both requirements, redefine ForeignScan.fs_relids as
base+OJ relids, and add a new field fs_base_relids that's set up by
the core planner.
Large though it is, this commit just does the minimum necessary to
install the new mechanisms and get check-world passing again.
Follow-up patches will perform some cleanup. (The README additions
and comments mention some stuff that will appear in the follow-up.)
Patch by me; thanks to Richard Guo for review.
Discussion: https://postgr.es/m/830269.1656693747@sss.pgh.pa.us
The rule system needs "old" and/or "new" pseudo-RTEs in rule actions
that are ON INSERT/UPDATE/DELETE. Historically it's put such entries
into the ON SELECT rules of views as well, but those are really quite
vestigial. The only thing we've used them for is to carry the
view's relid forward to AcquireExecutorLocks (so that we can
re-lock the view to verify it hasn't changed before re-using a plan)
and to carry its relid and permissions data forward to execution-time
permissions checks. What we can do instead of that is to retain
these fields of the RTE_RELATION RTE for the view even after we
convert it to an RTE_SUBQUERY RTE. This requires a tiny amount of
extra complication in the planner and AcquireExecutorLocks, but on
the other hand we can get rid of the logic that moves that data from
one place to another.
The principal immediate benefit of doing this, aside from a small
saving in the pg_rewrite data for views, is that these pseudo-RTEs
no longer trigger ruleutils.c's heuristic about qualifying variable
names when the rangetable's length is more than 1. That results
in quite a number of small simplifications in regression test outputs,
which are all to the good IMO.
Bump catversion because we need to dump a few more fields of
RTE_SUBQUERY RTEs. While those will always be zeroes anyway in
stored rules (because we'd never populate them until query rewrite)
they are useful for debugging, and it seems like we'd better make
sure to transmit such RTEs accurately in plans sent to parallel
workers. I don't think the executor actually examines these fields
after startup, but someday it might.
This is a second attempt at committing 1b4d280ea. The difference
from the first time is that now we can add some filtering rules to
AdjustUpgrade.pm to allow cross-version upgrade testing to pass
despite all the cosmetic changes in CREATE VIEW outputs.
Amit Langote (filtering rules by me)
Discussion: https://postgr.es/m/CA+HiwqEf7gPN4Hn+LoZ4tP2q_Qt7n3vw7-6fJKOf92tSEnX6Gg@mail.gmail.com
Discussion: https://postgr.es/m/891521.1673657296@sss.pgh.pa.us
Avoid explicitly grouping by columns that we know are redundant
for sorting, for example we need group by only one of x and y in
SELECT ... WHERE x = y GROUP BY x, y
This comes up more often than you might think, as shown by the
changes in the regression tests. It's nearly free to detect too,
since we are just piggybacking on the existing logic that detects
redundant pathkeys. (In some of the existing plans that change,
it's visible that a sort step preceding the grouping step already
didn't bother to sort by the redundant column, making the old plan
a bit silly-looking.)
To do this, build processed_groupClause and processed_distinctClause
lists that omit any provably-redundant sort items, and consult those
not the originals where relevant. This means that within the
planner, one should usually consult root->processed_groupClause or
root->processed_distinctClause if one wants to know which columns
are to be grouped on; but to check whether grouping or distinct-ing
is happening at all, check non-NIL-ness of parse->groupClause or
parse->distinctClause. This is comparable to longstanding rules
about handling the HAVING clause, so I don't think it'll be a huge
maintenance problem.
nodeAgg.c also needs minor mods, because it's now possible to generate
AGG_PLAIN and AGG_SORTED Agg nodes with zero grouping columns.
Patch by me; thanks to Richard Guo and David Rowley for review.
Discussion: https://postgr.es/m/185315.1672179489@sss.pgh.pa.us
In 1349d2790, we gave the planner the ability to provide ORDER BY/DISTINCT
Aggrefs with presorted input so that nodeAgg would not have to perform
sorts during execution. That commit failed to properly consider the
implications of if the Aggref had a volatile function in its ORDER
BY/DISTINCT clause. As it happened, this resulted in an ERROR about the
volatile function being missing from the targetlist.
Here, instead of adding the volatile function to the targetlist, we just
never consider an Aggref with a volatile function in its ORDER BY/DISTINCT
clause when choosing which Aggrefs we should sort by. We do this as if we
were to choose a plan which provided these aggregates with presorted
input, then if there were many such aggregates which could all share the
same sort order, then it may be surprising if they all shared the same
sort sometimes and didn't at other times when some other set of aggregates
were given presorted results. We can avoid this inconsistency by just
never providing these volatile function aggregates with presorted input.
Reported-by: Dean Rasheed
Discussion: https://postgr.es/m/CAEZATCWETioXs5kY8vT6BVguY41_wD962VDk=u_Nvd7S1UXzuQ@mail.gmail.com
This reverts commit 1b4d280ea1.
It's broken the buildfarm members that run cross-version-upgrade tests,
because they're not prepared to deal with cosmetic differences between
CREATE VIEW commands emitted by older servers and HEAD. Even if we had
a solution to that, which we don't, it'd take some time to roll it out
to the affected animals. This improvement isn't valuable enough to
justify addressing that problem on an emergency basis, so revert it
for now.
The rule system needs "old" and/or "new" pseudo-RTEs in rule actions
that are ON INSERT/UPDATE/DELETE. Historically it's put such entries
into the ON SELECT rules of views as well, but those are really quite
vestigial. The only thing we've used them for is to carry the
view's relid forward to AcquireExecutorLocks (so that we can
re-lock the view to verify it hasn't changed before re-using a plan)
and to carry its relid and permissions data forward to execution-time
permissions checks. What we can do instead of that is to retain
these fields of the RTE_RELATION RTE for the view even after we
convert it to an RTE_SUBQUERY RTE. This requires a tiny amount of
extra complication in the planner and AcquireExecutorLocks, but on
the other hand we can get rid of the logic that moves that data from
one place to another.
The principal immediate benefit of doing this, aside from a small
saving in the pg_rewrite data for views, is that these pseudo-RTEs
no longer trigger ruleutils.c's heuristic about qualifying variable
names when the rangetable's length is more than 1. That results
in quite a number of small simplifications in regression test outputs,
which are all to the good IMO.
Bump catversion because we need to dump a few more fields of
RTE_SUBQUERY RTEs. While those will always be zeroes anyway in
stored rules (because we'd never populate them until query rewrite)
they are useful for debugging, and it seems like we'd better make
sure to transmit such RTEs accurately in plans sent to parallel
workers. I don't think the executor actually examines these fields
after startup, but someday it might.
Amit Langote
Discussion: https://postgr.es/m/CA+HiwqEf7gPN4Hn+LoZ4tP2q_Qt7n3vw7-6fJKOf92tSEnX6Gg@mail.gmail.com
Prior to this, we only considered a full sort on the cheapest input path
and uniquifying any path which was already sorted in the required sort
order. Here we adjust create_final_distinct_paths() so that it also
adds an Incremental Sort path on any path which has presorted keys.
Additionally, this adjusts the parallel distinct code so that we now
consider sorting the cheapest partial path and incrementally sorting any
partial paths with presorted keys. Previously we didn't consider any
sorting for parallel distinct and only added a unique path atop any path
which had the required pathkeys already.
Author: David Rowley
Reviewed-by: Richard Guo
Discussion: https://postgr.es/m/CAApHDvo8Lz2H=42urBbfP65LTcEUOh288MT7DsG2_EWtW1AXHQ@mail.gmail.com
During the development of 728202b63, which was aimed at reducing the
number of sorts required to evaluate multiple window functions with
different WindowClause definitions, the code written sorted the
WindowClauses in reverse tleSortGroupRef order. There appears to be no
discussion in the thread which was opened to discuss the development of
this patch and no comments mentioning the fact that having the
WindowClauses in reverse tleSortGroupRef order makes it more likely that
the final WindowClause to be evaluated will provide presorted input to
the query's DISTINCT or ORDER BY clause. The reason for this is that the
tleSortGroupRef indexes are assigned for the DISTINCT and ORDER BY clauses
before they are for the WindowClauses PARTITION BY and ORDER BY clauses.
Putting the WindowClause with the lowest tleSortGroupRef last means that
it's more likely that no additional sorting is required for the query's
DISTINCT or ORDER BY clause.
All we're doing here is adding some tests and a comment to help ensure
that remains true and that we don't accidentally forget to consider this
again should we ever rewrite that code.
Author: Ankit Kumar Pandey, David Rowley
Discussion: https://postgr.es/m/CAApHDvq=g2=ny59f1bvwRVvupsgPHK-KjLPBsSL25fVuGZ4idQ@mail.gmail.com
WindowFuncs such as row_number() don't care if it's called with ROWS
UNBOUNDED PRECEDING AND CURRENT ROW or with RANGE UNBOUNDED PRECEDING AND
CURRENT ROW. The latter is less efficient as the RANGE option requires
that the executor check for peer rows, so using the ROW option instead
would cause less overhead. Because RANGE is part of the default frame
options for WindowClauses, it means WindowAgg is, by default, working much
harder than it needs to for window functions where the ROWS / RANGE option
has no effect on the window function's result.
On a test query from the discussion thread, a performance improvement of
344% was seen by using ROWS instead of RANGE.
Here we add a new support function node type to allow support functions to
be called for window functions so that the most optimal version of the
frame options can be set. The planner has been adjusted so that the frame
options are changed only if all window functions sharing the same window
clause agree on what the optimized frame options are.
Here we give the ability for row_number(), rank(), dense_rank(),
percent_rank(), cume_dist() and ntile() to alter their WindowClause's
frameOptions.
Reviewed-by: Vik Fearing, Erwin Brandstetter, Zhihong Yu
Discussion: https://postgr.es/m/CAGHENJ7LBBszxS+SkWWFVnBmOT2oVsBhDMB1DFrgerCeYa_DyA@mail.gmail.com
Discussion: https://postgr.es/m/CAApHDvohAKEtTXxq7Pc-ic2dKT8oZfbRKeEJP64M0B6+S88z+A@mail.gmail.com
1349d279 added query planner support to allow more efficient execution of
aggregate functions which have an ORDER BY or a DISTINCT clause. Prior to
that commit, the planner would only request that the lower planner produce
a plan with the order required for the GROUP BY clause and it would be
left up to nodeAgg.c to perform the final sort of records within each
group so that the aggregate transition functions were called in the
correct order. Now that the planner requests the lower planner produce a
plan with the GROUP BY and the ORDER BY / DISTINCT aggregates in mind,
there is the possibility that the planner chooses a plan which could be
less efficient than what would have been produced before 1349d279.
While developing 1349d279, I had in mind that Incremental Sort would help
us in cases where an index exists only on the GROUP BY column(s).
Incremental Sort would just replace the implicit tuplesorts which are
being performed in nodeAgg.c. However, because the planner has the
flexibility to instead choose a plan which just performs a full sort on
both the GROUP BY and ORDER BY / DISTINCT aggregate columns, there is
potential for the planner to make a bad choice. The costing for
Incremental Sort is not perfect as it assumes an even distribution of rows
to sort within each sort group.
Here we add an escape hatch in the form of the enable_presorted_aggregate
GUC. This will allow users to get the pre-PG16 behavior in cases where
they have no other means to convince the query planner to produce a plan
which only sorts on the GROUP BY column(s).
Discussion: https://postgr.es/m/CAApHDvr1Sm+g9hbv4REOVuvQKeDWXcKUAhmbK5K+dfun0s9CvA@mail.gmail.com
When incremental sorts were added in v13 a 1.5x pessimism factor was added
to the cost modal. Seemingly this was done because the cost modal only
has an estimate of the total number of input rows and the number of
presorted groups. It assumes that the input rows will be evenly
distributed throughout the presorted groups. The 1.5x pessimism factor
was added to slightly reduce the likelihood of incremental sorts being
used in the hope to avoid performance regressions where an incremental
sort plan was picked and turned out slower due to a large skew in the
number of rows in the presorted groups.
An additional quirk with the path generation code meant that we could
consider both a sort and an incremental sort on paths with presorted keys.
This meant that with the pessimism factor, it was possible that we opted
to perform a sort rather than an incremental sort when the given path had
presorted keys.
Here we remove the 1.5x pessimism factor to allow incremental sorts to
have a fairer chance at being chosen against a full sort.
Previously we would generally create a sort path on the cheapest input
path (if that wasn't sorted already) and incremental sort paths on any
path which had presorted keys. This meant that if the cheapest input path
wasn't completely sorted but happened to have presorted keys, we would
create a full sort path *and* an incremental sort path on that input path.
Here we change this logic so that if there are presorted keys, we only
create an incremental sort path, and create sort paths only when a full
sort is required.
Both the removal of the cost pessimism factor and the changes made to the
path generation make it more likely that incremental sorts will now be
chosen. That, of course, as with teaching the planner any new tricks,
means an increased likelihood that the planner will perform an incremental
sort when it's not the best method. Our standard escape hatch for these
cases is an enable_* GUC. enable_incremental_sort already exists for
this.
This came out of a report by Pavel Luzanov where he mentioned that the
master branch was choosing to perform a Seq Scan -> Sort -> Group
Aggregate for his query with an ORDER BY aggregate function. The v15 plan
for his query performed an Index Scan -> Group Aggregate, of course, the
aggregate performed the final sort internally in nodeAgg.c for the
aggregate's ORDER BY. The ideal plan would have been to use the index,
which provided partially sorted input then use an incremental sort to
provide the aggregate with the sorted input. This was not being chosen
due to the pessimism in the incremental sort cost modal, so here we remove
that and rationalize the path generation so that sort and incremental sort
plans don't have to needlessly compete. We assume that it's senseless
to ever use a full sort on a given input path where an incremental sort
can be performed.
Reported-by: Pavel Luzanov
Reviewed-by: Richard Guo
Discussion: https://postgr.es/m/9f61ddbf-2989-1536-b31e-6459370a6baa%40postgrespro.ru
9d9c02ccd added code to allow WindowAgg to take some shortcuts when a
monotonic WindowFunc reached some value that it could never come back
from due to the function's monotonic nature. That commit added a
runCondition field to WindowClause to store the condition which, when it
becomes false we can start taking shortcuts in nodeWindowAgg.c.
Here we fix an issue where subquery pullups didn't properly update the
runCondition to update the Vars to properly reference the new query level.
Here we also add a missing call to preprocess_expression() for the
WindowClause's runCondtion. The WindowFuncs in the targetlist will have
had this process done, so we must also do it for the WindowFuncs in the
runCondition so that they can be correctly found in the targetlist
during setrefs.c
Bug: #17709
Reported-by: Alexey Makhmutov
Author: Richard Guo, David Rowley
Discussion: https://postgr.es/m/17709-4f557160e3e8ee9a@postgresql.org
Backpatch-through: 15, where 9d9c02ccd was introduced
Currently, information about the permissions to be checked on relations
mentioned in a query is stored in their range table entries. So the
executor must scan the entire range table looking for relations that
need to have permissions checked. This can make the permission checking
part of the executor initialization needlessly expensive when many
inheritance children are present in the range range. While the
permissions need not be checked on the individual child relations, the
executor still must visit every range table entry to filter them out.
This commit moves the permission checking information out of the range
table entries into a new plan node called RTEPermissionInfo. Every
top-level (inheritance "root") RTE_RELATION entry in the range table
gets one and a list of those is maintained alongside the range table.
This new list is initialized by the parser when initializing the range
table. The rewriter can add more entries to it as rules/views are
expanded. Finally, the planner combines the lists of the individual
subqueries into one flat list that is passed to the executor for
checking.
To make it quick to find the RTEPermissionInfo entry belonging to a
given relation, RangeTblEntry gets a new Index field 'perminfoindex'
that stores the corresponding RTEPermissionInfo's index in the query's
list of the latter.
ExecutorCheckPerms_hook has gained another List * argument; the
signature is now:
typedef bool (*ExecutorCheckPerms_hook_type) (List *rangeTable,
List *rtePermInfos,
bool ereport_on_violation);
The first argument is no longer used by any in-core uses of the hook,
but we leave it in place because there may be other implementations that
do. Implementations should likely scan the rtePermInfos list to
determine which operations to allow or deny.
Author: Amit Langote <amitlangote09@gmail.com>
Discussion: https://postgr.es/m/CA+HiwqGjJDmUhDSfv-U2qhKJjt9ST7Xh9JXC_irsAQ1TAUsJYg@mail.gmail.com
Just because I'm a neatnik, and I'm currently working on
code in this area. It annoys me to not be able to pgindent
my patches without working around unrelated changes.
Tom Lane
Alvaro Herrera
Michael Paquier
David Rowley
Bruce Momjian
Andrew Dunstan