The standard way to check for list emptiness is to compare the
List pointer to NIL; our list code goes out of its way to ensure
that that is the only representation of an empty list. (An
acceptable alternative is a plain boolean test for non-null
pointer, but explicit mention of NIL is usually preferable.)
Various places didn't get that memo and expressed the condition
with list_length(), which might not be so bad except that there
were such a variety of ways to check it exactly: equal to zero,
less than or equal to zero, less than one, yadda yadda. In the
name of code readability, let's standardize all those spellings
as "list == NIL" or "list != NIL". (There's probably some
microscopic efficiency gain too, though few of these look to be
at all performance-critical.)
A very small number of cases were left as-is because they seemed
more consistent with other adjacent list_length tests that way.
Peter Smith, with bikeshedding from a number of us
Discussion: https://postgr.es/m/CAHut+PtQYe+ENX5KrONMfugf0q6NHg4hR5dAhqEXEc2eefFeig@mail.gmail.com
Commit fac1b470a thought we could check for set-returning functions
by testing only the top-level node in an expression tree. This is
wrong in itself, and to make matters worse it encouraged others
to make the same mistake, by exporting tlist.c's special-purpose
IS_SRF_CALL() as a widely-visible macro. I can't find any evidence
that anyone's taken the bait, but it was only a matter of time.
Use expression_returns_set() instead, and stuff the IS_SRF_CALL()
genie back in its bottle, this time with a warning label. I also
added a couple of cross-reference comments.
After a fair amount of fooling around, I've despaired of making
a robust test case that exposes the bug reliably, so no test case
here. (Note that the test case added by fac1b470a is itself
broken, in that it doesn't notice if you remove the code change.
The repro given by the bug submitter currently doesn't fail either
in v15 or HEAD, though I suspect that may indicate an unrelated bug.)
Per bug #17564 from Martijn van Oosterhout. Back-patch to v13,
as the faulty patch was.
Discussion: https://postgr.es/m/17564-c7472c2f90ef2da3@postgresql.org
ORDER BY / DISTINCT aggreagtes have, since implemented in Postgres, been
executed by always performing a sort in nodeAgg.c to sort the tuples in
the current group into the correct order before calling the transition
function on the sorted tuples. This was not great as often there might be
an index that could have provided pre-sorted input and allowed the
transition functions to be called as the rows come in, rather than having
to store them in a tuplestore in order to sort them once all the tuples
for the group have arrived.
Here we change the planner so it requests a path with a sort order which
supports the most amount of ORDER BY / DISTINCT aggregate functions and
add new code to the executor to allow it to support the processing of
ORDER BY / DISTINCT aggregates where the tuples are already sorted in the
correct order.
Since there can be many ORDER BY / DISTINCT aggregates in any given query
level, it's very possible that we can't find an order that suits all of
these aggregates. The sort order that the planner chooses is simply the
one that suits the most aggregate functions. We take the most strictly
sorted variation of each order and see how many aggregate functions can
use that, then we try again with the order of the remaining aggregates to
see if another order would suit more aggregate functions. For example:
SELECT agg(a ORDER BY a),agg2(a ORDER BY a,b) ...
would request the sort order to be {a, b} because {a} is a subset of the
sort order of {a,b}, but;
SELECT agg(a ORDER BY a),agg2(a ORDER BY c) ...
would just pick a plan ordered by {a} (we give precedence to aggregates
which are earlier in the targetlist).
SELECT agg(a ORDER BY a),agg2(a ORDER BY b),agg3(a ORDER BY b) ...
would choose to order by {b} since two aggregates suit that vs just one
that requires input ordered by {a}.
Author: David Rowley
Reviewed-by: Ronan Dunklau, James Coleman, Ranier Vilela, Richard Guo, Tom Lane
Discussion: https://postgr.es/m/CAApHDvpHzfo92%3DR4W0%2BxVua3BUYCKMckWAmo-2t_KiXN-wYH%3Dw%40mail.gmail.com
The select_outer_pathkeys_for_merge function made an attempt to build the
merge join pathkeys in the same order as query_pathkeys. This was done as
it may have led to no sort being required for an ORDER BY or GROUP BY
clause in the upper planner. However, this restriction seems overly
strict as it required that we match the query_pathkeys entirely or we
don't bother putting the merge join pathkeys in that order.
Here we relax this rule so that we use a prefix of the query_pathkeys
providing that prefix matches all of the join quals. This may provide the
upper planner with partially sorted input which will allow the use of
incremental sorts instead of full sorts.
Author: David Rowley
Reviewed-by: Richard Guo
Discussion: https://postgr.es/m/CAApHDvrtZu0PHVfDPFM4Yx3jNR2Wuwosv+T2zqa7LrhhBr2rRg@mail.gmail.com
Two callers of generate_useful_gather_paths were testing the wrong
thing when deciding whether to call that function: they checked for
being at the top of the current join subproblem, rather than being at
the actual top join. This'd result in failing to construct parallel
paths for a sub-join for which they might be useful.
While set_rel_pathlist() isn't actively broken, it seems best to
make its identical-in-intention test for this be like the other two.
This has been wrong all along, but given the lack of field complaints
I'm hesitant to back-patch into stable branches; we usually prefer
to avoid non-bug-fix changes in plan choices in minor releases.
It seems not too late for v15 though.
Richard Guo, reviewed by Antonin Houska and Tom Lane
Discussion: https://postgr.es/m/CAMbWs4-mH8Zf87-w+3P2J=nJB+5OyicO28ia9q_9o=Lamf_VHg@mail.gmail.com
Commit 4f658dc8 provided the traditional BSD fls() function in
src/port/fls.c so it could be used in several places. Later we added a
bunch of similar facilities in pg_bitutils.h, based on compiler
builtins that map to hardware instructions. It's a bit confusing to
have both 1-based and 0-based variants of this operation in use in
different parts of the tree, and neither is blessed by a standard.
Let's drop fls.c and the configure probe, and reuse the newer code.
Reviewed-by: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CA%2BhUKG%2B7dSX1XF8yFGmYk-%3D48dbjH2kmzZj16XvhbrWP-9BzRg%40mail.gmail.com
setrefs.c contains logic to discard no-op SubqueryScan nodes, that is,
ones that have no qual to check and copy the input targetlist unchanged.
(Formally it's not very nice to be applying such optimizations so late
in the planner, but there are practical reasons for it; mostly that we
can't unify relids between the subquery and the parent query until we
flatten the rangetable during setrefs.c.) This behavior falsifies our
previous cost estimates, since we would've charged cpu_tuple_cost per
row just to pass data through the node. Most of the time that's little
enough to not matter, but there are cases where this effect visibly
changes the plan compared to what you would've gotten with no
sub-select.
To improve the situation, make the callers of cost_subqueryscan tell
it whether they think the targetlist is trivial. cost_subqueryscan
already has the qual list, so it can check the other half of the
condition easily. It could make its own determination of tlist
triviality too, but doing so would be repetitive (for callers that
may call it several times) or unnecessarily expensive (for callers
that can determine this more cheaply than a general test would do).
This isn't a 100% solution, because createplan.c also does things
that can falsify any earlier estimate of whether the tlist is
trivial. However, it fixes nearly all cases in practice, if results
for the regression tests are anything to go by.
setrefs.c also contains logic to discard no-op Append and MergeAppend
nodes. We did have knowledge of that behavior at costing time, but
somebody failed to update it when a check on parallel-awareness was
added to the setrefs.c logic. Fix that while we're here.
These changes result in two minor changes in query plans shown in
our regression tests. Neither is relevant to the purposes of its
test case AFAICT.
Patch by me; thanks to Richard Guo for review.
Discussion: https://postgr.es/m/2581077.1651703520@sss.pgh.pa.us
This replaces all MemSet() calls with struct initialization where that
is easily and obviously possible. (For example, some cases have to
worry about padding bits, so I left those.)
(The same could be done with appropriate memset() calls, but this
patch is part of an effort to phase out MemSet(), so it doesn't touch
memset() calls.)
Reviewed-by: Ranier Vilela <ranier.vf@gmail.com>
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://www.postgresql.org/message-id/9847b13c-b785-f4e2-75c3-12ec77a3b05c@enterprisedb.com
No members of the buildfarm are using this version of Visual Studio,
resulting in all the code cleaned up here as being mostly dead, and
VS2017 is the oldest version still supported.
More versions could be cut, but the gain would be minimal, while
removing only VS2013 has the advantage to remove from the core code all
the dependencies on the value defined by _MSC_VER, where compatibility
tweaks have accumulated across the years mostly around locales and
strtof(), so that's a nice isolated cleanup.
Note that this commit additionally allows a revert of 3154e16. The
versions of Visual Studio now supported range from 2015 to 2022.
Author: Michael Paquier
Reviewed-by: Juan José Santamaría Flecha, Tom Lane, Thomas Munro, Justin
Pryzby
Discussion: https://postgr.es/m/YoH2IMtxcS3ncWn+@paquier.xyz
Truncating off the end of a freshly copied List is not a very efficient
way of copying the first N elements of a List.
In many of the cases that are updated here, the pattern was only being
used to remove the final element of a List. That's about the best case
for it, but there were many instances where the truncate trimming the List
down much further.
4cc832f94 added list_copy_head(), so let's use it in cases where it's
useful.
Author: David Rowley
Discussion: https://postgr.es/m/1986787.1657666922%40sss.pgh.pa.us
There are a few things that we could do a little better within
get_cheapest_group_keys_order():
1. We should be using list_free() rather than pfree() on a List.
2. We should use for_each_from() instead of manually coding a for loop to
skip the first n elements of a List
3. list_truncate(list_copy(...), n) is not a great way to copy the first n
elements of a list. Let's invent list_copy_head() for that. That way we
don't need to copy the entire list just to truncate it directly
afterwards.
4. We can simplify finding the cheapest cost by setting the cheapest cost
variable to DBL_MAX. That allows us to skip special-casing the initial
iteration of the loop.
Author: David Rowley
Discussion: https://postgr.es/m/CAApHDvrGyL3ft8waEkncG9y5HDMu5TFFJB1paoTC8zi9YK97Nw@mail.gmail.com
Backpatch-through: 15, where get_cheapest_group_keys_order was added.
9d9c02ccd added code to allow the executor to take shortcuts when quals
on monotonic window functions guaranteed that once the qual became false
it could never become true again. When possible, baserestrictinfo quals
are converted to become these quals, which we call run conditions.
Unfortunately, in 9d9c02ccd, I forgot to update
remove_unused_subquery_outputs to teach it about these run conditions.
This could cause a WindowFunc column which was unused in the target list
but referenced by an upper-level WHERE clause to be removed from the
subquery when the qual in the WHERE clause was converted into a window run
condition. Because of this, the entire WindowClause would be removed from
the query resulting in additional rows making it into the resultset when
they should have been filtered out by the WHERE clause.
Here we fix this by recording which target list items in the subquery have
run conditions. That gets passed along to remove_unused_subquery_outputs
to tell it not to remove these items from the target list.
Bug: #17495
Reported-by: Jeremy Evans
Reviewed-by: Richard Guo
Discussion: https://postgr.es/m/17495-7ffe2fa0b261b9fa@postgresql.org
Several places in the planner tried to clamp a double value to fit
in a "long" by doing
(long) Min(x, (double) LONG_MAX);
This is subtly incorrect, because it casts LONG_MAX to double and
potentially back again. If long is 64 bits then the double value
is inexact, and the platform might round it up to LONG_MAX+1
resulting in an overflow and an undesirably negative output.
While it's not hard to rewrite the expression into a safe form,
let's put it into a common function to reduce the risk of someone
doing it wrong in future.
In principle this is a bug fix, but since the problem could only
manifest with group count estimates exceeding 2^63, it seems unlikely
that anyone has actually hit this or will do so anytime soon. We're
fixing it mainly to satisfy fuzzer-type tools. That being the case,
a HEAD-only fix seems sufficient.
Andrey Lepikhov
Discussion: https://postgr.es/m/ebbc2efb-7ef9-bf2f-1ada-d6ec48f70e58@postgrespro.ru
In order to estimate the cache hit ratio of a Memoize node, one of the
inputs we require is the estimated number of times the Memoize node will
be rescanned. The higher this number, the large the cache hit ratio is
likely to become. Unfortunately, the value being passed as the number of
"calls" to the Memoize was incorrectly using the Nested Loop's
outer_path->parent->rows instead of outer_path->rows. This failed to
account for the fact that the outer_path might be parameterized by some
upper-level Nested Loop.
This problem could lead to Memoize plans appearing more favorable than
they might actually be. It could also lead to extended executor startup
times when work_mem values were large due to the planner setting overly
large MemoizePath->est_entries resulting in the Memoize hash table being
initially made much larger than might be required.
Fix this simply by passing outer_path->rows rather than
outer_path->parent->rows. Also, adjust the expected regression test
output for a plan change.
Reported-by: Pavel Stehule
Author: David Rowley
Discussion: https://postgr.es/m/CAFj8pRAMp%3DQsMi6sPQJ4W3hczoFJRvyXHJV3AZAZaMyTVM312Q%40mail.gmail.com
Backpatch-through: 14, where Memoize was introduced
SubqueryScan was always getting labeled with a rowcount estimate
appropriate for non-parallel cases. However, nodes that are
underneath a Gather should be treated as processing only one
worker's share of the rows, whether the particular node is explicitly
parallel-aware or not. Most non-scan-level node types get this
right automatically because they base their rowcount estimate on
that of their input sub-Path(s). But SubqueryScan didn't do that,
instead using the whole-relation rowcount estimate as if it were
a non-parallel-aware scan node. If there is a parallel-aware node
below the SubqueryScan, this is wrong, and it results in inflating
the cost estimates for nodes above the SubqueryScan, which can cause
us to not choose a parallel plan, or choose a silly one --- as indeed
is visible in the one regression test whose results change with this
patch. (Although that plan tree appears to contain no SubqueryScans,
there were some in it before setrefs.c deleted them.)
To fix, use path->subpath->rows not baserel->tuples as the number
of input tuples we'll process. This requires estimating the quals'
selectivity afresh, which is slightly annoying; but it shouldn't
really add much cost thanks to the caching done in RestrictInfo.
This is pretty clearly a bug fix, but I'll refrain from back-patching
as people might not appreciate plan choices changing in stable branches.
The fact that it took us this long to identify the bug suggests that
it's not a major problem.
Per report from bucoo, though this is not his proposed patch.
Discussion: https://postgr.es/m/202204121457159307248@sohu.com
inline_cte() expected to find exactly as many references to the
target CTE as its cterefcount indicates. While that should be
accurate for the tree as emitted by the parser, there are some
optimizations that occur upstream of here that could falsify it,
notably removal of unused subquery output expressions.
Trying to make the accounting 100% accurate seems expensive and
doomed to future breakage. It's not really worth it, because
all this code is protecting is downstream assumptions that every
referenced CTE has a plan. Let's convert those assertions to
regular test-and-elog just in case there's some actual problem,
and then drop the failing assertion.
Per report from Tomas Vondra (thanks also to Richard Guo for
analysis). Back-patch to v12 where the faulty code came in.
Discussion: https://postgr.es/m/29196a1e-ed47-c7ca-9be2-b1c636816183@enterprisedb.com
Window functions such as row_number() always return a value higher than
the previously returned value for tuples in any given window partition.
Traditionally queries such as;
SELECT * FROM (
SELECT *, row_number() over (order by c) rn
FROM t
) t WHERE rn <= 10;
were executed fairly inefficiently. Neither the query planner nor the
executor knew that once rn made it to 11 that nothing further would match
the outer query's WHERE clause. It would blindly continue until all
tuples were exhausted from the subquery.
Here we implement means to make the above execute more efficiently.
This is done by way of adding a pg_proc.prosupport function to various of
the built-in window functions and adding supporting code to allow the
support function to inform the planner if the window function is
monotonically increasing, monotonically decreasing, both or neither. The
planner is then able to make use of that information and possibly allow
the executor to short-circuit execution by way of adding a "run condition"
to the WindowAgg to allow it to determine if some of its execution work
can be skipped.
This "run condition" is not like a normal filter. These run conditions
are only built using quals comparing values to monotonic window functions.
For monotonic increasing functions, quals making use of the btree
operators for <, <= and = can be used (assuming the window function column
is on the left). You can see here that once such a condition becomes false
that a monotonic increasing function could never make it subsequently true
again. For monotonically decreasing functions the >, >= and = btree
operators for the given type can be used for run conditions.
The best-case situation for this is when there is a single WindowAgg node
without a PARTITION BY clause. Here when the run condition becomes false
the WindowAgg node can simply return NULL. No more tuples will ever match
the run condition. It's a little more complex when there is a PARTITION
BY clause. In this case, we cannot return NULL as we must still process
other partitions. To speed this case up we pull tuples from the outer
plan to check if they're from the same partition and simply discard them
if they are. When we find a tuple belonging to another partition we start
processing as normal again until the run condition becomes false or we run
out of tuples to process.
When there are multiple WindowAgg nodes to evaluate then this complicates
the situation. For intermediate WindowAggs we must ensure we always
return all tuples to the calling node. Any filtering done could lead to
incorrect results in WindowAgg nodes above. For all intermediate nodes,
we can still save some work when the run condition becomes false. We've
no need to evaluate the WindowFuncs anymore. Other WindowAgg nodes cannot
reference the value of these and these tuples will not appear in the final
result anyway. The savings here are small in comparison to what can be
saved in the top-level WingowAgg, but still worthwhile.
Intermediate WindowAgg nodes never filter out tuples, but here we change
WindowAgg so that the top-level WindowAgg filters out tuples that don't
match the intermediate WindowAgg node's run condition. Such filters
appear in the "Filter" clause in EXPLAIN for the top-level WindowAgg node.
Here we add prosupport functions to allow the above to work for;
row_number(), rank(), dense_rank(), count(*) and count(expr). It appears
technically possible to do the same for min() and max(), however, it seems
unlikely to be useful enough, so that's not done here.
Bump catversion
Author: David Rowley
Reviewed-by: Andy Fan, Zhihong Yu
Discussion: https://postgr.es/m/CAApHDvqvp3At8++yF8ij06sdcoo1S_b2YoaT9D4Nf+MObzsrLQ@mail.gmail.com
postgres_fdw would push ORDER BY clauses to the remote side without
verifying that the sort operator is safe to ship. Moreover, it failed
to print a suitable USING clause if the sort operator isn't default
for the sort expression's type. The net result of this is that the
remote sort might not have anywhere near the semantics we expect,
which'd be disastrous for locally-performed merge joins in particular.
We addressed similar issues in the context of ORDER BY within an
aggregate function call in commit 7012b132d, but failed to notice
that query-level ORDER BY was broken. Thus, much of the necessary
logic already existed, but it requires refactoring to be usable
in both cases.
Back-patch to all supported branches. In HEAD only, remove the
core code's copy of find_em_expr_for_rel, which is no longer used
and really should never have been pushed into equivclass.c in the
first place.
Ronan Dunklau, per report from David Rowley;
reviews by David Rowley, Ranier Vilela, and myself
Discussion: https://postgr.es/m/CAApHDvr4OeC2DBVY--zVP83-K=bYrTD7F8SZDhN4g+pj2f2S-A@mail.gmail.com
When evaluating a query with a multi-column GROUP BY clause using sort,
the cost may be heavily dependent on the order in which the keys are
compared when building the groups. Grouping does not imply any ordering,
so we're allowed to compare the keys in arbitrary order, and a Hash Agg
leverages this. But for Group Agg, we simply compared keys in the order
as specified in the query. This commit explores alternative ordering of
the keys, trying to find a cheaper one.
In principle, we might generate grouping paths for all permutations of
the keys, and leave the rest to the optimizer. But that might get very
expensive, so we try to pick only a couple interesting orderings based
on both local and global information.
When planning the grouping path, we explore statistics (number of
distinct values, cost of the comparison function) for the keys and
reorder them to minimize comparison costs. Intuitively, it may be better
to perform more expensive comparisons (for complex data types etc.)
last, because maybe the cheaper comparisons will be enough. Similarly,
the higher the cardinality of a key, the lower the probability we’ll
need to compare more keys. The patch generates and costs various
orderings, picking the cheapest ones.
The ordering of group keys may interact with other parts of the query,
some of which may not be known while planning the grouping. E.g. there
may be an explicit ORDER BY clause, or some other ordering-dependent
operation, higher up in the query, and using the same ordering may allow
using either incremental sort or even eliminate the sort entirely.
The patch generates orderings and picks those minimizing the comparison
cost (for various pathkeys), and then adds orderings that might be
useful for operations higher up in the plan (ORDER BY, etc.). Finally,
it always keeps the ordering specified in the query, on the assumption
the user might have additional insights.
This introduces a new GUC enable_group_by_reordering, so that the
optimization may be disabled if needed.
The original patch was proposed by Teodor Sigaev, and later improved and
reworked by Dmitry Dolgov. Reviews by a number of people, including me,
Andrey Lepikhov, Claudio Freire, Ibrar Ahmed and Zhihong Yu.
Author: Dmitry Dolgov, Teodor Sigaev, Tomas Vondra
Reviewed-by: Tomas Vondra, Andrey Lepikhov, Claudio Freire, Ibrar Ahmed, Zhihong Yu
Discussion: https://postgr.es/m/7c79e6a5-8597-74e8-0671-1c39d124c9d6%40sigaev.ru
Discussion: https://postgr.es/m/CA%2Bq6zcW_4o2NC0zutLkOJPsFt80megSpX_dVRo6GK9PC-Jx_Ag%40mail.gmail.com
This introduces the SQL/JSON functions for querying JSON data using
jsonpath expressions. The functions are:
JSON_EXISTS()
JSON_QUERY()
JSON_VALUE()
All of these functions only operate on jsonb. The workaround for now is
to cast the argument to jsonb.
JSON_EXISTS() tests if the jsonpath expression applied to the jsonb
value yields any values. JSON_VALUE() must return a single value, and an
error occurs if it tries to return multiple values. JSON_QUERY() must
return a json object or array, and there are various WRAPPER options for
handling scalar or multi-value results. Both these functions have
options for handling EMPTY and ERROR conditions.
Nikita Glukhov
Reviewers have included (in no particular order) Andres Freund, Alexander
Korotkov, Pavel Stehule, Andrew Alsup, Erik Rijkers, Zihong Yu,
Himanshu Upadhyaya, Daniel Gustafsson, Justin Pryzby.
Discussion: https://postgr.es/m/cd0bb935-0158-78a7-08b5-904886deac4b@postgrespro.ru
Up to now, the planner estimated the size of a recursive query's
worktable as 10 times the size of the non-recursive term. It's hard
to see how to do significantly better than that automatically, but
we can give users control over the multiplier to allow tuning for
specific use-cases. The default behavior remains the same.
Simon Riggs
Discussion: https://postgr.es/m/CANbhV-EuaLm4H3g0+BSTYHEGxJj3Kht0R+rJ8vT57Dejnh=_nA@mail.gmail.com
The planner needs to treat GroupingFunc like Aggref for many purposes,
in particular with respect to processing of the argument expressions,
which are not to be evaluated at runtime. A few places hadn't gotten
that memo, notably including subselect.c's processing of outer-level
aggregates. This resulted in assertion failures or wrong plans for
cases in which a GROUPING() construct references an outer aggregation
level.
Also fix missing special cases for GroupingFunc in cost_qual_eval
(resulting in wrong cost estimates for GROUPING(), although it's
not clear that that would affect plan shapes in practice) and in
ruleutils.c (resulting in excess parentheses in pretty-print mode).
Per bug #17088 from Yaoguang Chen. Back-patch to all supported
branches.
Richard Guo, Tom Lane
Discussion: https://postgr.es/m/17088-e33882b387de7f5c@postgresql.org
In sort_inner_and_outer we iterate a list of PathKey elements, but the
variable is declared as (List *). This mistake is benign, because we
only pass the pointer to lcons() and never dereference it.
This exists since ~2004, but it's confusing. So fix and backpatch to all
supported branches.
Backpatch-through: 10
Discussion: https://postgr.es/m/bf3a6ea1-a7d8-7211-0669-189d5c169374%40enterprisedb.com
When building append paths, we've been looking only at startup and total
costs for the paths. When building fractional paths that may eliminate
the cheapest one, because it may be dominated by two separate paths (one
for startup, one for total cost).
This extends generate_orderedappend_paths() to also consider which paths
have lowest fractional cost. Currently we only consider paths matching
pathkeys - in the future this may be improved by also considering paths
that are only partially sorted, with an incremental sort on top.
Original report of an issue by Arne Roland, patch by me (based on a
suggestion by Tom Lane).
Reviewed-by: Arne Roland, Zhihong Yu
Discussion: https://postgr.es/m/e8f9ec90-546d-e948-acce-0525f3e92773%40enterprisedb.com
Discussion: https://postgr.es/m/1581042da8044e71ada2d6e3a51bf7bb%40index.de
We can revert the code changes of commit b5febc1d1 now, because
commit 9a3ddeb51 installed a real solution for the difficulty
that b5febc1d1 just dodged, namely that the planner might pick
the wrong one of several index columns nominally containing the
same value. It only matters which one we pick if we pick one
that's not returnable, and that mistake is now foreclosed.
Although both of the aforementioned commits were back-patched,
I don't feel a need to take any risk by back-patching this one.
The cases that it improves are very corner-ish.
Discussion: https://postgr.es/m/3179992.1641150853@sss.pgh.pa.us
Memoize would always use the hash equality operator for the cache key
types to determine if the current set of parameters were the same as some
previously cached set. Certain types such as floating points where -0.0
and +0.0 differ in their binary representation but are classed as equal by
the hash equality operator may cause problems as unless the join uses the
same operator it's possible that whichever join operator is being used
would be able to distinguish the two values. In which case we may
accidentally return in the incorrect rows out of the cache.
To fix this here we add a binary mode to Memoize to allow it to the
current set of parameters to previously cached values by comparing
bit-by-bit rather than logically using the hash equality operator. This
binary mode is always used for LATERAL joins and it's used for normal
joins when any of the join operators are not hashable.
Reported-by: Tom Lane
Author: David Rowley
Discussion: https://postgr.es/m/3004308.1632952496@sss.pgh.pa.us
Backpatch-through: 14, where Memoize was added
In v14, because we don't have a field in RestrictInfo to cache both the
left and right type's hash equality operator, we just restrict the scope
of Memoize to only when the left and right types of a RestrictInfo are the
same.
In master we add another field to RestrictInfo and cache both hash
equality operators.
Reported-by: Jaime Casanova
Author: David Rowley
Discussion: https://postgr.es/m/20210929185544.GB24346%40ahch-to
Backpatch-through: 14
Up to now the size of a query's rangetable has been limited by the
constants INNER_VAR et al, which mustn't be equal to any real
rangetable index. 65000 doubtless seemed like enough for anybody,
and it still is orders of magnitude larger than the number of joins
we can realistically handle. However, we need a rangetable entry
for each child partition that is (or might be) processed by a query.
Queries with a few thousand partitions are getting more realistic,
so that the day when that limit becomes a problem is in sight,
even if it's not here yet. Hence, let's raise the limit.
Rather than just increase the values of INNER_VAR et al, this patch
adopts the approach of making them small negative values, so that
rangetables could theoretically become as long as INT_MAX.
The bulk of the patch is concerned with changing Var.varno and some
related variables from "Index" (unsigned int) to plain "int". This
is basically cosmetic, with little actual effect other than to help
debuggers print their values nicely. As such, I've only bothered
with changing places that could actually see INNER_VAR et al, which
the parser and most of the planner don't. We do have to be careful
in places that are performing less/greater comparisons on varnos,
but there are very few such places, other than the IS_SPECIAL_VARNO
macro itself.
A notable side effect of this patch is that while it used to be
possible to add INNER_VAR et al to a Bitmapset, that will now
draw an error. I don't see any likelihood that it wouldn't be a
bug to include these fake varnos in a bitmapset of real varnos,
so I think this is all to the good.
Although this touches outfuncs/readfuncs, I don't think a catversion
bump is required, since stored rules would never contain Vars
with these fake varnos.
Andrey Lepikhov and Tom Lane, after a suggestion by Peter Eisentraut
Discussion: https://postgr.es/m/43c7f2f5-1e27-27aa-8c65-c91859d15190@postgrespro.ru
959d00e9d added the ability to make use of an Append node instead of a
MergeAppend when we wanted to perform a scan of a partitioned table and
the required sort order was the same as the partitioned keys and the
partitioned table was defined in such a way that earlier partitions were
guaranteed to only contain lower-order values than later partitions.
However, previously we didn't allow these ordered partition scans for
LIST partitioned table when there were any partitions that allowed
multiple Datums. This was a very cheap check to make and we could likely
have done a little better by checking if there were interleaved
partitions, but at the time we didn't have visibility about which
partitions were pruned, so we still may have disallowed cases where all
interleaved partitions were pruned.
Since 475dbd0b7, we now have knowledge of pruned partitions, we can do a
much better job inside partitions_are_ordered().
Here we pass which partitions survived partition pruning into
partitions_are_ordered() and, for LIST partitioning, have it check to see
if any live partitions exist that are also in the new "interleaved_parts"
field defined in PartitionBoundInfo.
For RANGE partitioning we can relax the code which caused the partitions
to be unordered if a DEFAULT partition existed. Since we now know which
partitions were pruned, partitions_are_ordered() now returns true when the
DEFAULT partition was pruned.
Reviewed-by: Amit Langote, Zhihong Yu
Discussion: https://postgr.es/m/CAApHDvrdoN_sXU52i=QDXe2k3WAo=EVry29r2+Tq2WYcn2xhEA@mail.gmail.com
For partitioned tables with large numbers of partitions where queries are
able to prune all but a very small number of partitions, the time spent in
the planner looping over RelOptInfo.part_rels checking for non-NULL
RelOptInfos could become a large portion of the overall planning time.
Here we add a Bitmapset that records the non-pruned partitions. This
allows us to more efficiently skip the pruned partitions by looping over
the Bitmapset.
This will cause a very slight slow down in cases where no or not many
partitions could be pruned, however, those cases are already slow to plan
anyway and the overhead of looping over the Bitmapset would be
unmeasurable when compared with the other tasks such as path creation for
a large number of partitions.
Reviewed-by: Amit Langote, Zhihong Yu
Discussion: https://postgr.es/m/CAApHDvqnPx6JnUuPwaf5ao38zczrAb9mxt9gj4U1EKFfd4AqLA@mail.gmail.com
The point of introducing the hash_mem_multiplier GUC was to let users
reproduce the old behavior of hash aggregation, i.e. that it could use
more than work_mem at need. However, the implementation failed to get
the job done on Win64, where work_mem is clamped to 2GB to protect
various places that calculate memory sizes using "long int". As
written, the same clamp was applied to hash_mem. This resulted in
severe performance regressions for queries requiring a bit more than
2GB for hash aggregation, as they now spill to disk and there's no
way to stop that.
Getting rid of the work_mem restriction seems like a good idea, but
it's a big job and could not conceivably be back-patched. However,
there's only a fairly small number of places that are concerned with
the hash_mem value, and it turns out to be possible to remove the
restriction there without too much code churn or any ABI breaks.
So, let's do that for now to fix the regression, and leave the
larger task for another day.
This patch does introduce a bit more infrastructure that should help
with the larger task, namely pg_bitutils.h support for working with
size_t values.
Per gripe from Laurent Hasson. Back-patch to v13 where the
behavior change came in.
Discussion: https://postgr.es/m/997817.1627074924@sss.pgh.pa.us
Discussion: https://postgr.es/m/MN2PR15MB25601E80A9B6D1BA6F592B1985E39@MN2PR15MB2560.namprd15.prod.outlook.com
"Result Cache" was never a great name for this node, but nobody managed
to come up with another name that anyone liked enough. That was until
David Johnston mentioned "Node Memoization", which Tom Lane revised to
just "Memoize". People seem to like "Memoize", so let's do the rename.
Reviewed-by: Justin Pryzby
Discussion: https://postgr.es/m/20210708165145.GG1176@momjian.us
Backpatch-through: 14, where Result Cache was introduced
Code added in 9e215378d to disable building of Result Cache paths when
not all join conditions are part of the parameterization of a unique
join failed to first check if the inner path's param_info was set before
checking the param_info's ppi_clauses.
Add a check for NULL values here and just bail on trying to build the
path if param_info is NULL. lateral_vars are not considered when
deciding if the join is unique, so we're not missing out on doing the
optimization when there are lateral_vars and no param_info.
Reported-by: Coverity, via Tom Lane
Discussion: https://postgr.es/m/457998.1621779290@sss.pgh.pa.us
When the planner considered using a Result Cache node to cache results
from the inner side of a Nested Loop Join, it failed to consider that the
inner path's parameterization may not be the entire join condition. If
the join was marked as inner_unique then we may accidentally put the cache
in singlerow mode. This meant that entries would be marked as complete
after caching the first row. That was wrong as if only part of the join
condition was parameterized then the uniqueness of the unique join was not
guaranteed at the Result Cache's level. The uniqueness is only guaranteed
after Nested Loop applies the join filter. If subsequent rows were found,
this would lead to:
ERROR: cache entry already complete
This could have been fixed by only putting the cache in singlerow mode if
the entire join condition was parameterized. However, Nested Loop will
only read its inner side so far as the first matching row when the join is
unique, so that might mean we never get an opportunity to mark cache
entries as complete. Since non-complete cache entries are useless for
subsequent lookups, we just don't bother considering a Result Cache path
in this case.
In passing, remove the XXX comment that claimed the above ERROR might be
better suited to be an Assert. After there being an actual case which
triggered it, it seems better to keep it an ERROR.
Reported-by: David Christensen
Discussion: https://postgr.es/m/CAOxo6X+dy-V58iEPFgst8ahPKEU+38NZzUuc+a7wDBZd4TrHMQ@mail.gmail.com
I didn't particularly like this function name, as it fails to
express what's going on. Also, returning the sort expression
alone isn't too helpful --- typically, a caller would also
need some other fields of the EquivalenceMember. But the
sole caller really only needs a bool result, so let's make
it "bool relation_can_be_sorted_early()".
Discussion: https://postgr.es/m/91f3ec99-85a4-fa55-ea74-33f85a5c651f@swarm64.com
An oversight introduced by the incremental-sort patches caused
"could not find pathkey item to sort" errors in some situations
where a sort key involves an aggregate or window function.
The basic problem here is that find_em_expr_usable_for_sorting_rel
isn't properly modeling what prepare_sort_from_pathkeys will do
later. Rather than hoping we can keep those functions in sync,
let's refactor so that they actually share the code for
identifying a suitable sort expression.
With this refactoring, tlist.c's tlist_member_ignore_relabel
is unused. I removed it in HEAD but left it in place in v13,
in case any extensions are using it.
Per report from Luc Vlaming. Back-patch to v13 where the
problem arose.
James Coleman and Tom Lane
Discussion: https://postgr.es/m/91f3ec99-85a4-fa55-ea74-33f85a5c651f@swarm64.com
ScalarArrayOpExprs with "useOr=true" and a set of Consts on the righthand
side have traditionally been evaluated by using a linear search over the
array. When these arrays contain large numbers of elements then this
linear search could become a significant part of execution time.
Here we add a new method of evaluating ScalarArrayOpExpr expressions to
allow them to be evaluated by first building a hash table containing each
element, then on subsequent evaluations, we just probe that hash table to
determine if there is a match.
The planner is in charge of determining when this optimization is possible
and it enables it by setting hashfuncid in the ScalarArrayOpExpr. The
executor will only perform the hash table evaluation when the hashfuncid
is set.
This means that not all cases are optimized. For example CHECK constraints
containing an IN clause won't go through the planner, so won't get the
hashfuncid set. We could maybe do something about that at some later
date. The reason we're not doing it now is from fear that we may slow
down cases where the expression is evaluated only once. Those cases can
be common, for example, a single row INSERT to a table with a CHECK
constraint containing an IN clause.
In the planner, we enable this when there are suitable hash functions for
the ScalarArrayOpExpr's operator and only when there is at least
MIN_ARRAY_SIZE_FOR_HASHED_SAOP elements in the array. The threshold is
currently set to 9.
Author: James Coleman, David Rowley
Reviewed-by: David Rowley, Tomas Vondra, Heikki Linnakangas
Discussion: https://postgr.es/m/CAAaqYe8x62+=wn0zvNKCj55tPpg-JBHzhZFFc6ANovdqFw7-dA@mail.gmail.com
Here we add a new executor node type named "Result Cache". The planner
can include this node type in the plan to have the executor cache the
results from the inner side of parameterized nested loop joins. This
allows caching of tuples for sets of parameters so that in the event that
the node sees the same parameter values again, it can just return the
cached tuples instead of rescanning the inner side of the join all over
again. Internally, result cache uses a hash table in order to quickly
find tuples that have been previously cached.
For certain data sets, this can significantly improve the performance of
joins. The best cases for using this new node type are for join problems
where a large portion of the tuples from the inner side of the join have
no join partner on the outer side of the join. In such cases, hash join
would have to hash values that are never looked up, thus bloating the hash
table and possibly causing it to multi-batch. Merge joins would have to
skip over all of the unmatched rows. If we use a nested loop join with a
result cache, then we only cache tuples that have at least one join
partner on the outer side of the join. The benefits of using a
parameterized nested loop with a result cache increase when there are
fewer distinct values being looked up and the number of lookups of each
value is large. Also, hash probes to lookup the cache can be much faster
than the hash probe in a hash join as it's common that the result cache's
hash table is much smaller than the hash join's due to result cache only
caching useful tuples rather than all tuples from the inner side of the
join. This variation in hash probe performance is more significant when
the hash join's hash table no longer fits into the CPU's L3 cache, but the
result cache's hash table does. The apparent "random" access of hash
buckets with each hash probe can cause a poor L3 cache hit ratio for large
hash tables. Smaller hash tables generally perform better.
The hash table used for the cache limits itself to not exceeding work_mem
* hash_mem_multiplier in size. We maintain a dlist of keys for this cache
and when we're adding new tuples and realize we've exceeded the memory
budget, we evict cache entries starting with the least recently used ones
until we have enough memory to add the new tuples to the cache.
For parameterized nested loop joins, we now consider using one of these
result cache nodes in between the nested loop node and its inner node. We
determine when this might be useful based on cost, which is primarily
driven off of what the expected cache hit ratio will be. Estimating the
cache hit ratio relies on having good distinct estimates on the nested
loop's parameters.
For now, the planner will only consider using a result cache for
parameterized nested loop joins. This works for both normal joins and
also for LATERAL type joins to subqueries. It is possible to use this new
node for other uses in the future. For example, to cache results from
correlated subqueries. However, that's not done here due to some
difficulties obtaining a distinct estimation on the outer plan to
calculate the estimated cache hit ratio. Currently we plan the inner plan
before planning the outer plan so there is no good way to know if a result
cache would be useful or not since we can't estimate the number of times
the subplan will be called until the outer plan is generated.
The functionality being added here is newly introducing a dependency on
the return value of estimate_num_groups() during the join search.
Previously, during the join search, we only ever needed to perform
selectivity estimations. With this commit, we need to use
estimate_num_groups() in order to estimate what the hit ratio on the
result cache will be. In simple terms, if we expect 10 distinct values
and we expect 1000 outer rows, then we'll estimate the hit ratio to be
99%. Since cache hits are very cheap compared to scanning the underlying
nodes on the inner side of the nested loop join, then this will
significantly reduce the planner's cost for the join. However, it's
fairly easy to see here that things will go bad when estimate_num_groups()
incorrectly returns a value that's significantly lower than the actual
number of distinct values. If this happens then that may cause us to make
use of a nested loop join with a result cache instead of some other join
type, such as a merge or hash join. Our distinct estimations have been
known to be a source of trouble in the past, so the extra reliance on them
here could cause the planner to choose slower plans than it did previous
to having this feature. Distinct estimations are also fairly hard to
estimate accurately when several tables have been joined already or when a
WHERE clause filters out a set of values that are correlated to the
expressions we're estimating the number of distinct value for.
For now, the costing we perform during query planning for result caches
does put quite a bit of faith in the distinct estimations being accurate.
When these are accurate then we should generally see faster execution
times for plans containing a result cache. However, in the real world, we
may find that we need to either change the costings to put less trust in
the distinct estimations being accurate or perhaps even disable this
feature by default. There's always an element of risk when we teach the
query planner to do new tricks that it decides to use that new trick at
the wrong time and causes a regression. Users may opt to get the old
behavior by turning the feature off using the enable_resultcache GUC.
Currently, this is enabled by default. It remains to be seen if we'll
maintain that setting for the release.
Additionally, the name "Result Cache" is the best name I could think of
for this new node at the time I started writing the patch. Nobody seems
to strongly dislike the name. A few people did suggest other names but no
other name seemed to dominate in the brief discussion that there was about
names. Let's allow the beta period to see if the current name pleases
enough people. If there's some consensus on a better name, then we can
change it before the release. Please see the 2nd discussion link below
for the discussion on the "Result Cache" name.
Author: David Rowley
Reviewed-by: Andy Fan, Justin Pryzby, Zhihong Yu, Hou Zhijie
Tested-By: Konstantin Knizhnik
Discussion: https://postgr.es/m/CAApHDvrPcQyQdWERGYWx8J%2B2DLUNgXu%2BfOSbQ1UscxrunyXyrQ%40mail.gmail.com
Discussion: https://postgr.es/m/CAApHDvq=yQXr5kqhRviT2RhNKwToaWr9JAN5t+5_PzhuRJ3wvg@mail.gmail.com
This removes "Add Result Cache executor node". It seems that something
weird is going on with the tracking of cache hits and misses as
highlighted by many buildfarm animals. It's not yet clear what the
problem is as other parts of the plan indicate that the cache did work
correctly, it's just the hits and misses that were being reported as 0.
This is especially a bad time to have the buildfarm so broken, so
reverting before too many more animals go red.
Discussion: https://postgr.es/m/CAApHDvq_hydhfovm4=izgWs+C5HqEeRScjMbOgbpC-jRAeK3Yw@mail.gmail.com
