This patch adjusts the treatment of parameterized paths so that all paths
with the same parameterization (same set of required outer rels) for the
same relation will have the same rowcount estimate. We cache the rowcount
estimates to ensure that property, and hopefully save a few cycles too.
Doing this makes it practical for add_path_precheck to operate without
a rowcount estimate: it need only assume that paths with different
parameterizations never dominate each other, which is close enough to
true anyway for coarse filtering, because normally a more-parameterized
path should yield fewer rows thanks to having more join clauses to apply.
In add_path, we do the full nine yards of comparing rowcount estimates
along with everything else, so that we can discard parameterized paths that
don't actually have an advantage. This fixes some issues I'd found with
add_path rejecting parameterized paths on the grounds that they were more
expensive than not-parameterized ones, even though they yielded many fewer
rows and hence would be cheaper once subsequent joining was considered.
To make the same-rowcounts assumption valid, we have to require that any
parameterized path enforce *all* join clauses that could be obtained from
the particular set of outer rels, even if not all of them are useful for
indexing. This is required at both base scans and joins. It's a good
thing anyway since the net impact is that join quals are checked at the
lowest practical level in the join tree. Hence, discard the original
rather ad-hoc mechanism for choosing parameterization joinquals, and build
a better one that has a more principled rule for when clauses can be moved.
The original rule was actually buggy anyway for lack of knowledge about
which relations are part of an outer join's outer side; getting this right
requires adding an outer_relids field to RestrictInfo.
For some reason, in the original coding of the PlaceHolderVar mechanism
I had supposed that PlaceHolderVars couldn't propagate into subqueries.
That is of course entirely possible. When it happens, we need to treat
an outer-level PlaceHolderVar much like an outer Var or Aggref, that is
SS_replace_correlation_vars() needs to replace the PlaceHolderVar with
a Param, and then when building the finished SubPlan we have to provide
the PlaceHolderVar expression as an actual parameter for the SubPlan.
The handling of the contained expression is a bit delicate but it can be
treated exactly like an Aggref's expression.
In addition to the missing logic in subselect.c, prepjointree.c was failing
to search subqueries for PlaceHolderVars that need their relids adjusted
during subquery pullup. It looks like everyplace else that touches
PlaceHolderVars got it right, though.
Per report from Mark Murawski. In 9.1 and HEAD, queries affected by this
oversight would fail with "ERROR: Upper-level PlaceHolderVar found where
not expected". But in 9.0 and 8.4, you'd silently get possibly-wrong
answers, since the value transmitted into the subquery wouldn't go to null
when it should.
In commit 57664ed25e I tried to fix a bug
reported by Teodor Sigaev by making non-simple-Var output columns distinct
(by wrapping their expressions with dummy PlaceHolderVar nodes). This did
not work too well. Commit b28ffd0fcc fixed
some ensuing problems with matching to child indexes, but per a recent
report from Claus Stadler, constraint exclusion of UNION ALL subqueries was
still broken, because constant-simplification didn't handle the injected
PlaceHolderVars well either. On reflection, the original patch was quite
misguided: there is no reason to expect that EquivalenceClass child members
will be distinct. So instead of trying to make them so, we should ensure
that we can cope with the situation when they're not.
Accordingly, this patch reverts the code changes in the above-mentioned
commits (though the regression test cases they added stay). Instead, I've
added assorted defenses to make sure that duplicate EC child members don't
cause any problems. Teodor's original problem ("MergeAppend child's
targetlist doesn't match MergeAppend") is addressed more directly by
revising prepare_sort_from_pathkeys to let the parent MergeAppend's sort
list guide creation of each child's sort list.
In passing, get rid of add_sort_column; as far as I can tell, testing for
duplicate sort keys at this stage is dead code. Certainly it doesn't
trigger often enough to be worth expending cycles on in ordinary queries.
And keeping the test would've greatly complicated the new logic in
prepare_sort_from_pathkeys, because comparing pathkey list entries against
a previous output array requires that we not skip any entries in the list.
Back-patch to 9.1, like the previous patches. The only known issue in
this area that wasn't caused by the ill-advised previous patches was the
MergeAppend planning failure, which of course is not relevant before 9.1.
It's possible that we need some of the new defenses against duplicate child
EC entries in older branches, but until there's some clear evidence of that
I'm going to refrain from back-patching further.
Further reflection shows that a single callback isn't very workable if we
desire to let FDWs generate multiple Paths, because that forces the FDW to
do all work necessary to generate a valid Plan node for each Path. Instead
split the former PlanForeignScan API into three steps: GetForeignRelSize,
GetForeignPaths, GetForeignPlan. We had already bit the bullet of breaking
the 9.1 FDW API for 9.2, so this shouldn't cause very much additional pain,
and it's substantially more flexible for complex FDWs.
Add an fdw_private field to RelOptInfo so that the new functions can save
state there rather than possibly having to recalculate information two or
three times.
In addition, we'd not thought through what would be needed to allow an FDW
to set up subexpressions of its choice for runtime execution. We could
treat ForeignScan.fdw_private as an executable expression but that seems
likely to break existing FDWs unnecessarily (in particular, it would
restrict the set of node types allowable in fdw_private to those supported
by expression_tree_walker). Instead, invent a separate field fdw_exprs
which will receive the postprocessing appropriate for expression trees.
(One field is enough since it can be a list of expressions; also, we assume
the corresponding expression state tree(s) will be held within fdw_state,
so we don't need to add anything to ForeignScanState.)
Per review of Hanada Shigeru's pgsql_fdw patch. We may need to tweak this
further as we continue to work on that patch, but to me it feels a lot
closer to being right now.
The original API specification only allowed an FDW to create a single
access path, which doesn't seem like a terribly good idea in hindsight.
Instead, move the responsibility for building the Path node and calling
add_path() into the FDW's PlanForeignScan function. Now, it can do that
more than once if appropriate. There is no longer any need for the
transient FdwPlan struct, so get rid of that.
Etsuro Fujita, Shigeru Hanada, Tom Lane
This patch fixes the planner so that it can generate nestloop-with-
inner-indexscan plans even with one or more levels of joining between
the indexscan and the nestloop join that is supplying the parameter.
The executor was fixed to handle such cases some time ago, but the
planner was not ready. This should improve our plans in many situations
where join ordering restrictions formerly forced complete table scans.
There is probably a fair amount of tuning work yet to be done, because
of various heuristics that have been added to limit the number of
parameterized paths considered. However, we are not going to find out
what needs to be adjusted until the code gets some real-world use, so
it's time to get it in there where it can be tested easily.
Note API change for index AM amcostestimate functions. I'm not aware of
any non-core index AMs, but if there are any, they will need minor
adjustments.
In commit e2c2c2e8b1 I made use of nested
list structures to show which clauses went with which index columns, but
on reflection that's a data structure that only an old-line Lisp hacker
could love. Worse, it adds unnecessary complication to the many places
that don't much care which clauses go with which index columns. Revert
to the previous arrangement of flat lists of clauses, and instead add a
parallel integer list of column numbers. The places that care about the
pairing can chase both lists with forboth(), while the places that don't
care just examine one list the same as before.
The only real downside to this is that there are now two more lists that
need to be passed to amcostestimate functions in case they care about
column matching (which btcostestimate does, so not passing the info is not
an option). Rather than deal with 11-argument amcostestimate functions,
pass just the IndexPath and expect the functions to extract fields from it.
That gets us down to 7 arguments which is better than 11, and it seems
more future-proof against likely additions to the information we keep
about an index path.
It's potentially useful for an index to repeat the same indexable column
or expression in multiple index columns, if the columns have different
opclasses. (If they share opclasses too, the duplicate column is pretty
useless, but nonetheless we've allowed such cases since 9.0.) However,
the planner failed to cope with this, because createplan.c was relying on
simple equal() matching to figure out which index column each index qual
is intended for. We do have that information available upstream in
indxpath.c, though, so the fix is to not flatten the multi-level indexquals
list when putting it into an IndexPath. Then we can rely on the sublist
structure to identify target index columns in createplan.c. There's a
similar issue for index ORDER BYs (the KNNGIST feature), so introduce a
multi-level-list representation for that too. This adds a bit more
representational overhead, but we might more or less buy that back by not
having to search for matching index columns anymore in createplan.c;
likewise btcostestimate saves some cycles.
Per bug #6351 from Christian Rudolph. Likely symptoms include the "btree
index keys must be ordered by attribute" failure shown there, as well as
"operator MMMM is not a member of opfamily NNNN".
Although this is a pre-existing problem that can be demonstrated in 9.0 and
9.1, I'm not going to back-patch it, because the API changes in the planner
seem likely to break things such as index plugins. The corner cases where
this matters seem too narrow to justify possibly breaking things in a minor
release.
The need for this was debated when we put in the index-only-scan feature,
but at the time we had no near-term expectation of having AMs that could
support such scans for only some indexes; so we kept it simple. However,
the SP-GiST AM forces the issue, so let's fix it.
This patch only installs the new API; no behavior actually changes.
If we use a PlaceHolderVar from the outer relation in an inner indexscan,
we need to reference the PlaceHolderVar as such as the value to be passed
in from the outer relation. The previous code effectively tried to
reconstruct the PHV from its component expression, which doesn't work since
(a) the Vars therein aren't necessarily bubbled up far enough, and (b) it
would be the wrong semantics anyway because of the possibility that the PHV
is supposed to have gone to null at some point before the current join.
Point (a) led to "variable not found in subplan target list" planner
errors, but point (b) would have led to silently wrong answers.
Per report from Roger Niederland.
The uniqueness condition might fail to hold intra-transaction, and assuming
it does can give incorrect query results. Per report from Marti Raudsepp,
though this is not his proposed patch.
Back-patch to 9.0, where both these features were introduced. In the
released branches, add the new IndexOptInfo field to the end of the struct,
to try to minimize ABI breakage for third-party code that may be examining
that struct.
Add a column pg_class.relallvisible to remember the number of pages that
were all-visible according to the visibility map as of the last VACUUM
(or ANALYZE, or some other operations that update pg_class.relpages).
Use relallvisible/relpages, instead of an arbitrary constant, to estimate
how many heap page fetches can be avoided during an index-only scan.
This is pretty primitive and will no doubt see refinements once we've
acquired more field experience with the index-only scan mechanism, but
it's way better than using a constant.
Note: I had to adjust an underspecified query in the window.sql regression
test, because it was changing answers when the plan changed to use an
index-only scan. Some of the adjacent tests perhaps should be adjusted
as well, but I didn't do that here.
This commit changes index-only scans so that data is read directly from the
index tuple without first generating a faux heap tuple. The only immediate
benefit is that indexes on system columns (such as OID) can be used in
index-only scans, but this is necessary infrastructure if we are ever to
support index-only scans on expression indexes. The executor is now ready
for that, though the planner still needs substantial work to recognize
the possibility.
To do this, Vars in index-only plan nodes have to refer to index columns
not heap columns. I introduced a new special varno, INDEX_VAR, to mark
such Vars to avoid confusion. (In passing, this commit renames the two
existing special varnos to OUTER_VAR and INNER_VAR.) This allows
ruleutils.c to handle them with logic similar to what we use for subplan
reference Vars.
Since index-only scans are now fundamentally different from regular
indexscans so far as their expression subtrees are concerned, I also chose
to change them to have their own plan node type (and hence, their own
executor source file).
When a btree index contains all columns required by the query, and the
visibility map shows that all tuples on a target heap page are
visible-to-all, we don't need to fetch that heap page. This patch depends
on the previous patches that made the visibility map reliable.
There's a fair amount left to do here, notably trying to figure out a less
chintzy way of estimating the cost of an index-only scan, but the core
functionality seems ready to commit.
Robert Haas and Ibrar Ahmed, with some previous work by Heikki Linnakangas.
Formerly, set_subquery_pathlist and other creators of plans for subqueries
saved only the rangetable and rowMarks lists from the lower-level
PlannerInfo. But there's no reason not to remember the whole PlannerInfo,
and indeed this turns out to simplify matters in a number of places.
The immediate reason for doing this was so that the subroot will still be
accessible when we're trying to extract column statistics out of an
already-planned subquery. But now that I've done it, it seems like a good
code-beautification effort in its own right.
I also chose to get rid of the transient subrtable and subrowmark fields in
SubqueryScan nodes, in favor of having setrefs.c look up the subquery's
RelOptInfo. That required changing all the APIs in setrefs.c to pass
PlannerInfo not PlannerGlobal, which was a large but quite mechanical
transformation.
One side-effect not foreseen at the beginning is that this finally broke
inheritance_planner's assumption that replanning the same subquery RTE N
times would necessarily give interchangeable results each time. That
assumption was always pretty risky, but now we really have to make a
separate RTE for each instance so that there's a place to carry the
separate subroots.
The previous coding failed to account properly for the costs of evaluating
the input expressions of aggregates and window functions, as seen in a
recent gripe from Claudio Freire. (I said at the time that it wasn't
counting these costs at all; but on closer inspection, it was effectively
charging these costs once per output tuple. That is completely wrong for
aggregates, and not exactly right for window functions either.)
There was also a hard-wired assumption that aggregates and window functions
had procost 1.0, which is now fixed to respect the actual cataloged costs.
The costing of WindowAgg is still pretty bogus, since it doesn't try to
estimate the effects of spilling data to disk, but that seems like a
separate issue.
Instead of playing cute games with pathkeys, just build a direct
representation of the intended sub-select, and feed it through
query_planner to get a Path for the index access. This is a bit slower
than 9.1's previous method, since we'll duplicate most of the overhead of
query_planner; but since the whole optimization only applies to rather
simple single-table queries, that probably won't be much of a problem in
practice. The advantage is that we get to do the right thing when there's
a partial index that needs the implicit IS NOT NULL clause to be usable.
Also, although this makes planagg.c be a bit more closely tied to the
ordering of operations in grouping_planner, we can get rid of some coupling
to lower-level parts of the planner. Per complaint from Marti Raudsepp.
All expression nodes now have an explicit output-collation field, unless
they are known to only return a noncollatable data type (such as boolean
or record). Also, nodes that can invoke collation-aware functions store
a separate field that is the collation value to pass to the function.
This avoids confusion that arises when a function has collatable inputs
and noncollatable output type, or vice versa.
Also, replace the parser's on-the-fly collation assignment method with
a post-pass over the completed expression tree. This allows us to use
a more complex (and hopefully more nearly spec-compliant) assignment
rule without paying for it in extra storage in every expression node.
Fix assorted bugs in the planner's handling of collations by making
collation one of the defining properties of an EquivalenceClass and
by converting CollateExprs into discardable RelabelType nodes during
expression preprocessing.
This patch implements data-modifying WITH queries according to the
semantics that the updates all happen with the same command counter value,
and in an unspecified order. Therefore one WITH clause can't see the
effects of another, nor can the outer query see the effects other than
through the RETURNING values. And attempts to do conflicting updates will
have unpredictable results. We'll need to document all that.
This commit just fixes the code; documentation updates are waiting on
author.
Marko Tiikkaja and Hitoshi Harada
This commit provides the core code and documentation needed. A contrib
module test case will follow shortly.
Shigeru Hanada, Jan Urbanski, Heikki Linnakangas
That function was supposing that indexoid == 0 for a hypothetical index,
but that is not likely to be true in any non-toy implementation of an index
adviser, since assigning a fake OID is the only way to know at EXPLAIN time
which hypothetical index got selected. Fix by adding a flag to
IndexOptInfo to mark hypothetical indexes. Back-patch to 9.0 where
get_actual_variable_range() was added.
Gurjeet Singh
Flattening of subquery range tables during setrefs.c could lead to the
rangetable indexes in PlanRowMark nodes not matching up with the column
names previously assigned to the corresponding resjunk ctid (resp. tableoid
or wholerow) columns. Typical symptom would be either a "cannot extract
system attribute from virtual tuple" error or an Assert failure. This
wasn't a problem before 9.0 because we didn't support FOR UPDATE below the
top query level, and so the final flattening could never renumber an RTE
that was relevant to FOR UPDATE. Fix by using a plan-tree-wide unique
number for each PlanRowMark to label the associated resjunk columns, so
that the number need not change during flattening.
Per report from David Johnston (though I'm darned if I can see how this got
past initial testing of the relevant code). Back-patch to 9.0.
This adds collation support for columns and domains, a COLLATE clause
to override it per expression, and B-tree index support.
Peter Eisentraut
reviewed by Pavel Stehule, Itagaki Takahiro, Robert Haas, Noah Misch
This is a heavily revised version of builtin_knngist_core-0.9. The
ordering operators are no longer mixed in with actual quals, which would
have confused not only humans but significant parts of the planner.
Instead, ordering operators are carried separately throughout planning and
execution.
Since the API for ambeginscan and amrescan functions had to be changed
anyway, this commit takes the opportunity to rationalize that a bit.
RelationGetIndexScan no longer forces a premature index_rescan call;
instead, callers of index_beginscan must call index_rescan too. Aside from
making the AM-side initialization logic a bit less peculiar, this has the
advantage that we do not make a useless extra am_rescan call when there are
runtime key values. AMs formerly could not assume that the key values
passed to amrescan were actually valid; now they can.
Teodor Sigaev and Tom Lane
Formerly we looked up the operators associated with each index (caching
them in relcache) and then the planner looked up the btree opfamily
containing such operators in order to build the btree-centric pathkey
representation that describes the index's sort order. This is quite
pointless for btree indexes: we might as well just use the index's opfamily
information directly. That saves syscache lookup cycles during planning,
and furthermore allows us to eliminate the relcache's caching of operators
altogether, which may help in reducing backend startup time.
I added code to plancat.c to perform the same type of double lookup
on-the-fly if it's ever faced with a non-btree amcanorder index AM.
If such a thing actually becomes interesting for production, we should
replace that logic with some more-direct method for identifying the
corresponding btree opfamily; but it's not worth spending effort on now.
There is considerably more to do pursuant to my recent proposal to get rid
of sort-operator-based representations of sort orderings, but this patch
grabs some of the low-hanging fruit. I'll look at the remainder of that
work after the current commitfest.
This commit adds columns amoppurpose and amopsortfamily to pg_amop, and
column amcanorderbyop to pg_am. For the moment all the entries in
amcanorderbyop are "false", since the underlying support isn't there yet.
Also, extend the CREATE OPERATOR CLASS/ALTER OPERATOR FAMILY commands with
[ FOR SEARCH | FOR ORDER BY sort_operator_family ] clauses to allow the new
columns of pg_amop to be populated, and create pg_dump support for dumping
that information.
I also added some documentation, although it's perhaps a bit premature
given that the feature doesn't do anything useful yet.
Teodor Sigaev, Robert Haas, Tom Lane
We no longer need the terminating zero entry in opfamily[], so get rid of
it. Also replace assorted ad-hoc looping logic with simple for and foreach
constructs. This code is now noticeably more readable than it was an hour
ago; credit to Robert for seeing that it could be simplified.
Fix things so that top-N sorting can be used in child Sort nodes of a
MergeAppend node, when there is a LIMIT and no intervening joins or
grouping. Actually doing this on the executor side isn't too bad,
but it's a bit messier to get the planner to cost it properly.
Per gripe from Robert Haas.
In passing, fix an oversight in the original top-N-sorting patch:
query_planner should not assume that a LIMIT can be used to make an
explicit sort cheaper when there will be grouping or aggregation in
between. Possibly this should be back-patched, but I'm not sure the
mistake is serious enough to be a real problem in practice.
Per my recent proposal, get rid of all the direct inspection of indexes
and manual generation of paths in planagg.c. Instead, set up
EquivalenceClasses for the aggregate argument expressions, and let the
regular path generation logic deal with creating paths that can satisfy
those sort orders. This makes planagg.c a bit more visible to the rest
of the planner than it was originally, but the approach is basically a lot
cleaner than before. A major advantage of doing it this way is that we get
MIN/MAX optimization on inheritance trees (using MergeAppend of indexscans)
practically for free, whereas in the old way we'd have had to add a whole
lot more duplicative logic.
One small disadvantage of this approach is that MIN/MAX aggregates can no
longer exploit partial indexes having an "x IS NOT NULL" predicate, unless
that restriction or something that implies it is specified in the query.
The previous implementation was able to use the added "x IS NOT NULL"
condition as an extra predicate proof condition, but in this version we
rely entirely on indexes that are considered usable by the main planning
process. That seems a fair tradeoff for the simplicity and functionality
gained.
This patch eliminates the former need to sort the output of an Append scan
when an ordered scan of an inheritance tree is wanted. This should be
particularly useful for fast-start cases such as queries with LIMIT.
Original patch by Greg Stark, with further hacking by Hans-Jurgen Schonig,
Robert Haas, and Tom Lane.
The point of a PlaceHolderVar is to allow a non-strict expression to be
evaluated below an outer join, after which its value bubbles up like a Var
and can be forced to NULL when the outer join's semantics require that.
However, there was a serious design oversight in that, namely that we
didn't ensure that there was actually a correct place in the plan tree
to evaluate the placeholder :-(. It may be necessary to delay evaluation
of an outer join to ensure that a placeholder that should be evaluated
below the join can be evaluated there. Per recent bug report from Kirill
Simonov.
Back-patch to 8.4 where the PlaceHolderVar mechanism was introduced.
relation using the general PARAM_EXEC executor parameter mechanism, rather
than the ad-hoc kluge of passing the outer tuple down through ExecReScan.
The previous method was hard to understand and could never be extended to
handle parameters coming from multiple join levels. This patch doesn't
change the set of possible plans nor have any significant performance effect,
but it's necessary infrastructure for future generalization of the concept
of an inner indexscan plan.
ExecReScan's second parameter is now unused, so it's removed.
constraint exclusion on an inheritance set that is the target of an UPDATE
or DELETE query. Per gripe from Marc Cousin. Back-patch to 8.4 where
the feature was introduced.
This patch only supports seq_page_cost and random_page_cost as parameters,
but it provides the infrastructure to scalably support many more.
In particular, we may want to add support for effective_io_concurrency,
but I'm leaving that as future work for now.
Thanks to Tom Lane for design help and Alvaro Herrera for the review.
to be just a minor extension of the previous patch that made "x IS NULL"
indexable, because we can treat the IS NOT NULL condition as if it were
"x < NULL" or "x > NULL" (depending on the index's NULLS FIRST/LAST option),
just like IS NULL is treated like "x = NULL". Aside from any possible
usefulness in its own right, this is an important improvement for
index-optimized MAX/MIN aggregates: it is now reliably possible to get
a column's min or max value cheaply, even when there are a lot of nulls
cluttering the interesting end of the index.
by adding a requirement that build_join_rel add new join RelOptInfos to the
appropriate list immediately at creation. Per report from Robert Haas,
the list_concat_unique_ptr() calls that this change eliminates were taking
the lion's share of the runtime in larger join problems. This doesn't do
anything to fix the fundamental combinatorial explosion in large join
problems, but it should push out the threshold of pain a bit further.
Note: because this changes the order in which joinrel lists are built,
it might result in changes in selected plans in cases where different
alternatives have exactly the same costs. There is one example in the
regression tests.
mergejoin to shield it from doing mark/restore and refetches. Put an explicit
flag in MergePath so we can centralize the logic that knows about this,
and add costing logic that considers using Materialize even when it's not
forced by the previously-existing considerations. This is in response to
a discussion back in August that suggested that materializing an inner
indexscan can be helpful when the refetch percentage is high enough.
a lot of strange behaviors that occurred in join cases. We now identify the
"current" row for every joined relation in UPDATE, DELETE, and SELECT FOR
UPDATE/SHARE queries. If an EvalPlanQual recheck is necessary, we jam the
appropriate row into each scan node in the rechecking plan, forcing it to emit
only that one row. The former behavior could rescan the whole of each joined
relation for each recheck, which was terrible for performance, and what's much
worse could result in duplicated output tuples.
Also, the original implementation of EvalPlanQual could not re-use the recheck
execution tree --- it had to go through a full executor init and shutdown for
every row to be tested. To avoid this overhead, I've associated a special
runtime Param with each LockRows or ModifyTable plan node, and arranged to
make every scan node below such a node depend on that Param. Thus, by
signaling a change in that Param, the EPQ machinery can just rescan the
already-built test plan.
This patch also adds a prohibition on set-returning functions in the
targetlist of SELECT FOR UPDATE/SHARE. This is needed to avoid the
duplicate-output-tuple problem. It seems fairly reasonable since the
other restrictions on SELECT FOR UPDATE are meant to ensure that there
is a unique correspondence between source tuples and result tuples,
which an output SRF destroys as much as anything else does.
