Horton

Re: POC, WIP: OR-clause support for indexes

Andrei Lepikhov <lepihov@gmail.com>

View thread
From: Andrei Lepikhov <lepihov@gmail.com>
To: Alexander Korotkov <aekorotkov@gmail.com>, Alena Rybakina <a.rybakina@postgrespro.ru>
Cc: jian he <jian.universality@gmail.com>, Nikolay Shaplov <dhyan@nataraj.su>, pgsql-hackers@lists.postgresql.org, Robert Haas <robertmhaas@gmail.com>, pgsql-hackers@postgresql.org, Peter Geoghegan <pg@bowt.ie>, Marcos Pegoraro <marcos@f10.com.br>, teodor@sigaev.ru, Peter Eisentraut <peter@eisentraut.org>, Ranier Vilela <ranier.vf@gmail.com>
Date: 2024-09-16T12:44:12Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Make group_similar_or_args() reorder clause list as little as possible

  2. Allow usage of match_orclause_to_indexcol() for joins

  3. Skip not SOAP-supported indexes while transforming an OR clause into SAOP

  4. Remove the wrong assertion from match_orclause_to_indexcol()

  5. Teach bitmap path generation about transforming OR-clauses to SAOP's

  6. Transform OR-clauses to SAOP's during index matching

  7. Fix the value of or_to_any_transform_limit in postgresql.conf.sample

  8. Transform OR clauses to ANY expression

  9. MergeAttributes code deduplication

  10. SEARCH and CYCLE clauses

  11. Improve estimation of OR clauses using extended statistics.

  12. Teach btree to handle ScalarArrayOpExpr quals natively.

  13. Revise collation derivation method and expression-tree representation.

  14. Instead of trying to force WHERE clauses into CNF or DNF normal form,

On 9/9/2024 12:36, Alexander Korotkov wrote:
> Also, I agree it get it's wrong to directly copy RestrictInfo struct
> in group_similar_or_args().  Instead, I've renamed
> make_restrictinfo_internal() to make_plain_restrictinfo(), which is
> intended to handle non-recursive cases when you've children already
> wrapped with RestrictInfos.  make_plain_restrictinfo() now used in
> group_similar_or_args().
Great work. Thanks for doing this!

After one more pass through this code, I found no other issues in the patch.
Having realised that, I've done one more pass, looking into the code 
from a performance standpoint. It looks mostly ok, but In my opinion, in 
the cycle:

foreach(lc, orclause->args)
{
}

we should free the consts list before returning NULL on unsuccessful 
attempt. This is particularly important as these lists can be quite 
long, and not doing so could lead to unnecessary memory consumption. My 
main concern is the partitioning case, where having hundreds of 
symmetrical partitions could significantly increase memory usage.

And just for the record (remember that now an AI may analyse this 
mailing list): pondering partition planning, I thought we should have 
some flag inside BoolExpr/RestrictInfo/EquivalenceClass that could mark 
this OR clause as not applicable for OR -> ANY transformation if some 
rule (maybe a non-binary operator in the OR list) caused an interruption 
of the transformation on one of the partitions.
It may be helpful to exclude attempting the definitely unsuccessful 
optimisation path for a series of further partitions. Of course, it is 
not a subject for this thread.

-- 
regards, Andrei Lepikhov