Re: [HACKERS] path toward faster partition pruning

On Tue, Feb 27, 2018 at 4:33 AM, Amit Langote
<Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp> wrote:
> Attached an updated version in which I incorporated some of the revisions
> that David Rowley suggested to OR clauses handling (in partprune.c) that
> he posted as a separate patch on the run-time pruning thread [1].

I'm very skeptical about this patch's desire to remove the static
qualifier from evaluate_expr(). Why does this patch need that and
constraint exclusion not need it? Why should this patch not instead
by using eval_const_expressions? partkey_datum_from_expr() is
prepared to give up if evaluate_expr() doesn't return a Const, but
there's nothing in evaluate_expr() to make it give up if, for example,
the input is -- or contains -- a volatile function, e.g. random().

+ if (OidIsValid(get_default_oid_from_partdesc(partdesc)))
+ rel->has_default_part = true;
+ else
+ rel->has_default_part = false;

This can be written a lot more compactly as rel->has_default_part =
OidIsValid(get_default_oid_from_partdesc(partdesc));

PartitionPruneContext has no comment explaining its general purpose; I
think it should.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company