Mario Weilguni <mweilguni@sime.com> writes:
> Thanks for the info. Changing this to use extract is no real problem, I was
> just curious if this is intendend behaviour.
From the CVS logs:
2006-11-28 14:18 tgl
* src/include/catalog/: pg_proc.h (REL7_3_STABLE), pg_proc.h(REL7_4_STABLE), pg_proc.h (REL8_1_STABLE),
pg_proc.h(REL8_0_STABLE):Mark to_number() and the numeric-type variants ofto_char() as stable, not immutable, because
theirresults depend onlc_numeric; this is a longstanding oversight. We cannot forceinitdb for this in the back
branches,but we can at least providecorrect catalog entries for future installations.
2006-11-28 14:18 tgl
* src/include/catalog/pg_proc.h: Mark to_char(timestamp withouttimezone) as stable, not immutable, since its result now
dependsonthe lc_messages setting, as noted by Bruce. Also, mark to_number()and the numeric-type variants of to_char()
asstable, because theirresults depend on lc_numeric; this is a longstanding oversight. Also, mark to_date() and
to_char(interval)as stable; althoughthese appear not to depend on any GUC variables as of CVS HEAD,that seems a
propertyunlikely to survive future improvements. Itseems best to mark all the formatting functions stable and be
donewithit. catversion not bumped, because this does not seemcritical enough to force a post-RC1 initdb, and anyway we
cannotdoso in the back branches.
regards, tom lane