On Wed, Feb 8, 2012 at 06:21, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Marti Raudsepp <marti@juffo.org> writes:
>> Case #1 uses the normal textcat(text, text) operator by automatically
>> coercing 'x' as text.
>> However, case #2 uses the anytextcat(anynonarray, text), which is
>> marked as volatile thus acts as an optimization barrier.
>
> Hmm ... since those operators were invented (in 8.3), we have adopted a
> policy that I/O functions are presumed to be no worse than stable:
> http://archives.postgresql.org/pgsql-committers/2010-07/msg00307.php
> ISTM that would justify relabeling anytextcat/textanycat as stable,
> which should fix this.
Yes, we should definitely take advantage of that.
I scanned through all of pg_proc, there are 4 functions like this that
can be changed: textanycat, anytextcat, quote_literal and
quote_nullable. All of these have SQL wrappers to cast their argument
to ::text.
quote_literal | select pg_catalog.quote_literal($1::pg_catalog.text)
quote_nullable | select pg_catalog.quote_nullable($1::pg_catalog.text)
textanycat | select $1 || $2::pg_catalog.text
anytextcat | select $1::pg_catalog.text || $2
Patch attached (in git am format). Passes all regression tests (except
'json' which fails on my machine even on git master).
No documentation changes necessary AFAICT.
Regards,
Marti