On Thu, Oct 26, 2017 at 10:01 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Laurenz Albe <laurenz.albe@cybertec.at> writes:
>> Also, to have PostgreSQL inline the function, which would be good
>> for performance, it should be declared IMMUTABLE.
>
> Actually, if you hope to have a SQL function be inlined, it's better
> not to decorate it at all --- not with IMMUTABLE, and not with STRICT
> either. Both of those restrict the parser's ability to inline unless
> it can prove the contained expression is equally immutable/strict.
> With the default attributes of volatile/not strict, there's nothing
> to prove.
This is extremely obnoxious. Is it possible to raise a warning on
function creation?
> (In any case, it's usually easy enough to tell from EXPLAIN output
> whether inlining has happened.)
No it isn't. The explain syntax is arcane and inlining as a general
concept is only very indirectly expressed. I really think we ought to
do better here; I was not able to find any treatment of inlining given
in the 'Performance Tips' or the 'Functions and Operators' section, or
anywhere really (except the wiki). This is really a disservice to the
users, I think.
merlin
--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
