Dimitri Fontaine <dimitri@2ndQuadrant.fr> writes:
> Tom Lane <tgl@sss.pgh.pa.us> writes:
>> Has anybody stopped to look at the SQL standard for this? In-line
>> trigger definitions are actually what they intend, IIRC.
> In which language? Do we need to include PL/PSM to be compliant, and
> use that by default?
Darn if I know. But let's make sure we don't paint ourselves into a
corner such that we couldn't support the standard's syntax sometime
in the future.
> In that case we might want to force people to
> spell out LANGUAGE plpgsql when we don't provide for PSM yet, so that we
> avoid some backwards compatibility problems down the road.
I suspect that we can avoid that as long as the command is based around
a string literal for the function body. OTOH, CREATE FUNCTION has never
had a default for LANGUAGE, and we don't get many complaints about that,
so maybe insisting that LANGUAGE be supplied for an in-line trigger
isn't unreasonable.
regards, tom lane