"Pavel Stehule" <pavel.stehule@gmail.com> writes:
> who write this patch?
Well, like I said, I'm willing to adjust the patch to whatever syntax
we come up with.
After sleeping on it I'm a bit less excited about using the SQL/PSM
SIGNAL syntax; the reason being that if we use that, and then sometime
in the future we read the spec more closely and find out that it demands
different behavior than RAISE has, we'd have a compatibility problem.
Inventing PG-only additions to RAISE doesn't carry that risk.
So right now I'm thinking I like my original proposal
http://archives.postgresql.org/pgsql-hackers/2008-05/msg00357.php
with the exception that we should go withSQLSTATE 'xyzzy'
as the syntax in EXCEPTION lists. Also I'm willing to go with
ERRCODE rather than CODE as the name of the USING option, since
Pavel didn't like CODE. (I don't want to use SQLSTATE for it,
because with this syntax it's pretty clear that SQLSTATE means
one of the 5-letter codes, *not* a condition name.)
regards, tom lane