Re: Clobbered parameter names via DECLARE in PL/PgSQL
Pavel Stehule <pavel.stehule@gmail.com>
From: Pavel Stehule <pavel.stehule@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Brendan Jurd <direvus@gmail.com>, pgsql-hackers@postgresql.org
Date: 2012-04-17T09:50:55Z
Lists: pgsql-hackers
Attachments
- plpgsql_check_function-2012-04-17-1.patch.gz (application/x-gzip) patch 2012
Hello there is VIP patch of plpgsql_check_function that supports this warning Regards Pavel 2012/4/15 Pavel Stehule <pavel.stehule@gmail.com>: > 2012/4/15 Tom Lane <tgl@sss.pgh.pa.us>: >> Pavel Stehule <pavel.stehule@gmail.com> writes: >>> We can raise warning from CREATE OR REPLACE FUNCTION - but I would to >>> like have plpgsql_check_function inside core - and it is better place >>> for this and similar issues. >> >> I agree. This is a perfectly legal use of nested declaration scopes, >> so it would be totally inappropriate to complain about it in normal >> use of a plpgsql function. On the other hand, it would probably be >> sane and useful for CHECK FUNCTION to flag any case where an inner >> declaration shadows an outer-scope name (not only the specific case >> of topmost block vs function parameter). > > yes, it is very simple check there. There should be "levels" of > warnings in future and performance or semantic warnings. > > But, we don't need to increase complexity of CHECK FUNCTION now. A > design of CHECK FUNCTION was rich for this purposes. And we need to > find way to push plpgsql_check_function to core first. > > Regards > > Pavel > > > > > >> >> regards, tom lane