2008/8/15 Tom Lane <tgl@sss.pgh.pa.us>:
> Hannu Krosing <hannu@2ndQuadrant.com> writes:
>> How is this supposed to interact with argument names ?
>
> Yeah, the real problem with this proposal is that it conscripts a syntax
> that we'll probably want to use in the future for argument-name-based
> parameter matching. The proposed behavior is not nearly as useful as
> that would be.
It isn't. As Hannu showed these features should live in harmony (if we
will accept Oracle's syntax). I see as real problem new column in
pg_proc, that allow quickly chose between labeled and non labeled
functions - and then collect labels in parse time. This is way for
adding parameter info into variadic function - it's main goal of this
proposal. Without column in pg_proc it could same slowdowns like first
variadic function's implementation.
Regards
Pavel Stehule
>
> regards, tom lane
>
