Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> Since we ran out of time/ideas on this, I would propose just committing
> the part that breaks ties based on the number of arguments, which
> already solves a large part of the problem (at least in a pre-default
> values world) and would very likely be a part of any possible future
> utterly complete solution.

Seems safe enough.

			regards, tom lane

In response to

• Re: pg_dump sort order for functions at 2010-02-15 14:48:11 from Peter Eisentraut

pgsql-hackers by date

Next:	From: Euler Taveira de Oliveira	Date: 2010-02-15 15:15:24
	Subject: Re: TCP keepalive support for libpq
Previous:	From: Robert Haas	Date: 2010-02-15 15:10:57
	Subject: Re: plperl message style on newly added messages