On Wed, 9 Jan 2019 at 13:36, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> David Rowley <david.rowley@2ndquadrant.com> writes:
> > Why can't we just remove the !noError check in the location where the
> > error is raised?
>
> I don't like that a bit --- the point of noError is to prevent throwing
> errors, and it doesn't seem like it should be LookupFuncName's business
> to decide it's smarter than its callers. Maybe we need another flag
> argument?
Well, I guess you didn't have backpatching this in mind. The reason I
thought it was okay to hijack that flag was that the ambiguous error
was only raised when the function parameters were not defined. I
chased around and came to the conclusion this only happened during
DROP. Maybe that's a big assumption as it certainly might not help
future callers passing nargs as -1.
I've attached another version with a newly added flag.
--
David Rowley http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services