Re: Drop all overloads of a function without knowing parameter types

Sergey Konoplev <gray(dot)ru(at)gmail(dot)com> writes:
> On Tue, Feb 4, 2014 at 8:35 AM, Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> wrote:
>> Sergey Konoplev escribi:
>>> On Mon, Feb 3, 2014 at 10:09 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>>> People periodically ask for extensions flavored more or less like this,
>>>> but I'm suspicious of building any such thing into the core. There's too
>>>> little commonality in the exact conditions they want to search on.
>>>> Leaving it at the level of a scripting problem, as above, allows arbitrary
>>>> customization of the search condition.

>>> I understand the POV of both Evan and you here. However, I think that
>>> there might be a good solution for this particular case - to allow
>>> dropping functions by name only if it has the only signature, but if
>>> there are 2 or more signatures then print an error specifying all the
>>> forms of the function, eg.:
>>>
>>> ERROR: Can not drop function 'foo' because it has more then one
>>> signature: foo(integer), foo(text).

>> But that doesn't solve Evan's request. He would want both functions
>> gone, not an error.

> I was writing about some kind of a compromise.

My point was precisely that a compromise would satisfy nobody. There
would be a few cases for which it was Exactly The Right Thing, and many
more for which you'd still need to learn how to do the EXECUTE trick.

I wonder whether we shouldn't address this by adding a few examples
of that type of trick to the docs. Not sure where, though ...

regards, tom lane