Re: Passing CopyMultiInsertInfo structure to CopyMultiInsertInfoNextFreeSlot()

On 2019-May-14, Michael Paquier wrote:

> On Tue, May 14, 2019 at 01:19:30PM +1200, David Rowley wrote:
> > When I wrote the code I admit that I was probably wearing my
> > object-orientated programming hat. I had in mind that the whole
> > function series would have made a good class. Passing the
> > CopyMultiInsertInfo was sort of the non-OOP equivalent to having
> > this/Me/self available, as it would be for any instance method of a
> > class. Back to reality, this isn't OOP, so I was wearing the wrong
> > hat. I think the unused parameter should likely be removed. It's
> > probably not doing a great deal of harm since the function is static
> > inline and the compiler should be producing any code for the unused
> > param, but for the sake of preventing confusion, it should be removed.
> > Ashutosh had to ask about it, so it wasn't immediately clear what the
> > purpose of it was. Since there's none, be gone with it, I say.
>
> Sounds fair to me. This has been introduced by 86b8504, so let's see
> what's Andres take.

If this were up to me, I'd leave the function signature alone, and just add
(void) miinfo; /* unused parameter */
to the function code. It seems perfectly reasonable to have that
function argument, and a little weird not to have it.

--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services