Re: PQconninfo function for libpq

On Nov 28, 2012 1:54 AM, "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> writes:
> > Peter Eisentraut wrote:
> >> There is already the PQconninfoOption.dispchar field for this purpose.
>
> > I had the impression that that field would go away with this patch, but
> > then again it might not be worth the compatibility break. I find the
> > dispchar thingy somewhat unsightly.
>
> It is that, without a doubt, but that API has been that way longer than
> any of us have been working on the code. I'm not excited about changing
> it just to have an allegedly-cleaner API. And we cannot have the field
> simply "go away", because it's been exposed to applications for lo these
> many years, and surely some of them are using it. So in practice we'd
> be maintaining both the old API and the new one.
>
> I think we should leave it as-is until there are more reasons to change
> it than seem to be provided in this thread.

I think removing that would be a really bad idea. I'm not sure anybody is
actually relying on it, but it would also change the size of the struct and
thus break things for anybody using those functions.

If people prefer we remove the password classifier for the new function
since it at least partially duplicates that field we can certainly do that,
but I think leaving it in allows those who write new code to use a slightly
neater api, at pretty much no cost in maintenance for us.

/Magnus