[MPlayer-dev-eng] [PATCH] Add giflib 5 support

[Reimar Döffinger]

On Sat, May 04, 2013 at 08:47:45PM +0100, Andïï wrote:
> On May 4, 2013 6:52 PM, "Reimar Döffinger" <Reimar.Doeffinger at gmx.de> wrote:
> >
> > On 04.05.2013, at 19:21, Andïï <gnu.andrew.rocks at gmail.com> wrote:
> > > On May 4, 2013 7:34 AM, "Reimar Döffinger" <Reimar.Doeffinger at gmx.de>
> wrote:
> > >>
> > >> On 02.05.2013, at 11:50, Andrew John Hughes <gnu_andrew at member.fsf.org>
> > > wrote:
> > >>
> > >>> Trying again...
> > >
> http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/2013-April/071595.html
> > >>>
> > >>> The attached patch allows mplayer to be built with giflib 5.  Tested
> > >>> against 1.1 and SVN.
> > >>
> > >> Sorry, I forgot to ask a few things.
> > >> I was wondering if there is much "need" for it, since FFmpeg does
> support
> > > gif quite fully I think. Basically I am curious why you care :-)
> > >
> > > If giflib has been upgraded to version 5, mplayer won't even detect the
> > > presence of giflib and support won't be compiled. The patch fixes
> configure
> > > to detect version 5 too, and the consequent build errors from code.
> >
> > So far it's obvious, but who cares if giflib is enabled or not.
> >
> 
> People who use it?  I assume there's some userbase for it or you'd have
> just removed it from the codebase altogether.

We very rarely remove feature if we do not have to.
In the case of the decoder it's main purpose should now be as an alternative/reference
for the FFmpeg implementation.
Nothing non-developers should care much about.
I wanted to know if you are using giflib in MPlayer or know someone who
does and would disagree with this characterization.

> > >> This is related to whether there is much of a point in keeping support
> > > for the old version.
> > >
> > > Support for <5 needs to be kept for those who haven't yet upgraded.
> >
> > I see no obvious "need", we can just drop support for it.
> > I wanted to know what reasons there are to continue to support versions <
> 4.
> > Since I consider our giflib support fairly pointless (less relevant than
> aalib even for example) there should be a fairly good reason to add to the
> maintenance burden by keeping support for old versions instead of telling
> users "upgrade if you really need it".
> 
> Fair enough.  The patch was originally written for the existing 1.1 so
> there are different considerations than if you're doing a new release where
> you can say "giflib needs version 5 or later".

We very rarely do updates to existing releases, and anything beyond
security issues is out of scope for it.

> It would certainly simplify the patch.

I think that would be reasonable, even though it is a bit unfortunate
that Debian and Ubuntu seem to be stuck at giflib 4.
If it causes lots of complaints it's still possible to add compatibility to v4
back at that point.