[Debian GNUstep maintainers] Bug#885786: gnustep-back: Please drop art backend package

[Yavor Doganov]

Jeremy Bicha wrote:
> On Fri, Dec 29, 2017 at 7:59 PM, Yavor Doganov <yavor at gnu.org> wrote:
> >   3) Take libart under our umbrella.  (Provided it's a viable option
> >      in the first place.)
> 
> As you can see, libart has had zero maintenance since 2010:

It doesn't bother me much.  Almost half of the GNUstep stack (non-core
packages and libraries) is unmaintained, some of them since 2001.
We're used to it.  As long as we can cope with security and other
important bugs, it is manageable.  But it's a last resort.

> But if we can remove libgnome, then it probably makes sense to
> remove libgnomecanvas too and if we can remove that, then this is
> one of the only packages left using libart.

Right, that's ironclad logic and I don't dispute your effort to get
rid of libraries that you have to maintain but have no future and/or
are abandoned upstream.  I'm just sharing our concerns.

> https://bugs.debian.org/885800 (and dia is unmaintained).

Dia was a flagman package once upon a time, I'm surprised to see it
unmaintained.  I'll take a look to see if I can help with this bug or
at least get some QA job done.  It's a real pity.

Ivan Vučica wrote:
> > On 30 Dec 2017, at 00:59, Yavor Doganov <yavor at gnu.org> wrote:
> > 
> >  1) Wait for the opal or wayland backends to be labelled "ready for
> >     release".  We can drop art immediately then.
> 
> It would be nice if Opal (Core Graphics) would be packaged anyway (a
> requirement for the -back opal). I’m not aware of apps that require
> it, but it would not be terrible to have a dev package either.

Some time ago I filed an ITP bug for CoreBase but discovered some
ABI-related issues and couldn't upload the package back then.
Afterwards I went MIA and probably haven't even reported it upstream.
It has always been my intention to package Opal.

 > Wayland backend has not been merged at this time, and it’d be a
 > “windowing” backend, not a “drawing” backend anyway. That is: I
 > think it still uses Cairo.

Right.  It's still useful to have it as a separate backend.  There's a
truckload of windowing and rendering issues, different behavior,
different code paths and different font choices between the backends.