Final word on OpenAL?

Duncan Coutts duncan.coutts at worc.ox.ac.uk
Wed Apr 29 05:46:27 EDT 2009


On Tue, 2009-04-28 at 18:25 +0200, Sven Panne wrote:
> Am Dienstag, 28. April 2009 03:24:23 schrieb Mikhail Glushenkov:
> > what is the final word regarding inclusion of OpenAL? I for one
> > would vote against including it, since it is not essential IMO
> > and can be always added later. [...]
> 
> I've just uploaded new versions of the OpenGL/GLUT/OpenAL/ALUT packages to 
> Hackage. The fix for the wrong calling convention for OpenAL is included, so 
> perhaps you can give it a try. But if there is not enough time, I agree with 
> you that the inclusion of OpenAL is not essential (although it would be nice 
> :-).
> 
> OTOH, I strongly propose to use the new OpenGL/GLUT packages, they contain 
> tons of bug fixes and a few additions which I've been asked for several times.

Sven, I don't want to sound like I'm picking on your packages but I
think we should stick to the current versions for the first release.
It's not about the merits of your recent fixes. Here's the rationale:

The platform is supposed to be doing time-based release not feature
ones. Now obviously for the first release we've not exactly been doing
that but we should be trying to. We need to get away from the idea that
releases should be delayed because of some cool new feature in one or
two components. The release cycles allow for fixes and features to get
into subsequent releases and there should not be too long between
releases. If we can actually get to the stage where we set a release
date many months in advance then there are great coordination advantages
as the GNOME project has demonstrated. It means we don't constantly need
to communicate between a few dozen maintainers and release managers on
when the releases will be. We just have a fixed date and a schedule for
when packages have to be on Hackage prior to a release. If packages are
not ready then it's not problem. Fixes can be rolled into the next minor
release and new features rolled into the next major release. This will
become all the more important as the platform gets bigger in terms of
numbers of packages and numbers of maintainers.

So I suggest we stick to the existing package list for the 2009.2.0
release and that we include your updated OpenGL/GLUT versions for
2009.2.1 in 4-6 weeks time. The OpenAL/ALUT packages can be proposed for
the next major release in ~6 months time (once the procedures to do that
are established by the folks on the libraries mailing list).

I hope that sounds reasonable. We don't want to discourage you or other
maintainers from polishing your packages, on the contrary. But we also
have to think about the kind of release process that we want and from
the experience of other projects it seems that time-based is the best
approach for low-stress releases (rather than making nail-biting
last-minute decisions about packages going in or out, or delaying
releases).

Duncan




More information about the Haskell-platform mailing list