On Mon, Apr 18, 2011 at 6:21 PM, Jens Petersen wrote:

On 7 April 2011 06:51, Don Stewart wrote:

...

Moving those modules under OpenGL.*.*.StateVar etc would remove the problems with upgrading to the HP.

Perhaps it would be a good time now to revisit again whether we really need to include OpenGL support inside HP or not.

I grant that it may be convenient for users who might find it hard to build OpenGL and GLUT with cabal-install; on the other hand not that many haskell packages actually need those libraries AFAICT so I am still not convinced that they really belong in HP.

Let's look at some numbers: http://www.galois.com/~dons/hackage/march-2011/mar-10-to-feb-11.txt http://www.galois.com/~dons/hackage/march-2011/mar-10-to-feb-11.txtSearch for opengl and you'll see that it's in the top 25, meaning it's used by a relatively high number of packages. That means we're discussing a library that provides a unique capability (opengl support) that is in the top 25 of the Haskell libraries. Not making it easy to get seems like a bit of a strategic error to me. The reverse package index shows what depends on it: http://bifunctor.homelinux.net/~roel/cgi-bin/hackage-scripts/revdeps/OpenGL-... And some things depend on the raw bindings instead: http://bifunctor.homelinux.net/~roel/cgi-bin/hackage-scripts/revdeps/OpenGLR... Additionally, I've had 1 person contact me off list saying that they wouldn't even be a Haskell programmer if the opengl bindings were not in the HP. If there had been any barrier to getting good opengl support when they started learning Haskell that would have been a show stopper for them. Jason