Hmm, none of the ghc-* packages should be emitting debuginfo packages:

# ghc does not emit debug information
%global debug_package %{nil}

(In reply to comment #1)
> Hmm, none of the ghc-* packages should be emitting debuginfo packages:
> 
> # ghc does not emit debug information
> %global debug_package %{nil}  

Sorry to be blunt, but would you mind actually checking things before closing bugs?  All of the packages mentioned in the initial comment *do* emit debuginfo packages.  They have the line you cited commented out.

http://cvs.fedoraproject.org/viewvc/devel/ghc-GLUT/ghc-GLUT.spec?revision=1.6&view=markup
http://cvs.fedoraproject.org/viewvc/devel/ghc-OpenGL/ghc-OpenGL.spec?revision=1.6&view=markup
http://cvs.fedoraproject.org/viewvc/devel/ghc-tar/ghc-tar.spec?revision=1.4&view=markup
http://cvs.fedoraproject.org/viewvc/devel/ghc-zlib/ghc-zlib.spec?revision=1.18&view=markup

It's perfectly fine to be blunt, but I did actually check, but I looked at the F-12 branch rather than the devel branch.  It might be a good policy not to make accusations about what someone did or didn't do when you don't actually know. That could easily be misinterpreted as rudeness, which would not be so perfectly fine.

Jens, what's the story with the debuginfo generation being re-enabled?

This bug is reported against rawhide, the initial comment specifically mentions a NEVR containing f13, there's no way I could tell that you checked F-12, and I think it's quite fair to assume that if rawhide had been actually checked, this bug would have not been closed as NOTABUG along with the text in comment 1.  But nevermind, mistakes do happen, I'll shut up now and let you find out the cause and what to do about it.

That's me sorry.

I reenabled debuginfo in rawhide for a few libraries as an experiment
since it strips the new shared libraries, but if that is stepping on
sanity script's I better do it right I guess.

bos: BTW feel free to reassign bugs relating to my changes to me. :)

eg Compare 
http://koji.fedoraproject.org/koji/fileinfo?rpmID=1732945&filename=/usr/lib/ghc-6.12.1/zlib-0.5.0.0/libHSzlib-0.5.0.0-ghc6.12.1.so (118745 bytes) vs http://koji.fedoraproject.org/koji/fileinfo?rpmID=1743252&filename=/usr/lib/ghc-6.12.1/zlib-0.5.0.0/libHSzlib-0.5.0.0-ghc6.12.1.so (78372 bytes)

So creating debuginfo strips the .so more than brp-strip?

Thanks Ville for catching and reporting it.

I reverted my little experiment in the latest builds.