[Pkg-haskell-maintainers] Bug#600691: libghc6-texmath-dev: should link against libghc6-parsec3-dev

[John MacFarlane]

+++ Jonas Smedegaard [Oct 19 10 12:44 ]:
> On Tue, Oct 19, 2010 at 12:29:47PM +0200, Joachim Breitner wrote:
> >Am Dienstag, den 19.10.2010, 11:03 +0200 schrieb Jonas Smedegaard:
> >>Package: libghc6-texmath-dev
> >>Version: 0.3.0.2-2+b1
> >>Severity: important
> >>
> >>Hi,
> >>
> >>Currently texmath links against parsec2, and highlighting-kate
> >>against parsec3.
> >>
> >>Recent Pandoc needs both and fails to build when linked against
> >>different versions of parsec, so the wonderful new Pandoc cannot
> >>be packaged for Debian :-(
> >>
> >>Simply bumping libghc6-texmath build-dependency on
> >>libghc6-parsec3-dev and rebuilding makes the Pandoc build
> >>succeed.
> >
> >I thought this would be no problem as long as pandoc does not use
> >parsec-types from both texmath and highlighting-kate. Does it do
> >that? Did you test building pandoc with cabal (which is more picky
> >about such issues) or regular Setup.hs?
> 
> Sorry - I don't know the answers to above questions.  Cc'ing John,
> who should know.

I don't know the answer to the question about building. I think a standard
debian-haskell build script is being used, and I assume it uses Setup.hs
directly instead of cabal. But I'm not positive.

Pandoc does use a parser defined in texmath.  So if texmath links
against parsec2, and the rest of pandoc links against parsec3 (as
it currently does in the debian package), there might be a problem
there.

Jonas, have you tried changing the pandoc debian/rules to link against
parsec2?  Pandoc 1.6 should work with either version of parsec.

It seems to me that it would be best if all of these inter-operating
packages linked against the same version of parsec.  All three of
them will currently work with either version.  Would the maintainer
of texmath consider bumping the build dependency to parsec3?

> >Do you plan to upload pandoc to unstable or experimental? Is it
> >targeted to squeeze?
> >
> >(But no objections to doing the bump, the question is whether we
> >want to do it now in unstable, now in experimental, or after the
> >release in unstable)
> 
> It is a new upstream release, so unless John can help raise concern
> about security bugs in older releases, I believe it is too late in
> the game for Squeeze.

Unfortunately, I don't see any security fixes in the changelog.

John