Re: rpath issue

[Sven Bartscher <sven.bartscher@weltraumschlangen.de>]

On Sun, 23 Aug 2015 19:43:25 +0200
Joachim Breitner <nomeata@debian.org> wrote:

> Hi,
> 
> Am Sonntag, den 23.08.2015, 16:42 +0200 schrieb Sven Bartscher:
> > We recently got a lot of rejects because of rpath lintian warnings.
> > After looking up what those mean, I'm confused why our packages set
> > the rpath. I would expect them not to bother with the rpath at all,
> > since they're statically linked.
> > Can anyone explain, why we have that issue?
> 
> it’s simply a bug in Cabal:
> https://github.com/haskell/cabal/issues/2625
> https://github.com/haskell/cabal/pull/2626
> 
> It is fixed upstream, but it takes a while to make it into a GHC
> release. We could try to backport that change to the Cabal in
> GHC-7.10, or simply try to include the latest Cabal in GHC-7.10, but
> previously, I was not comfortable patching GHC to such an extend.
> Now, with the ability to run "make-all" to verify the changes, such
> fixes are actually reasonable – but still quite a bit of work and
> waiting time.

Instead of fixing GHC (and thus doing work that was already done by
upstream), we could patch haskell-devscripts, to use chrpath, to remove
the rpath from the binaries.

Regards
Sven