Re: ghc will ever reach testing?
On Tue, Jul 16, 2013 at 09:35:24AM +0100, Gianfranco Costamagna wrote:
> > Da: Joachim Breitner <nomeata@debian.org>
> >we are well aware of the problem, see
> >http://lists.debian.org/debian-haskell/2013/07/msg00019.html
> >and http://ghc.haskell.org/trac/ghc/ticket/7993
> >
> >Unfortunately, debugging the RTS of an advanced compiler on a strange
> >architecture is not something that is easily done, and everyone has also
> >other things to do.
> >
> >So unless someone comes up with a fix soon, we will have to „fix“ it by
> >removing Haskell from s390x.
>
> I already read before asking this bug and message...
>
> Quoting from the bug
> "sometimes hangs at child process exit on s390x"
>
> So why not just issue rebuilds until it succeedes?
No, that's hopeless. When I say "sometimes", I mean that each ghc
invocation has some probability of hanging; but attempting to build
something large like ghc itself runs ghc so many times that the
cumulative probability of a hang somewhere is near enough certain.
I think we might as well remove the stack on s390x at this point. Given
the nature of the bug, we're going to have to rebootstrap ghc/s390x
anyway in order to recover if and when somebody fixes it, and it isn't
really all that much harder to rebootstrap the whole stack.
--
Colin Watson [cjwatson@debian.org]
Reply to: