Re: Haskell migration exp to unstable
On Fri, May 24, 2013 at 12:13:02PM +0100, Colin Watson wrote:
> On Fri, May 24, 2013 at 12:54:56PM +0200, Joachim Breitner wrote:
> > Am Freitag, den 24.05.2013, 10:56 +0100 schrieb Colin Watson:
> > > I don't suppose you'd consider disabling GHCi on ARM while you're here?
> > > If not, I suspect you're going to find you need to disable it later to
> > > get things that use doctests into testing, if you recall our earlier
> > > discussion.
> >
> > hmm, unfortunate timing, I got the mail right after uploading ghc to
> > unstable.
> >
> > Let’s sort this out separately; if we want to get stuff into testing
> > before we can still simply remove affected packages on arm (which we
> > probably have to do anyways).
>
> OK. It'll be a bunch of stuff up to and including git-annex (the whole
> persistent/yesod stack is affected), which was why I veered away from
> doing that on Ubuntu. Actually disabling GHCi only changes the
> libghc-ghc-dev ABI, so it's not utterly terrible to do later.
I can indeed confirm that we're starting to see failures due to broken
GHCi, for example:
https://buildd.debian.org/status/fetch.php?pkg=haskell-data-accessor-template&arch=armel&ver=0.2.1.10-3&stamp=1369656600
https://buildd.debian.org/status/fetch.php?pkg=haskell-distributive&arch=armhf&ver=0.3-2&stamp=1369561716
Note that some of the affected packages would succeed if GHCi were
disabled (because doctest detects that condition and skips), so it's not
quite true that we'd just have to remove all the affected packages
anyway.
--
Colin Watson [cjwatson@debian.org]
Reply to: