ghc-7.6.2 breaks haddock interface...

Joachim Breitner nomeata at
Wed Dec 12 09:40:42 CET 2012

Dear Ian,

Am Mittwoch, den 12.12.2012, 00:52 +0000 schrieb Ian Lynagh:
> Won't you have to rebuild everything anyway, due to the GHC version
> number in the .hi files changing?

good point; let me add a bit of information about the Debian
infrastructure: There are two ways of rebuilding stuff. Regular uploads
and binNMUs.

Regular uploads require manual work: Adding a changelog entry to the
source file, building it locally (which is required and means that the
developer has to make sure he builds everything in the right order, and
that he has to install previously built packages), signing the package
and then uploading the new source package together with the built
package. Tedious work.

binNMUs can be scheduled programatically and cause the Debian
autobuilder infrastructure to built the package again on the selected
architectures (possibily all). The infrastructure is smart enough to
figure out the build order itself and no further interaction is
required. Nice.

Unfortunately, the latter can only rebuild arch-dependent packages,
which in the context of Debian packages means -dev and -prof.
Architecture independent packages (here: -doc) cannot be built this way
and thus can only be rebuilt by full sourceful uploads.

The .hi files reside in the arch-dependent packages, while the .haddock
files don’t.

I know that you might think now: Well, then fix your infrastructure. A
valid point. But I am not in the position to do so, hence my hope for
only as many haddock interface number bumps as required. (The Debian
side thinks things about Haskell like: Well, get your ABIs stable so
that we can bugfix individual packages without breaking other


Joachim "nomeata" Breitner
Debian Developer
  nomeata at | ICQ# 74513189 | GPG-Keyid: 4743206C
  JID: nomeata at |
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <>

More information about the Glasgow-haskell-users mailing list