[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#570594: Haskell binNMUs



Hi,

Am Samstag, den 20.02.2010, 11:19 +0100 schrieb Marc 'HE' Brockschmidt:
> > Such a change is expected to happen very rarely, but we want to be
> > prepared. Therefore, from ghc6-6.12.1-10 on, ghc6 also provides virtual
> > packages corresponding to the ABIs contained in them. We need all
> > haskell library packages to be rebuild with ghc6-6.12.1-10.
> 
> Great. Is this scheme now allowing ghc transition without source
> uploads?

We already had such a scheme since about half a year or a year, and it
would work for our transition as well. Dependencies were tied to the
upstream version of a package, which usually works. But in some
circumstances, the ABI does break without a change of the upstream
version (e.g. a rebuild against a updated dependency), so this system
was still lacking.

But since ghc6-6.12.1, the compiler explicitly gives ABI hashes for each
package and each package dependencies. We transform these into a set of
Provides: and Depends: that guarantee that only working packages are
installed. But to implement this, we had to add the new
${haskell:Provides} substvar to the control files, therefore the round
of sourceful uploads this time. With the next compiler or library
upgrade, this can be handled by binNMUs and without manual Dep-Wait
queuing (thanks to the wanna-build support).

> > It does not give back any packages state Build-Attempted, though. I’ll
> > go through that list tomorrow...
> 
> Happy to hear that, will schedule the needed give-backs too :-)

I’m wondering a bit how to get the Dep-Waits correctly... I’ll report
back in a while.

Greetings,
Joachim

-- 
Joachim "nomeata" Breitner
Debian Developer
  nomeata@debian.org | ICQ# 74513189 | GPG-Keyid: 4743206C
  JID: nomeata@joachim-breitner.de | http://people.debian.org/~nomeata

Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil


Reply to: