Hi, Am Dienstag, den 05.04.2011, 22:54 +0530 schrieb Joachim Breitner: > Am Dienstag, den 05.04.2011, 21:01 +0530 schrieb Joachim Breitner: > > we are not uploading 7.0.3 yet > > looks like we are after all. > > It turns out that the ABI hash of the packages created by ghc changes > when we upload a new version. The reason can be found on > http://hackage.haskell.org/trac/ghc/ticket/4012#comment:1 > > So if we upload a new ghc revision now, we will have to rebuild all > libraries that have been uploaded so far anyways (thanks to > auto-signing, this should be faster than ever). So the only disadvantage > of ghc-7.0.3 is that it is not part of the platform. But if you look at > the release notes¹, you can see that there is nothing we do not want to > offer our users, and bad floating point results sound like something we > want to fix. So let’s invoke “we only follow the platform until we have > a reason not to” and upload 7.0.3. Any objections? > > Now for 7.0.3 we have three options to get a stable ABI: > 1. Upload 7.0.3-1, wait until it built on all arches, upload again. > 2. Upload 7.0.3-1, for each arch: Wait until it is built, schedule > binNMU. > 3. Remove cBooterVersion. > > I kinda like options 3, but there might be more causes for ABI changes > and it is an unnecessary diversion from upstream. I guess 2 is good. > > I’ll not do that today, though. ghc-7.0.3 does not have this problem, i.e. the base hash stays the same, regardless of whether 7.0.3 is built aginst 7.0.2 or 7.0.3. Therefore, I just uploaded ghc-7.0.3 and did not bother to rebuild it again. binNMUs for all libraries built against 7.0.2 are scheduled. i386 is pretty quick in building them, thanks to auto-signing. Hmm, that’s all for this announcement :-) 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: This is a digitally signed message part