On Thu, Jan 13, 2011 at 21:24, Peter Simons
Hi guys,
the AUR package haskell-haskore [1] was marked out-of-date by palmfron. Now, technically that package is not "out-of-date", because the version we distribute on AUR -- 0.1.0.4 -- is the latest one. Still, there is something wrong with that package: it doesn't compile [2]. As far as I can tell, that PKGBUILD is broken because its dependencies cannot be satisfied. That is a bug, and this problem actually feels *worse* than the package just being outdated. AUR does not offer a way to mark packages as "broken", so in my humble opinion it's quite alright to use the out-of-date flag for that purpose instead, because the PKGBUILD clearly needs updating.
Personally, I would like to keep the out-of-date marker on that package set, and be it for no other purpose than as a warning to AUR users who try to install it. I am interested in other people's opinion on this matter. Does anyone feel compelled to agree or disagree with this approach? And if so, for what reasons?
True, the package needs updating, but I don't think it's a good idea to expand the meaning of 'out-of-date' to also include 'broken'. It's not that difficult to find packages on AUR that were broken and then fixed by discussion in the comments, and at no point were those packages marked 'out-of-date' because of their brokenness. In fact this situation is one of the reasons why the issue tracker is enabled on github.com/archhaskell/habs. So, I'd argue it shouldn't be set 'out-of-date', because it isn't, instead I'll create an issue on the HABS project and add a link from AUR.
Furthermore, I wonder whether anyone would like to volunteer to fix the haskell-haskore package? I haven't looked closely at the problem, but it seems to me that the build failure can be remedied by any of the following means:
(1) Contact haskore's author and convince him to release an update that lifts the haskell-event-list<0.1 and haskell-non-negative<0.1 restrictions.
(2) Patch the Cabal file and the PKGBUILD to lift those restrictions.
I will attempt to see if (2) would fix it, and then proceed with (1).
(3) Upload older version of those packages to AUR using different names, and change the haskore package to depend on the older versions instead.
Would anyone of the Arch-Haskell team like to take care of that? Otherwise, I feel that we should offer palmfron to orphan the package so that he can take over maintenance.
That is one option, but let's save that for later. /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus