[aur-general] pandoc moved back to the AUR for the time being

Daniel Micay danielmicay at gmail.com
Thu Sep 19 15:45:44 EDT 2013


On Thu, Sep 19, 2013 at 12:34 PM, Karol Blazewicz <karol.blazewicz at gmail.com
> wrote:

> On Thu, Sep 19, 2013 at 1:11 PM, Rashif Ray Rahman <schiv at archlinux.org>
> wrote:
> > On 19 September 2013 05:40, Daniel Micay <danielmicay at gmail.com> wrote:
> >> I've moved pandoc back to the AUR because the latest version picked up
> too
> >> many new dependencies. It was convenient to have it in [community], but
> >> it's much saner to maintain these with the automation used by the
> >> arch-haskell project.
> >>
> >> I only use it for Rust's documentation generator, and don't want to be
> >> frequently rebuilding 30+ libraries for a language I don't use.
> >>
> >> Our packaging standards don't make much sense for Haskell because there
> are
> >> little gains from dynamic linking. There's not nearly enough sharing of
> >> code for it to come close to being efficient, and no stable ABI either.
> >
> > That's a shame, I am really dependent on it these days. But thanks for
> > maintaining it until now :)
> >
> >
> > --
> > GPG/PGP ID: C0711BF1
>
> [haskell] unofficial repo has haskell-pandoc 1.11.1-16.
>
> The unmaintained packages uploaded by arch-haskell guys to the AUR
> have been removed a couple months ago
> https://mailman.archlinux.org/pipermail/aur-general/2013-June/023951.html
>
> Do you want to keep the haskell packages you dropped to the AUR in the
> AUR because it makes it easier to move them to the official repos some
> time down the road or ...?
> Maybe the unofficial repo is enough and there's no need for the AUR
> packages?
>

1.11.1 was the last version packaged in the repositories, 1.12 is the one
picking the new dependencies.

I started on packaging/building them (aeson and attoparsec but not
semigroups, monad-control, base-unicode-symbols, etc.) and upgraded all the
existing dependencies.

It wouldn't be hard to finish 1.12, but doing rebuilds whenever a
dependency gets upgraded takes too much time, especially since Haskell's
community is very fickle about the combinator/string/stream libraries they
use.


More information about the aur-general mailing list