[haskell-llvm] pushing llvm-base with patched setup.hs to hackage

[Carter Schonwald]

because there is no simpler way to support both ranges of cabal versions in
setup.hs

by your own words, its bad to have needless version breakage, and theres no
way to do the conditional #if MIN_VERSION_BLAH stuff in a setup.hs files,
 and without this conditional compilation code, or some more complicated
config process that before the setup.hs, there is no way
to support both   both [Version >= 1.17] (and 1.18 will be out soon, and
will likely be the version in the next fall / winter haskell platform
release)
and [Version < 1.17].

I am all ears for a way to more nicely support a wider range of cabal
versions thats nicer. But until such a better solution that also works is
presented, this is the simplest solution.

LLVM lib bindings should *just work*, this is needed to hold that standard
up.


On Fri, May 17, 2013 at 4:19 PM, Henning Thielemann <
lemming at henning-thielemann.de> wrote:

>
> On Fri, 17 May 2013, Carter Schonwald wrote:
>
>  https://github.com/bos/llvm/**blob/master/base/Setup.hs#L90-**L116<https://github.com/bos/llvm/blob/master/base/Setup.hs#L90-L116>this is the change I made. It does not change
>> ANYTHING within the build process aside from making  llvm-base buildable
>> with  cabal > 1.16 as well as the
>> current <= 1.16 setup
>>
>
> I see. Since the comment qualifies the code to be a "horrible hack" and
> using Template Haskell in Setup.hs looks awful to me by itself - can you
> explain, what problem it shall solve and why do you think that this "hack"
> is the only or the best way to solve the problem? My experience is that it
> is even more difficult to get Template Haskell code running on different
> GHC versions, since TH changes with almost every GHC version.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://projects.haskell.org/pipermail/haskell-llvm/attachments/20130517/5e9cbdac/attachment.htm>