[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] gnu: Add ghc-unix-time.
From: |
Paul van der Walt |
Subject: |
Re: [PATCH] gnu: Add ghc-unix-time. |
Date: |
Thu, 22 Oct 2015 17:43:03 +0200 |
On 2015-10-22 at 10:47, quoth address@hidden:
>> + #:phases
>> + (alist-cons-before
>> + 'configure 'fix-/bin/sh
>> + (lambda _
>> + ;; Use `sh', not `/bin/sh'.
>> + (substitute* (find-files "." "Makefile|configure")
>> + (("/bin/sh")
>> + "sh")))
>> + %standard-phases)))
>
> See the ghc-x11 package for a, IMHO, more elegant solution.
Ah! Indeed.
> This appears to be a more general issue however. Any package that declares in
> its *.cabal file "build-type: Configure" is going to need CONFIG_SHELL and/or
> SHELL in the environment. See additionally the proposed ghc-sdl,
> ghc-sdl-image, ghc-sdl-mixer, and ghc-old-time patches.
>
> Perhaps we could patch haskell-build-system to export appropriate CONFIG_SHELL
> and SHELL variables if it sees a "configure" file in the top-level source
> directory. WDYT? Would you like to send a patch?
Right, i see what you mean. I agree that it'd probably be the right
thing to do. However, to expedite the merge of wip-haskell-… i propose
that i take the solution of ghc-x11 and apply it to the relevant
packages, and then look at fixing the haskell-build-system. At that
point we can simply remove the phases introduced to work around this
deficiency.
Do you think that's acceptable, or is it imperative* that we first solve
the problem in the build system?
I prefer the incremental approach (but i am not an objective outside
observer).
King regards,
p.
Re: [PATCH] gnu: Add ghc-xml., ericbavier, 2015/10/22
Re: [PATCH] gnu: Add ghc-uniplate., ericbavier, 2015/10/22
Re: [PATCH] gnu: Add Idris, the dependently-typed language., ericbavier, 2015/10/22