[Nix-dev] builderDefsPackage and propagatedBuildInputs

Alastair Pharo asppsa at gmail.com
Sun Oct 12 12:12:46 CEST 2014


Ok great that makes things easier!
On 12-Oct-2014 8:47 pm, "Domen Kožar" <domen at dev.si> wrote:

> This is exactly why I created https://github.com/NixOS/nixpkgs/issues/4210
> and would like to get rid of many ways to declare a package. I'd say plain
> mkDerivation is encouraged and whenever I upgrade a package using
> builderDefsPackage, I'd convert it into plain mkDerivation function.
>
> On Sun, Oct 12, 2014 at 10:48 AM, Alastair Pharo <asppsa at gmail.com> wrote:
>
>> Hi there,
>>
>> I am trying to put a patch together to upgrade the Pure
>> <https://github.com/NixOS/nixpkgs/tree/master/pkgs/development/interpreters/pure/default.nix>
>> package in nikpkgs to the latest version (0.62). In doing so, I’m also
>> trying to fix the package up so that Pure’s add-on modules
>> <http://puredocs.bitbucket.org/#addon-modules> can be packaged too.
>>
>> Regarding the latter objective, the main issue with the package at
>> present as I understand it is that it does not propagate any of its
>> buildInputs, so add-ons do not know where to look for the various
>> libraries and headers that Pure is built with. I have tried to fix this by
>> just splitting the dependencies into buildInputs and
>> propagatedBuildInputs, but this doesn’t seem to work, and I think the
>> reason is because the package’s nix expression is using the
>> builderDefsPackage function. From my very poor understanding of the
>> source code
>> <https://github.com/NixOS/nixpkgs/blob/master/pkgs/build-support/builder-defs/builder-defs.nix#L497>
>> this function has some special way of handling propagatedBuildInputs
>> (something to do with passthru?), however I have not been able to work
>> out what that is.
>>
>> I have success if I completely rewrite the nix expression for Pure using
>> stdenv.mkDerivation instead
>> <https://github.com/asppsa/nixpkgs/commit/f92a0281fa91ae6ca0a7a7ff701706ff2d6a0508>,
>> but I am supposing that my patch is more likely to be accepted if the
>> builderDefsPackage form is retained. As such, is someone able to help me
>> to understand this function better?
>>
>> Thanks in advance,
>>
>> Alastair
>>
>> _______________________________________________
>> nix-dev mailing list
>> nix-dev at lists.science.uu.nl
>> http://lists.science.uu.nl/mailman/listinfo/nix-dev
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.science.uu.nl/pipermail/nix-dev/attachments/20141012/6d7896e3/attachment.html 


More information about the nix-dev mailing list