Re: [PATCH 02/11] gnu: Add mingw-w64.

[Alex Kost <alezost@gmail.com>]

Hello, I understand nothing in all this mingw stuff, I just have a couple
of insignificant comments.

Jan Nieuwenhuizen (2016-08-09 09:41 +0300) wrote:

> * gnu/packages/patches/gcc-4.9.3-mingw-gthr-default.patch,
> gnu/packages/patches/mingw-w64-5.0rc2-gcc-4.9.3.patch,
> gnu/packages/mingw.scm: New files.
> * gnu/local.mk (dist_patch_DATA): Add them.
[...]
> +(define-public mingw-w64
> +  (package
> +    (name "mingw-w64")
> +    (version "5.0-rc2")
> +    (source (origin
> +              (method url-fetch)
> +              (uri (string-append
> +                    "https://sourceforge.net/projects/mingw-w64/files/mingw-w64/"
> +                    "mingw-w64-release/mingw-w64-v" version ".tar.bz2"))
> +              (sha256
> +               (base32 "0imdary8j07if8ih73pfgxiclpf2ax8h3mz8mxln07i8sbbd30c9"))
> +              (patches (search-patches "mingw-w64-5.0rc2-gcc-4.9.3.patch"))))
> +    (native-inputs `(("xgcc-core" ,xgcc-sans-libc-i686-w64-mingw32)
> +                     ("xbinutils" ,xbinutils-i686-w64-mingw32)))
> +    (build-system gnu-build-system)
> +    (search-paths
> +     (list (search-path-specification
> +            (variable "CROSS_C_INCLUDE_PATH")
> +            (files '("include" "i686-w64-mingw32/include")))
> +           (search-path-specification
> +            (variable "CROSS_LIBRARY_PATH")
> +            (files
> +             '("lib" "lib64" "i686-w64-mingw32/lib" "i686-w64-mingw32/lib64")))))
> +    (arguments
> +     `(#:configure-flags '("--host=i686-w64-mingw32")
> +       #:phases (modify-phases %standard-phases
> +         (add-before
> +             'configure 'setenv

The indentation of this 'modify-phases' is not good.  I would write it
like this:

  #:phases
  (modify-phases %standard-phases
    (add-before 'configure 'setenv
      (lambda ...)))


> +           (lambda _
> +             (let ((xgcc-core (assoc-ref %build-inputs "xgcc-core"))

Here, instead of referring '%build-inputs', it is better (I mean it
would be more "functional") to use 'inputs' argument passed to this
phase, like this:

  (lambda* (#:key inputs #:allow-other-keys)
    (let ((xgcc-core (assoc-ref inputs "xgcc-core"))

-- 
Alex