Re: Guix and FSDG

[Clément Lassieur]

Hi David,

"Thompson, David" <address@hidden> writes:

> I believe this is incorrect and I will explain why.  This exact
> circumstance was brought up in the early days of the Guix project when
> FSDG compliance was a big topic of discussion because Ludovic and RMS
> were making sure that Guix conformed to it.  It is true that Guix will
> download source archives for packages that *may* contain files with a
> nonfree license.  However, Guix has a special mechanism developed
> specifically to deal with this issue.  In Guix, the <origin> data type
> is used to store information about a package's source code.  In this
> data structure there is a field called "snippet" which may contain a
> custom procedure written by the person that wrote the package.  The
> role of the snippet procedure is to *remove* any files in the source
> archive that are not freely licensed.  The result is a new source
> archive that contains only freely licensed files.  The most important
> part of this process is that the original source archive is *never*
> accessible to the Guix user via any Guix tools.  The original archive
> is discarded and does not end up in the canonical location for Guix
> data: /gnu/store.  Thus, running `guix build --source
> problematic-package` will only ever return the cleaned archive, never
> the original with nonfree files.  Therefore, Guix has taken sufficient
> technical measures to avoid steering its user towards nonfree software
> and thus Guix is compliant with the FSDG.

Unfortunately I think this isn't true.  For example this is what 'guix
build -S zpaq' outputs:

--8<---------------cut here---------------start------------->8---
substitute: updating substitutes from 'https://ci.guix.gnu.org'... 100.0%
The following derivation will be built:
   /gnu/store/k9g90yqfcz8c0lnz1mcfd9k53llyw8wh-zpaq715.tar.xz.drv
2,7 MB will be downloaded:
   /gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip
substituting /gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip...
downloading from 
https://ci.guix.gnu.org/nar/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip...
 zpaq715.zip  2.6MiB                  2.4MiB/s 00:01 [##################] 100.0%

building /gnu/store/k9g90yqfcz8c0lnz1mcfd9k53llyw8wh-zpaq715.tar.xz.drv...
/gnu/store/cnqpra8vr2l5fz00rr4yj4bp3hr00cfw-tar-1.32/bin/tar: Option --mtime: 
Treating date '@0' as 1970-01-01 00:00:00
zpaq715.zip/
zpaq715.zip/COPYING
zpaq715.zip/Makefile
zpaq715.zip/libzpaq.cpp
zpaq715.zip/libzpaq.h
zpaq715.zip/readme.txt
zpaq715.zip/zpaq.cpp
zpaq715.zip/zpaq.pod
source is under 'zpaq715.zip'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/libzpaq.h' -> 
`zpaq715.zip/libzpaq.h'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/zpaq.pod' -> 
`zpaq715.zip/zpaq.pod'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/Makefile' -> 
`zpaq715.zip/Makefile'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/zpaq.exe' -> 
`zpaq715.zip/zpaq.exe'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/libzpaq.cpp' -> 
`zpaq715.zip/libzpaq.cpp'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/COPYING' -> 
`zpaq715.zip/COPYING'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/zpaq.cpp' -> 
`zpaq715.zip/zpaq.cpp'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/zpaq64.exe' -> 
`zpaq715.zip/zpaq64.exe'
`/gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip/readme.txt' -> 
`zpaq715.zip/readme.txt'
successfully built 
/gnu/store/k9g90yqfcz8c0lnz1mcfd9k53llyw8wh-zpaq715.tar.xz.drv
/gnu/store/ran43jk8ngc60zzxgzhzdhkrpyb9i0ji-zpaq715.tar.xz
--8<---------------cut here---------------end--------------->8---

Both /gnu/store/32xxhz1sf3v9wxkhlvf4b9493dy588gi-zpaq715.zip and
/gnu/store/ran43jk8ngc60zzxgzhzdhkrpyb9i0ji-zpaq715.tar.xz are
accessible.

The former allows me to access zpaq64.exe and zpaq.exe although the
snippet is:

--8<---------------cut here---------------start------------->8---
(snippet
 ;; Delete irrelevant pre-compiled binaries.
 '(begin
    (for-each delete-file (find-files "." "\\.exe$"))
    #t))
--8<---------------cut here---------------end--------------->8---

Clément