The Hurd, GNU, Guix, and GSoC

The Hurd, GNU, Guix, and GSoC

[Ludovic Courtès]

[-- Attachment #1: Type: text/plain, Size: 1100 bytes --]

Hello!

While browsing
<http://www.gnu.org/software/hurd/community/gsoc/project_ideas.html>, I
saw references to Stow{,fs}, and an item called “Hurdish Package Manager
for the GNU System”.

The mission of GNU Guix [0] is to be a package manager for, and
distribution of, the GNU system.  The distribution only works on
GNU/Linux currently, but the goal is to support the Hurd (when it has
source releases...), and possibly kFreeBSD.

Thus, it will most probably have to restrict itself to features provided
by all three kernels.  However, it could make use of Hurd-specific
features when running on it, such as unionfs.

GSoC-wise, possible projects include:

  • porting the distribution to GNU/Hurd (but again, I’d rather rely on
    published source tarballs of the Hurd, libc, etc. than on Git
    checkouts);

  • working on the integration of Guix on the Hurd, via unionfs/stowfs
    in particular.

WDYT?  Should we update the items in the ideas page to point to Guix?
Should we add new items?

Thanks,
Ludo’.

[0] http://www.gnu.org/software/guix/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: The Hurd, GNU, Guix, and GSoC

[Cyril Roelandt]

On 03/26/2013 06:24 PM, Ludovic Courtès wrote:
> Hello!
>
> While browsing
> <http://www.gnu.org/software/hurd/community/gsoc/project_ideas.html>, I
> saw references to Stow{,fs}, and an item called “Hurdish Package Manager
> for the GNU System”.
>
> The mission of GNU Guix [0] is to be a package manager for, and
> distribution of, the GNU system.  The distribution only works on
> GNU/Linux currently, but the goal is to support the Hurd (when it has
> source releases...), and possibly kFreeBSD.
>
> Thus, it will most probably have to restrict itself to features provided
> by all three kernels.  However, it could make use of Hurd-specific
> features when running on it, such as unionfs.
>

This makes perfect sense. Since FreeBSD is way more popular than the 
Hurd, maybe students would be more likely to be willing to work on a 
Guix + (k)FreeBSD project. Do we know (k)FreeBSD people who might aslo 
be interested in Guix ? :p

> GSoC-wise, possible projects include:
>
>    • porting the distribution to GNU/Hurd (but again, I’d rather rely on
>      published source tarballs of the Hurd, libc, etc. than on Git
>      checkouts);

What's so wrong with git checkouts ? Are you afraid the code might not 
be very stable ?

>
>    • working on the integration of Guix on the Hurd, via unionfs/stowfs
>      in particular.
>
> WDYT?  Should we update the items in the ideas page to point to Guix?
> Should we add new items?
>

I agree on the whole thing.


Cyril.

Re: The Hurd, GNU, Guix, and GSoC

[Ludovic Courtès]

Cyril Roelandt <tipecaml@gmail.com> skribis:

> On 03/26/2013 06:24 PM, Ludovic Courtès wrote:

[...]

>> The mission of GNU Guix [0] is to be a package manager for, and
>> distribution of, the GNU system.  The distribution only works on
>> GNU/Linux currently, but the goal is to support the Hurd (when it has
>> source releases...), and possibly kFreeBSD.
>>
>> Thus, it will most probably have to restrict itself to features provided
>> by all three kernels.  However, it could make use of Hurd-specific
>> features when running on it, such as unionfs.
>>
>
> This makes perfect sense. Since FreeBSD is way more popular than the
> Hurd, maybe students would be more likely to be willing to work on a
> Guix + (k)FreeBSD project. Do we know (k)FreeBSD people who might aslo
> be interested in Guix ? :p

Perhaps, but that’s off-topic.

>> GSoC-wise, possible projects include:
>>
>>    • porting the distribution to GNU/Hurd (but again, I’d rather rely on
>>      published source tarballs of the Hurd, libc, etc. than on Git
>>      checkouts);
>
> What's so wrong with git checkouts ? Are you afraid the code might not
> be very stable ?

No.  It’s just that, IMO, it’s the developers’ responsibility to
identify milestones in their project and to give it a version number.
Deferring that choice to people outside of the project may not be as
fruitful.

Ludo’.

Re: The Hurd, GNU, Guix, and GSoC

[Arne Babenhauserheide]

[-- Attachment #1: Type: text/plain, Size: 1431 bytes --]

Hi Ludovic,

Am Dienstag, 26. März 2013, 18:24:06 schrieb Ludovic Courtès:
> GSoC-wise, possible projects include:
> 
>   • porting the distribution to GNU/Hurd (but again, I’d rather rely on
>     published source tarballs of the Hurd, libc, etc. than on Git
>     checkouts);
> 
>   • working on the integration of Guix on the Hurd, via unionfs/stowfs
>     in particular.
> 
> WDYT?  Should we update the items in the ideas page to point to Guix?
> Should we add new items?

That sounds pretty nice. I watched your presentation on Guix and liked it a lot. The main reasons why I did not try it yet are time and being happy with Gentoo… and that I’m consistently out of diskspace… :)

But transparent binary packages sound really nice and the concept of defining packages as lisp files is quite close to the ebuilds I know from Gentoo - maybe even closer, because it might need less shell hacks. 

I think a really worthwhile goal would be to add a compatibility layer between debian source packages and Guix, so we can avoid splitting our efforts on different code organization methods.

Maybe a Guix package which essentially just refers to a debian source package - for getting the version as well as for getting the code.

Best wishes and the best of luck with Guix!
Arne
-- 
1w6 sie zu achten,
sie alle zu finden,
in Spiele zu leiten
und sacht zu verbinden.
→ http://1w6.org


[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 316 bytes --]

Re: The Hurd, GNU, Guix, and GSoC

[Ludovic Courtès]

ludo@gnu.org (Ludovic Courtès) skribis:

>   • working on the integration of Guix on the Hurd, via unionfs/stowfs
>     in particular.

I’ve pushed edits to the “project ideas” page toward that goal.

Comments welcome.

Ludo’.