[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Package test service for GNU maintainers

From: John Darrington
Subject: Re: Package test service for GNU maintainers
Date: Tue, 19 Aug 2014 08:04:40 +0200
User-agent: Mutt/1.5.21 (2010-09-15)

On Sun, Aug 17, 2014 at 04:53:52PM +0200, Ludovic Court??s wrote:
     As a part of a discussion about ???cool features for GNU maintainers??? at
     the GHM, someone (Andreas, IIRC) suggested having a service (possibly a
     Web service) that would allow maintainers to upload their (pre)release
     tarballs and then get some feedback:
       ??? If the package is not available in the distro, it would just say
         ???sorry, your package is not available in the distro, but you can
         help!???.  Something like that.  ;-)
       ??? If the package is available, it would do something equivalent to
         ???guix build foo --with-source=foo.tar.gz??? and eventually return the
         build result.
         Upon success, it could also return a generated patch that can be
         sent to guix-devel for inclusion.
         It could also use something similar to ???guix refresh
         --list-dependent??? to try building dependent packages, or at least
         tell the maintainer to check them.
     I think we have most of the tools to do that, and it may be a good way
     to entice GNU maintainers into contributing to Guix.  Since it???s a
     lightweight process, we could suggest to make it a recommendation in the
     maintainer???s guide.
     Thoughts?  Who want to give it a go?  :-)
I could give it a go.

However there are a couple of things which are of concern.

1. The way you describe it above, would it not be considered SaaS ?

2. By default, guix does not check that out-of-source builds are functional.
   "make distcheck" on the other-hand does.  Do we want to duplicate the 
   of "make distcheck"?  Perhaps we do, since I have recently seen a rather bad
   GNU release where it obviously wasn't run.

3. Should we check too that a package behaves sanely when cross-compiling?

4. Some years ago, somebody knocked up a similar "GNU package linter" and ran
   it on all GNU packages.  One maintainer got very shitty about a "deficiency"
   that was detected - started demanding an apology, blah, blah blah ...


PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See or any PGP keyserver for public key.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]