Re: Differing release name and source version.

[Rob Vermaas]

Hi Mats,

indeed inetutils was configured to use the latest successful tarball build as input for the coverage and build jobs. I have rewritten [0] the nix _expression_ to the generic gnu-job.nix function, which should solve your issue and make everything consistent again. 

Cheers,

Rob

[0] http://git.savannah.gnu.org/cgit/hydra-recipes.git/commit/?id=43cf3dd2e01226278b91abffc7cc82c0a4f661e2

On Tue, Mar 31, 2015 at 8:44 PM, Mats Erik Andersson <address@hidden> wrote:

Hello there,

I noticed the job 20977014, which is 'coverage' for GNU Inetutils,
that it is claimed to be for release

  inetutils-1.9.2.59-3e7d-dirty

but in fact it is using the change

  3e7df6 to ae6b5b   of inetutils

Thus the correct release name should be

  inetutils-1.9.2.60-ae6b-dirty

The problem is that the project GNU Inetutils happened to push
a further commit between 'tarball' and 'coverage'. In fact, the
very same incorrectness is true for 20977016/manual. The job
20977015/tarball is displaying correct versioning.

It seems as if the secondary jobs, depending an a previous tarball,
are using the most recent upstream source, ignoring the versioning
claimed by the tarball dependency. Since the job 'manual' is building
the Nix packages, you have a race condition resulting in faulty
source tracking.

Best regards,
  Mats Erik Andersson, on behalf of GNU Inetutils

--

Rob Vermaas

[email] address@hidden