targetcli-fb unreproducible: where's the diffoscope output gone?

[Christian Seiler]

On 02/18/2017 05:54 PM, Mattia Rizzolo wrote:
> On Sat, Feb 18, 2017 at 05:50:25PM +0100, Christian Seiler wrote:
>> Can I find that piece of information somewhere on that page?
> 
> Yes, at the bottom of the rbuild log.

Ah, now I understand the output. I initially thought rbuild
was just the first build, and didn't give it too much thought
since I didn't expect diffoscope to be called there.

>> It might be a good idea to change the code behind it to check
>> WIFSIGNALED(diffoscope_exit_code) and if that's the case consider
>> it to be a temporary test bed error, instead of a reproducibility
>> failure - which would result in just an error code.
> 
> Yes, you're totally right; it just take some work to do, it has been in
> the todo for a while, but it's rare enough to not in fact be a problem
> usually.
> For now all situations where we're not totally sure a package is
> reproducible (in our current framework), it'll be marked as unrep.

I could work on a patch for this. Could you point me in the right
direction? According to the wiki, you're using a modified version
of sbuild, but the link from the wiki leads via a mailing list
post [1] to

https://anonscm.debian.org/cgit/reproducible/sbuild.git/

which doesn't appear to exist (anymore).

Regards,
Christian

[1] https://lists.alioth.debian.org/pipermail/reproducible-builds/Week-of-Mon-20141229/000613.html