[Qa-jenkins-dev] Bug#878133: reproducible: please report active variations in pbuilder logs

[Simon McVittie]

Package: jenkins.debian.org
Severity: wishlist

One of my packages (tap.py) fails to build from source in the second
(more-varied) reproducible build, and it isn't immediately clear how to
reproduce that failure - I think it's something to do with locales and
character sets, but it isn't obvious which locale was used.
https://tests.reproducible-builds.org/debian/index_variations.html
documents some possible variations, but nothing in the logs seems to
say which ones were actually active in a particular rebuild.

It would be great if the pbuilder logs for the paired builds would log
as much as possible of the variation, probably by running some hook
script. If both builds logged this, that would be maximally useful for
diffing. To be useful in the FTBFS case, the variation should be logged
before doing the actual build, so that it's visible even when failing.

The attached script logs everything currently mentioned in
https://tests.reproducible-builds.org/debian/index_variations.html
other than the working directory and its filesystem type. I don't know
the mechanics of pbuilder hooks (I don't use pbuilder myself), but it
looks as though it might be most useful as an "A" hook, in which case
it would hopefully see $BUILDDIR in its environment?

Thanks,
    smcv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: reproducible-info.sh
Type: application/x-sh
Size: 336 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/qa-jenkins-dev/attachments/20171010/4f6c8481/attachment.sh>