Hi Lunar, thanks for rebasing our changes on the latest master branch!
On Freitag, 4. März 2016, Jérémy Bobbio wrote: > This versions implement changes discussed in #138409. One is that we are > now capturing some environment variables in .buildinfo files. In the > case of tests.reproducible-builds.org and the prebuilder script that means > that, as it is, every package would fail to be reproducible with the > current comparison. which environment variables are put into the .buildinfo files so that the comparison will fail? or maybe rather: why cant we make the .changes file include reproducible .buildinfo files? > So I refrained from uploading binaries yet. > > Couple of ways out I can think of: > > 1. Easy quick fix: use sed to remove .buildinfo from both .changes > file, then give .changes file to diffoscope. > 2. Slightly more complicated: change diffoscope to ignore most > fields in .buildinfo files. > 3. More involved: finally add “ignore modules” to diffoscope and > write a module to ignore most fields in .buildinfo files. I > would enable such a module by default. 4. teach diffoscope to ignore .buildinfo files when processing .changes files? > (I won't have time to work on any of this before probably 8-10 days.) > > Just for the record, Guillem said on IRC that he was waiting for us to > test the treewalk code before uploading dpkg/1.18.5 which ought to fix > file ordering issues. /me nods. cheers, Holger
Description: This is a digitally signed message part.
_______________________________________________ Reproducible-builds mailing list Reproducibleemail@example.com http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/reproducible-builds