On Wed, Apr 13, 2016 at 11:31:02PM +0200, Ludovic Courtès wrote: > Leo Famulari <l...@famulari.name> skribis: > > > On Sat, Apr 09, 2016 at 10:59:03PM +0300, Efraim Flashner wrote: > >> We do have some packages that have to be updated together, acme and > >> letsencrypt, python-cryptography and python-cryptography-vectors. But > >> those do have a commit message that they are updated together.
Python-cryptography-vectors is a test suite for python-cryptography [0]. I think it wouldn't be wrong for them to be integrated into the python-cryptography package if they are not used elsewhere. > > If those packages were updated separately, the build would break after > > the first commit. Neither of those package pairs works if the versions > > are not in sync [0]. > > > > In my opinion, that is a valid reason for updating multiple packages in > > the same commit. > > If that is the case (I wouldn’t be surprised), I agree it’s a valid > reason. I tried it, and pcmanfm 1.2.4 at least built successfully against libfm 1.2.3. But, pcmanfm only ran tests in the 'po' translations directory, where they warn about use of a deprecated regex syntax. > Anyway, it’s no big deal, so we shouldn’t block otherwise > uncontroversial and non-disrupting patches for 1 week just because of > this IMO. True, we got sidetracked. Sorry for the delay, Raymond! Applied as 58f738a1c8. [0] https://cryptography.io/en/latest/development/test-vectors/
