On 28 Jul 2015, at 5:14 pm, Michael Stapelberg <stapelb...@debian.org> wrote: > > Chris, thanks for the explanation, that makes sense! I totally missed the > fact that there were not-yet-uploaded changes in git. > > Tim, I just tried packaging the new upstream snapshot, and the fix you got > upstream to merge was indeed not enough. My first guess is that exampleStr > requires section1 and section2 to be serialized in order, but Dict.format() > does not sort the keys when iterating over the map.
Yes that’s correct. I think that entire test needs to be rewritten - perhaps by deserialising the generated string and then making assertions against the dict. > It’s surprising that this hasn’t caused issues elsewhere yet. It doesn’t happen very often so perhaps the failure is easily missed if the tests aren’t run often and automatically. > So, yes, if you could work with upstream on a proper solution and then we > could just package a new upstream snapshot, that’d be great. OK - I’ll work on that then. Tim. > > On Tue, Jul 28, 2015 at 3:02 AM, Potter, Tim (Cloud Services) > <timothy.pot...@hp.com> wrote: > On 28 Jul 2015, at 6:35 am, solo-debianb...@goeswhere.com wrote: > > > > On Mon, Jul 27, 2015 at 07:26:10PM +0200, Michael Stapelberg wrote: > >> control: tags -1 + unreproducible > >> > >> Chris, was this an issue on your end? Or am I misinterpreting something? > >> > > > > The problem seems to have gone away. I was running local builds in > > response to errors on the reproducible-builds builder. Their builder > > has rebuilt sucessfully since then, and I have also just rebuilt > > sucessfully. Perhaps it was fixed by dependency changes? > > > > Proof I'm not mad: my build log from local: > > https://paste.debian.net/286717/ > > Hi everyone. I think this bug is due to this test relying on the ordering > of keys retrieved from > a hash being the same as they were inserted. This seems to work most of the > time (at > least on amd64) but occasionally the keys come out in a different order and > the test breaks. > > I could disable the test so things work and work with upstream for a proper > fix. Would that > help out some? > > > Tim. > > > > -- > Best regards, > Michael
smime.p7s
Description: S/MIME cryptographic signature