(re-sending, as I forgot to include the bug last time) Hi Martin,
Thanks for the patch; I'll apply it to address this bug. On Sat, May 24, 2014 at 08:50:28AM +0200, Martin Pitt wrote: > Jelmer Vernooij [2014-05-24 2:50 +0200]: > > I can't reproduce this unless I run those commands from an unbuilt > > source tree. > Hm, I indeed just tried it again in schroot and it works now. So I > suppose what happened was that I checked out the subvertpy source and > as schroot keeps the current directory the import tried to import from > the (unbuilt) source tree. So, sorry for the false alarm! No worries. :-) > > I really like the idea of autopkgtest but it's still a pain to run > > these tests locally. > > Do you have suggestions how to make it easier? In the simplest case > you just need to run "debian/tests/testsuite", or with autopkgtest > itself with e. g. > > adt-run -B .// --- null # on local machine > adt-run -B subvertpy_0.9.1-5.dsc --- schroot sid # in "sid" schroot > > Due to historical/backwards compatible reasons this could certainly > look a bit prettier, but it's certainly not rocket science? adt-run indeed has a confusing command-line and is not very well documented (though I see the package now has a couple of .txt files that weren't there when I last looked). I tried setting up the lxc backend a couple of times, but ran into issues each time and gave up on it. It would be great if it was possible to use a cowbuilder/pbuilder chroot. I already have those set up and am not too keen on having yet another set of chroots to maintain. Maybe I should give the schroot backend a shot; I've never used it though, so that would mean becoming familiar with it first. > > For subvertpy we're just running the standard > > testsuite - installed - that already runs as part of the build anyway, > > so autopkgtest is just not worth the trouble at the moment for me. > > I'll disable or remove the autopkgtests with the next upload. > > As you wish, but it's trivial to fix, and will immediately tell you > when something in your dependencies or Python itself changes which > breaks subvertpy (and once we integrate autopkgtest into britney it > will even prevent that regression to land, as we already do in > Ubuntu). I appreciate what you're doing, and I look forward to the day we have that functionality! This is also the reason why I was keen to try out autopkgtest. That said... At the moment, using autopkgtest isn't worth the effort for me, at least for packages like subvertpy. I have not encountered bugs of this class for subvertpy so far, and currently I find the tools around DEP-8 just too frustrating to use (DEP-8 itself is quite sensible). Cheers, Jelmer -- Jelmer Vernooij <jel...@debian.org> Debian Developer http://samba.org/~jelmer/
signature.asc
Description: Digital signature
