Hello, On antradienis 26 Liepa 2011 23:50:24 Roger Leigh wrote: > On Tue, Jul 26, 2011 at 11:25:46PM +0300, Modestas Vainius wrote: > > Hello, > > > > On antradienis 26 Liepa 2011 22:59:46 Roger Leigh wrote: > > > On Tue, Jul 26, 2011 at 09:16:17PM +0200, Kurt Roeckx wrote: > > > > I'm also not sure we want to use the aptitude resolver in that > > > > case. > > > > > > We definitely don't. Its major flaw is that (unbelievably) it does > > > not return a nonzero exit status when it fails, so sbuild does not > > > know if installing the build deps worked or not. It typically fails > > > down the line some time after, but it's really unsuitable for building > > > unstable or anything else given this major flaw. It we can't > > > determine the build environment was set up correctly, we can't make > > > any guarantees about the quality of the build. > > > > Then check if sbuild-build-depends-packagename-dummy was actually > > installed after aptitude run and fail if it wasn't? > > Possibly. It's still not ideal though--we still end up failing at > a point other than where things went wrong. Sorry but I fail to understand the argument. The failure would still be in AptitudeResolver.pm, right after one additional check with dpkg-query -- status. It would not hurt to do this check in apt resolver too just as a precaution. > I would much rather have > the tool fixed to work properly rather than working around it (which > usually ends up failing at some point when one's assumptions are > broken). Sure, but then resolver would depend on the specific aptitude version which is not that desirable. Also getting the fix to stable might be complicated too since it is not clear how difficult is to fix this bug. Workarounds or additional safety checks do not hurt anyway. It's not like this one is a dirty hack or something. > Are there any circumstances were the dummy package is shown as being > installed where things are actually broken? Not possible since you implemented "install dummy package from the repository rather than force it via dpkg -i" feature. Neither apt nor aptitude would ever install a package when dependencies are unsatisfiable. > There's also cleanup to consider: we might also leave the chroot dirty > and broken at the end of a build, and be unaware of the fact. I still don't understand how this might happen if failure is always reported one way or another in the resolver code itself. -- Modestas Vainius <modax@debian.org>
Attachment:
signature.asc
Description: This is a digitally signed message part.