Dear Emmanuel, [I didn't see your reply until ginggs notified me, bug submitters aren't automatically subscribed to the bug].
On Thu, 25 Oct 2018 07:56:17 +0200 Emmanuel Promayon <emmanuel.proma...@univ-grenoble-alpes.fr> wrote: > From the autopkgtest log [1], it seems that vtk6 is installed along > side vtk7. Yes, as a bystander, some components from vtk6 and some from vtk7. > - I updated to camitk-4.1.2-2 so that it only depends on vtk7. vtk6 > should not be installed by autopkgtest (and is not when run inside > unstable). Than there is probably some (versioned) dependency wrong somewhere. autopkgtest installs only those packages that are listed in the autopkgtest control file of camitk version 4.1.2-2, but it tries to fulfill that with as many packages from testing as possible. Only those packages that must come from unstable, come from unstable. These dependency issues are often not found and/or triggered in unstable, so that is no surprise. > IMHO is that this autopkgtest regression is due to the fact that testing > has gdcm 2.8.7-1 (that depends on vtk6), while unstable has gdcm 2.8.7-5 > (uploaded 22 october, that depends on vtk7 only). > This problem should go away when gdcm 2.8.7-5 enters testing. If your analysis is correct, some package is missing a versioned dependency on gdcm >= 2.8.7-5~ then (or gdcm should have a versioned breaks). > Let me know what you think of this analysis and what I can do to help (I > just have a partial understanding on how these kind of problems should > be solved/displayed). This should be solved by adding that versioned dependency to the right package. Then autopkgtest will install the right version and everything should also work when testing in testing. Paul
signature.asc
Description: OpenPGP digital signature