On Sat, Jan 28, 2012 at 09:07:34PM +0100, Mathieu Malaterre wrote: > Steve, > > On Sat, Jan 28, 2012 at 7:27 PM, Steve M. Robbins <steve@sumost.ca> wrote: > > On Sat, Jan 28, 2012 at 03:11:18PM +0100, Mathieu Malaterre wrote: > > > >> > Since it's released, I was planning to upload straight to 'unstable'. > >> > Do you think there's a need to stage in 'experimental' first? > >> > >> ITK will be build against gdcm. I would prefer to see gdcm transition > >> (#657288) to have ended (ie. gdcm 2.2.x into unstable) first. > > > > I'm not sure what your concern is; can you elaborate? > > Just trying to avoir another set of #(655783 655784 655785 655786 > 655787 655788) because ITK will be build using gdcm 2.0 So, what was the root cause of these bugs? I can't see a change in ITK during this time period. The most likely suspect I can see is a new gdcm version 2.0.19 in early January. Did it change ABI without changing SOVERSION? In any event, the proximal cause of this problem is due to ants, igstk and the like build-depending on ITK (v3). Initially, ITK v4 will have ZERO such reverse dependencies and thus ZERO impact on a gdcm transition. > > ITK 4 builds with the gdcm in unstable so if it builds OK with the new > > gdcm, I don't see it will hinder the latter's transition. > > The fact that ITK builds against 2.0 does not mean it builds fine with > 2.2 from experimental. I would really like to have feedback on that > combination just as fast as you for ITK OK, that's a different matter: I can pull the gdcm 2.2 packages into a chroot and build ITK v4 against it. (Or, you can do it since the ITK svn repository now builds) > > Moreover, I expect ITK 4 to be undergoing repeated source uploads to > > get it building everywhere (3.20.0 got up to rev -17) so it's likely > > that gdcm 2.2 gets into 'testing' before ITK 4 does for this reason > > alone. > > As said above, during this time, people will get another set of RC > bugs identical to #655783 and al. Not until ITK v4 has packages that build-depend on it. I don't anticipate that will happen for quite some time. In the worst case, we can delay transition of ITK v4 to testing (by an artificial RC bug, if necessary) until gdcm 2.2 transitions. > Can you just do at least one upload to experimental first ? I will do a test build of ITK 4 against gdcm 2.2 and post the results for discussion. -Steve
Attachment:
signature.asc
Description: Digital signature