Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
Control: tags -1 - pending Removing tag pending like Emilio did it for #753044. DS -- 4096R/DF5182C8 46CB 1CA8 9EA3 B743 7676 1DB9 15E0 9AF4 DF51 82C8 LPI certified Linux admin (LPI000329859 64mz6f7kt4) http://www.danielstender.com/blog/ -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
The transition(s) of Vigra (1.10.0 and lib5v5) is just around the corner. We're checking the reverse-deps, all works/builds fine instead of current 3depict [1], and I've faced some problems with libreoffice ([2], also not related to libvigraimpex-dev, though). Yet little busy, but this is going to completed soon. Thanks, Daniel Stender [1] https://bugs.debian.org/798858 3depict: FTBFS against mathgl 2.3.3 [2] http://www.danielstender.com/vigra/libreoffice_5.0.1-1_amd64-20150911-2300.build -- 4096R/DF5182C8 46CB 1CA8 9EA3 B743 7676 1DB9 15E0 9AF4 DF51 82C8 LPI certified Linux admin (LPI000329859 64mz6f7kt4) http://www.danielstender.com/blog/ -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
On Thu, 13 Aug 2015 at 13:47:04 +0200, Daniel Stender wrote: We have some other serious issues open for Vigra (with the Lenna image set [2] and test suite problems in Mips), so I suggest we do it that way: I'm going to prepare a v5 1.9.0+dfsg-11 for unstable in the next days and check the reverse deps. Did this ever happen? I believe the current policy is that maintainers (and NMUers) should upload transitioning packages to unstable as soon as each library build-dependency that needs a transition has started it. If necessary, add versioned build-dependencies, to make sure that your package will go into Dep-Wait state on the buildds until their build-dependencies are at the transitioned version on the relevant architecture. The bug about the Lena sample/test images does not need to block this: while it is a bug that should be fixed, it isn't a regression (the version currently in testing is no better than the one in unstable in this respect), and unlike the libstdc++ transition it can't block work elsewhere in Debian. If the offending files are also present in stable/testing (which I suspect they are), please mark the bug as found in those versions so that the BTS knows what's going on. 1.10.0 in experimental seems to have built successfully on mips, hopefully that's a good sign for that bit. You shouldn't need to go through the NEW queue for a second time when 1.9.0+dfsg-11 is uploaded, because the new binary package name has already been approved. S -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
We have a SONAME bump happening due to the Vigra 1.10.0 transition [1] which generally could let spare a v5 package when the two transitions would be combined, isn't it? However, I think it's better to adopt what has been already changed for 1.10.0+dfsg-9ubuntu1 to update Vigra for 15.10 (renaming libvigraimpex5 to libvigraimpexv5 and rebuild) - there's really no need to avoid this. We have some other serious issues open for Vigra (with the Lenna image set [2] and test suite problems in Mips), so I suggest we do it that way: I'm going to prepare a v5 1.9.0+dfsg-11 for unstable in the next days and check the reverse deps. After that we go for a v5 1.10.0+dfsg-10 in experimental and check the reverse deps on that, that would close the stdc++6 transition as the next thing to do on Vigra. Daniel [1] https://bugs.debian.org/793044 (transition: libvigraimpex) [2] https://lists.debian.org/debian-devel/2015/08/msg00090.html -- http://www.danielstender.com/blog/ 4096R/DF5182C8 46CB 1CA8 9EA3 B743 7676 1DB9 15E0 9AF4 DF51 82C8 -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
On 2015-08-13 Daniel Stender deb...@danielstender.com wrote: We have a SONAME bump happening due to the Vigra 1.10.0 transition [1] which generally could let spare a v5 package when the two transitions would be combined, isn't it? However, I think it's better to adopt what has been already changed for 1.10.0+dfsg-9ubuntu1 to update Vigra for 15.10 (renaming libvigraimpex5 to libvigraimpexv5 and rebuild) - there's really no need to avoid this. I think that is sensible., too. We have some other serious issues open for Vigra (with the Lenna image set [2] and test suite problems in Mips), so I suggest we do it that way: I'm going to prepare a v5 1.9.0+dfsg-11 for unstable in the next days and check the reverse deps. After that we go for a v5 1.10.0+dfsg-10 in experimental and check the reverse deps on that, that would close the stdc++6 transition as the next thing to do on Vigra. [...] I would suggest to make a v5 for /experimental/ ASAP to minimize delay due to new processing. cu Andreas -- `What a good friend you are to him, Dr. Maturin. His other friends are so grateful to you.' `I sew his ears on from time to time, sure' -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
Package: src:libvigraimpex Version: 1.10.0+dfsg-9 Severity: serious Tags: sid stretch confirmed User: debian-...@lists.debian.org Usertags: libstdc++-cxx11 [ confirmed, for both 1.10.0+dfsg-9 in experimental and 1.9.0+dfsg-10 in unstable ] Background [1]: libstdc++6 introduces a new ABI to conform to the C++11 standard, but keeps the old ABI to not break existing binaries. Packages which are built with g++-5 from experimental (not the one from testing/unstable) are using the new ABI. Libraries built from this source package export some of the new __cxx11 or B5cxx11 symbols, and dropping other symbols. If these symbols are part of the API of the library, then this rebuild with g++-5 will trigger a transition for the library. What is needed: - Rebuild the library using g++/g++-5 from experimental. Note that most likely all C++ libraries within the build dependencies need a rebuild too. You can find the log for a rebuild in https://people.debian.org/~doko/logs/gcc5-20150701/ Search for BEGIN GCC CXX11 in the log. - Decide if the symbols matching __cxx11 or B5cxx11 are part of the library API, and are used by the reverse dependencies of the library. - If there are no symbols matching __cxx11 or B5cxx11 in the symbols forming the library API, you should close this issue with a short explanation. - If there are no reverse dependencies, it should be the package maintainers decision if a transition is needed. However this might break software which is not in the Debian archive, and built against these packages. - If a library transition is needed, please prepare for the change. Rename the library package, append v5 to the name of the package (e.g. libfoo2 - libfoo2v5). Such a change can be avoided, if you have a soversion bump and you upload this version instead of the renamed package. Prepare a patch and attach it to this issue (mark this issue with patch), so that it is possible to NMU such a package. We'll probably have more than hundred transitions triggered. Then reassign the issue to release.debian.org and properly tag it as a transition issue, by sending an email to cont...@bugs.debian.org: user release.debian@packages.debian.org usertag this issue + transition block this issue by 790756 reassign this issue release.debian.org - If unsure if a transition is needed, please tag the issue with help to ask for feedback from other Debian developers. The libstdc++6 transition will be a large one, and it will come with a lot of pain. Please help it by preparing the follow-up transitions. [1] https://wiki.debian.org/GCC5#libstdc.2B-.2B-_ABI_transition -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
Bug#794736: libvigraimpex: library transition is needed when GCC 5 is the default
On 2015-08-06 Matthias Klose d...@debian.org wrote: Package: src:libvigraimpex Version: 1.10.0+dfsg-9 Severity: serious Tags: sid stretch confirmed User: debian-...@lists.debian.org Usertags: libstdc++-cxx11 [ confirmed, for both 1.10.0+dfsg-9 in experimental and 1.9.0+dfsg-10 in unstable ] [...] Hello, how about combining this with #793044 (transition: libvigraimpex) instead of doing two transitions? Ubuntu already has a patch for the Vigra-1.10.0 C++ transitiion. cu Andreas -- `What a good friend you are to him, Dr. Maturin. His other friends are so grateful to you.' `I sew his ears on from time to time, sure' -- debian-science-maintainers mailing list debian-science-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-science-maintainers
