Il 26/11/24 16:32, Dan Horák ha scritto: > On Tue, 26 Nov 2024 08:10:57 -0700 > Jerry James <loganjerry@xxxxxxxxx> wrote: > >> On Tue, Nov 26, 2024 at 7:53 AM Dan Horák <dan@xxxxxxxx> wrote: >>> On Mon, 25 Nov 2024 18:03:37 +0000 >>> Mattia Verga via devel <devel@xxxxxxxxxxxxxxxxxxxxxxx> wrote: >>> >>>> I'd like some help to identify the issue behind Siril being FTB only on F40: >>>> >>>> https://koji.fedoraproject.org/koji/buildinfo?buildID=2590137 >>>> >>>> The error behind the failure is: >>>> >>>> ``` >>>> /usr/bin/ld: /tmp/ccPj3YgU.ltrans25.ltrans.o: in function `get_kernel': >>>> /builddir/build/BUILD/siril-1.2.5/redhat-linux-build/../src/gui/newdeconv.c:838:(.text+0x5f60): >>>> undefined reference to `estimate_kernel' >>>> /usr/bin/ld: >>>> /builddir/build/BUILD/siril-1.2.5/redhat-linux-build/../src/gui/newdeconv.c:835:(.text+0x6003): >>>> undefined reference to `gf_estimate_kernel' >>>> collect2: error: ld returned 1 exit status >>>> ``` >>>> >>>> Accordingly to Koschei [1], the F40 builds have started to fail in >>>> September with no change in the specfile. After that I enabled the >>>> optional ffms2 support and updated the version, but even reverting the >>>> ffms2 option doesn't fix the build issue. >>>> I can see some build flags are different between F41 and F40, but the >>>> flags used in the latest successful build on F40 seems consistent with >>>> those used now. >>> the question is what used to provide the estimate_kernel and >>> gf_estimate_kernel functions/symbols. The koschei output would suggest >>> it could be imath or libvpl (gcc, qt5-qtbase, python3, etc look unlikely >>> to me). >> It looks like siril itself provides both functions, estimate_kernel in >> src/filters/deconvolution/estimate_kernel.cpp and gf_estimate_kernel >> in src/filters/deconvolution/gf_estimate.cpp. So, some difference in >> the linker invoctions between F40 and F41/Rawhide? >> >> Although I will note that I just tried an F40 mock build of siril, and >> it succeeded. > it could be a parallel make/build issue when the linking runs before > the object files for the filters/deconvolution sources are ready > > > Dan > I have tried multiple times a 'fedpkg mockbuild' in F40, also disabling the parallel build (with '%global _smp_build_ncpus 1'), but I constantly get a build failure... so it doesn't seem a parallel build issue neither. Jerry, have you tried a normal mock build or a fedpkg mockbuild? (Could that change anything?) Mattia -- _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
