Hi Yavor, Am Tue, Oct 21, 2025 at 07:59:14PM +0300 schrieb Yavor Doganov: > Andreas Tille wrote: > > > Unfortunately the autopkgtest in my local pbuilder I get: > > Just out of curiosity, how do you run autopkgtests in pbuilder? I > couldn't find any documentation about it.
I think I've done my setup following https://wiki.debian.org/PbuilderTricks#Run_autopkgtest_to_run_tests_from_the_source_package To be honest: These days I would probably not start with this but would rather switch to sbuild since this is more actively developed and I was told its way more easy than it used to be. > > However, some tests are failing for i386. I decided to remove the > > package from all 32bit architectures to reduce the maintenance burden. > > I'm fine if someone might have a better idea to solve this problem. > > I think this is way too drastic and such approach should be used as a > last resort (ideally coordinated with upstream as well). The software is de facto orphaned so if we want to support 32bit we are on our own. > Only 2 test > files out of 310 fail which indicates that the package is far from > broken on i386 (cmdlineparseroutput is flaky and should be always > disabled). You are kindly invited to revert my admittedly drastic means. I simply was weighthing my personal time I could either spent for some quite rarely used software which is most probably not used on 32bit any more against fixing a heap of RC bugs in packages that are actually used. > I suggest to skip 2dimage-cost-lsd and 3d-vectorfield > tests on 32-bit architectures for the time being, at least until > someone with better C++ knowledge investigates the real culprit. Feel free to do so. As long as the package will migrate (which would not happen in case of i386 problems) that's perfectly fine. > > We also need to deactivate dh_auto_test for architecture=all builds > > (which I did not do yet). > > Right; I'll do this. I'll also remove the SALSA_BUILD stuff as > apparently it's not needed. ACK. > The autopkgtest and reprotest tests failed due to a libtool bug > (#1118477) which has been fixed in the meantime. Cool! Thank you for pointing this out. > The piuparts test fails for me locally too because the binary packages > in the archive (from -13) are not installable (they depend on old > library packages that are no longer available). Makes sense - so we can probably ignore this. > I think (but I'm not > sure) that when britney schedules this test, the inevitable failure > will not be a blocker for the migration of the package to testing. Thank you for the clarification. Just let me know if you finished what you said above. Kind regards Andreas. -- https://fam-tille.de
