| Jasem[m] | Help! It appears something changed in Launchpad armhf farms. I have CMake checks for armv6l and armv7l and aarch64 via CMAKE_SYSTEM_PROCESSOR and none of the them are working now | 15:54 |
|---|---|---|
| Jasem[m] | https://launchpadlibrarian.net/363768459/buildlog_ubuntu-artful-armhf.libsbig_2.1.3~201804081003~ubuntu17.10.1_BUILDING.txt.gz | 15:54 |
| Jasem[m] | This build log ends up copying a 32bit blob instead of an armv6 blob.. and it used to work before so something must have changed on the build farm itself, maybe recent upgrade changed something? | 15:55 |
| Jasem[m] | IF (CMAKE_SYSTEM_PROCESSOR MATCHES "armv6l") does not work on Launchpad now | 15:56 |
| cjwatson | Jasem[m]: When did it last build? Have you tried diffing the build logs? | 19:49 |
| cjwatson | Jasem[m]: It's much more likely to be a change in one of your build-dependencies. | 19:50 |
| cjwatson | Ah, or maybe not, hmm | 19:53 |
| cjwatson | -Kernel reported to sbuild: 4.4.0-101-generic #124-Ubuntu SMP Fri Nov 10 18:30:44 UTC 2017 armv7l | 19:53 |
| cjwatson | +Kernel reported to sbuild: 4.4.0-119-generic #143-Ubuntu SMP Mon Apr 2 16:10:02 UTC 2018 armv8l | 19:53 |
| cjwatson | Looks like we lost the compat-utsname patch for some reason? | 19:54 |
| cjwatson | Jasem[m]: This is still a bug in your package - it must be conditionalising on uname somewhere and failing to handle armv8l (though conditionalising on uname is wrong anyway - you should override that in the packaging to conditionalise on DEB_HOST_ARCH) | 19:55 |
| cjwatson | Jasem[m]: But it is apparently a regression at our end, since we intend to report armv7l to armhf builds rather than armv8l | 19:55 |
| Jasem[m] | cjwatson: it's "sorta" resolved now. It was reporting armv6l before and this check (CMAKE_SYSTEM_PROCESSOR MATCHES "armv6l") used to pass fine | 20:02 |
| Jasem[m] | cjwatson: I changed it to MATHCES "armv+" for now | 20:02 |
| cjwatson | Jasem[m]: No, it was reporting armv7l before. | 20:03 |
| cjwatson | Pretty certain of that, and you can see it in the diff fragment above. | 20:03 |
| Jasem[m] | cjwatson: At any rate, it seems to link just fine with an armv6l blob for now.. | 20:05 |
| cjwatson | Jasem[m]: Right, but armv7l would be more correct. Anyway, we'll get it fixed, thanks for the note | 20:06 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!