| infinity | cjwatson: I really want to accept this livecd-rootfs, but I'm scared. | 03:10 |
|---|---|---|
| infinity | cjwatson: Did you do before and after test builds and compare package lists and (more importantly, IMO), apt-auto-install states? | 03:10 |
| cjwatson | infinity: I compared package lists and they looked broadly reasonable. I put a requirement to check before-and-after auto-install states in the bug's test case. | 11:30 |
| cjwatson | infinity: As I said in the bug report, if I had a less invasive idea which would actually work then I'd use it :-/ | 11:35 |
| doko | cjwatson, infinity: https://launchpad.net/ubuntu/+archive/test-rebuild-20121221/+build/4096089 | 15:44 |
| doko | INFO Upload was rejected: | 15:45 |
| doko | INFO Component "non-free" is not allowed in raring | 15:45 |
| doko | ? | 15:45 |
| doko | cjwatson, infinity: https://launchpadlibrarian.net/128286296/buildlog_ubuntu-raring-i386.armhf-cross-toolchain-base_1.97_FAILEDTOBUILD.txt.gz | 18:30 |
| doko | the package builds fine on amd64 | 18:31 |
| doko | shouldn't incompatible libraries just be skipped instead? I assume on amd64 it is done because the host ist 64bit, and the target is 32bit | 18:32 |
| infinity | doko: Have you noticed all the hung builds that have been killed in the gcc-4.8 rebuild test? | 19:06 |
| infinity | doko: Sometimes cc1plus hanging, sometimes perl hanging, but they were all in the gcc-4.8 rebuild, none in the normal one. Maybe a bug in libgcc1 or something? | 19:06 |
| doko | infinity, example? | 19:07 |
| infinity | doko: (There's an example hung on komainu right now, if you want to try to get some debugging info from webops or something, though that might be tough on a weekend) | 19:10 |
| infinity | doko: One on lamiak too. Same package, so that might actually be reproducible. | 19:10 |
| infinity | doko: Oh, wait. pdl might be hanging for other reasons, it might not be an example of what I was talking about. :/ | 19:10 |
| infinity | doko: Well, like I said, those two ongoing pdl builds. But they might be something else, I can't remember if they also failed in the normal test rebuild. | 19:12 |
| infinity | Hrm, no, they didn't. | 19:12 |
| infinity | So, yeah. Those. | 19:12 |
| infinity | doko: There were a fair few of them, though. Pretty much any x86 build log in the rebuild with "Session terminated, terminating shell... ...terminated." in it, and then evidence of someone forcibly killing a hung process is probably the same issue, whatever the issue is. | 19:17 |
| doko | infinity, any idea about the ldd issue? | 19:17 |
| infinity | doko: (It's not x86-specific, mind you, we saw it on ARM too, but timeouts can happen there for other reasons too) | 19:18 |
| infinity | doko: I'm looking at the cross/shlibs thing right now, it piqued my curiosity. | 19:18 |
| infinity | doko: I'm also looking at a glibc testsuite regression that only happens on Ubuntu, not Debian. Haven't bisected the cause yet, but it worries me a bit. :/ | 19:18 |
| infinity | doko: Yeah, the pdl testsuite looks like a good reproducer for the hang. At least, I can't reproduce it on plain raring, and it's currently hung on two buildds with gcc-4.8 | 19:29 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!