IRC logs for #buildstream for Friday, 2017-09-01

*** bochecha_ has quit IRC01:09
*** tristan has quit IRC04:55
*** jude has quit IRC05:19
*** jude has joined #buildstream05:59
*** jude has quit IRC07:13
*** jude has joined #buildstream07:32
*** jonathanmaw has joined #buildstream07:55
*** bochecha_ has joined #buildstream08:00
*** ssam2 has joined #buildstream09:01
*** tlater has joined #buildstream09:11
ssam2i keep getting confused between "fetch" and "pull"10:00
ssam2like running a build on a new architecture and thinking "how is it fetching artifacts??"10:00
ssam2but of course it's fetching sources10:00
ssam2getting a strange buildstream issue on ppc64b ... I have a sysroot compiled for ppc64b which I can chroot into manually, but when buildstream tries to open a shell it gets 'Exec format error'10:34
*** tlater has quit IRC10:40
ssam2seems like something is up in the SafeHardlinkOps filesystem10:41
ssam2I can chroot into and run stuff from the ./root directory fine10:41
*** tlater has joined #buildstream10:41
ssam2but the scratch/_/mount directory is totally weird10:41
ssam2if I try to run any program from that directory manually it exits with success but does nothing10:42
ssam2I get 'permission denied' trying to chroot10:42
ssam2in fact all the files seem to be 0 bytes in size10:43
*** tlater has quit IRC11:33
ssam2i've added some tracing to the SafeHardlinkOps class, and the getattr() method seems to be returning the correct file sizes11:46
ssam2and yet, `ls -l` shows everything as zero bytes11:46
ssam2I tried sshfs (a different FUSE filesystem) and it seems to work fine11:47
ssam2so, maybe an issue in the Python FUSE bindings :-(11:47
ssam2https://github.com/terencehonles/fusepy does claim to support PPC11:48
ssam2hmm but perhaps not ppc64 !11:48
ssam2looking at the code11:49
ssam2maybe https://github.com/terencehonles/fusepy/blob/master/fuse.py#L142 needs some work11:49
ssam2i'll try after lunch :-)11:49
*** tlater has joined #buildstream12:18
*** tlater has quit IRC12:27
*** vgrade has joined #buildstream12:35
*** tlater has joined #buildstream12:47
*** benbrown has joined #buildstream12:48
gitlab-br-botpush on buildstream@cross_platform (by Tristan Maat): 10 commits (last: Make sandboxchroot mount root as read-only) https://gitlab.com/BuildStream/buildstream/commit/96a81986164fe67393d677819ca99557bfe7fc9d13:15
gitlab-br-botbuildstream: merge request (cross_platform->master: WIP: Cross platform) #81 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/8113:15
ssam2i've tried to fix fusepy for ppc64 ... it seems to work now on ppc64b but not on ppc64l13:49
ssam2however the `struct stat` layout looks identical on both platforms so I'm a bit confused13:49
ssam2ah, ppc64l reports itself as ppc64le not ppc6413:50
ssam2https://github.com/terencehonles/fusepy/pull/7213:55
ssam2not much hope of that being merged though, the maintainer seems to have been AWOL for about a year13:55
ssam2maybe we'll have to bundle fusepy with buildstream. it's only 1 file anyway13:55
gitlab-br-botbuildstream: merge request (interactive-bst-shell-cmd->master: Allow forcing an interactive 'bst shell') #85 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/8514:05
gitlab-br-botbuildstream: issue #83 ("FUSE SafeHardlinkOps filesystem doesn't work on ppc64") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/8314:06
tlaterUgh, so close to finishing the sandbox problems :|14:56
*** tlater has quit IRC15:19
*** vgrade has quit IRC15:28
*** tristan has joined #buildstream15:50
ssam2i have a feeling that FUSE on ppc64 might also be super slow15:50
ssam2hi tristan15:51
ssam2hmm, perhaps not FUSE -- it's really slow if I `chroot` directly and run the commands myself too15:58
ssam2something isn't right though. i guess i have something to investigate next week ...15:58
*** jonathanmaw has quit IRC16:31
tristanssam2, ummm I agree with copying fuse.py into buildstream16:40
tristanssam2, there were some concerns raised about mixed licensing but we should have a strategy for this16:40
tristanit's not the first time, I need python click to be awesome but the project is in the mud16:41
tristanso I forked a module out of it with my bash completion changes, I still need to be sure about the licensing because the project is not totally clear on that :-/16:42
tristanit expresses a bsd license and then has a mostly empty LICENSE file prefixed with "this applies only to the parser.py module"16:42
gitlab-br-botbuildstream: merge request (interactive-bst-shell-cmd->master: Allow forcing an interactive 'bst shell') #85 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/8516:57
ssam2ok17:05
*** ssam2 has quit IRC17:14
*** tiagogomes has quit IRC17:19
*** jude has quit IRC17:39
*** jude has joined #buildstream18:16
gitlab-br-botpush on buildstream@master (by Tristan Van Berkom): 3 commits (last: documentation: Set page titles manually in python modules.) https://gitlab.com/BuildStream/buildstream/commit/3b35fc99c3b817fe3c07ead1beb93ed9a3d37f3c18:16
gitlab-br-botpush on buildstream@master (by Tristan Van Berkom): 1 commit (last: source-bundle: fix filtering out of non-build elements) https://gitlab.com/BuildStream/buildstream/commit/40ab6c994d2a7a045e47574c1c70148452899edd18:17
gitlab-br-botbuildstream: merge request (sb-elem-filter->master: source-bundle: fix filtering out of non-build elements) #83 changed state ("closed"): https://gitlab.com/BuildStream/buildstream/merge_requests/8318:19
tristanwhats with the docker image18:48
tristanintegration tests are not getting off the ground: https://gitlab.com/BuildStream/buildstream/-/jobs/3107034118:48
*** tristan has quit IRC19:07
*** bochecha_ has quit IRC19:32
*** bochecha_ has joined #buildstream19:35
*** tristan has joined #buildstream20:04
*** ChanServ sets mode: +o tristan20:04
*** jude has quit IRC23:54

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!