IRC logs for #baserock for Wednesday, 2017-04-26

*** gtristan has joined #baserock06:04
*** rdale has joined #baserock06:38
*** jude has joined #baserock07:02
*** gtristan has quit IRC07:32
*** ctbruce has joined #baserock07:34
*** gtristan has joined #baserock07:44
*** toscalix has joined #baserock07:49
*** gtristan has quit IRC08:02
*** paulw has joined #baserock08:03
*** gtristan has joined #baserock08:06
*** jonathanmaw has joined #baserock08:21
*** ssam2 has joined #baserock09:12
*** ChanServ sets mode: +v ssam209:12
*** tiagogomes has joined #baserock09:20
*** locallycompact has joined #baserock10:12
*** jude has quit IRC11:39
*** jude has joined #baserock12:07
*** gtristan has quit IRC13:45
*** ctbruce has quit IRC14:01
*** ctbruce has joined #baserock14:02
*** paulw has joined #baserock15:20
paulsherwoodtrying latest ybd, on latest definitions of gnome...15:36
paulsherwoodi see a *lot* of 'WARNING: Ignoring symlink "/src/tmp/tmpP9GFTr/usr/sbin" which purges non-empty directory '15:37
paulsherwoodis this a sign that the definitions are duplicating loads of stuff, or something else?15:37
paulsherwoodwin 4815:40
paulsherwoodbah15:40
ironfootthe /usr/sbin symlink is being unpacked after some other unpacked artifact created /usr/sbin/ folder?15:40
paulsherwoodyup... but i'm seeing hundreds of similar messages15:40
paulsherwoodquestion is why would so many artifacts all create the same things?15:41
paulsherwoods/all/all expect to/15:41
ironfootso, you are seeing more than one warning per artifact build?15:41
paulsherwoodyes, but for differnet files15:42
ironfoothm.. can you paste a chunk of this build log somewhre please?15:43
paulsherwoodactually they all appear to be /usr/bin /bin and /sbin15:44
paulsherwoodhttp://paste.baserock.org/zubebameto15:45
ironfootright, so fhs-dirs is being unpacked after /usr/sbin /bin and /sbin folder have been created. I don't know if that's an issue or not, tbh15:47
ironfootI don't know how those folders are being created beforehand. Maybe the unpacking is not being done in order?15:47
ironfootin any case, I would expect the final system to work if at least, when unpacking the artifacts, this is done in order15:48
ironfootwhen unpacking the artifacts when constructing the final system*15:48
ironfootI believe jjardon implemented the usr-merge in definitions, he may have some useful input on this15:49
*** ctbruce has quit IRC16:19
*** jonathanmaw has quit IRC16:26
paulsherwoodactually, that's now the least of my worries... http://paste.baserock.org/poteliduha16:49
paulsherwoodand http://paste.baserock.org/inovenaker16:50
paulsherwoodand http://paste.baserock.org/avodalowic16:51
ironfootwat16:53
paulsherwoodquite :)16:53
ironfootso, checks thats autogen.sh is there, and then executes it and says not found16:53
ironfootmight be a consequence of the warnings shown before16:57
*** toscalix has quit IRC16:59
*** gtristan has joined #baserock17:00
*** locallycompact has quit IRC17:02
ironfootso, for example, the autogen.sh in libnl needs /bin/bash: http://git.baserock.org/cgit/delta/libnl.git/tree/autogen.sh17:02
ironfootis that present in when building it? is the actual bash? or the not-real-bash generated in stage2 (which is only shell)17:03
paulsherwoodthe latter, i believe17:03
ironfootthe fact that the symlinks are being unpacked late, might be causing that some things are not being overwritten17:05
*** jude has quit IRC17:24
jjardonwierd, lets see if we can reproduce here: https://gitlab.com/baserock/definitions/pipelines/788768317:27
gtristanpaulsherwood, that is all due to random staging order and things in stage3 of build-essential not explicitly depending on fhs-dirs17:56
gtristanwell, I presume the staging order will not be that random if everything in stage3 of build-essential at least depends on fhs-dirs17:57
gtristanif the symlinks are being unpacked late, then you will have /usr/bin/bash, but no /bin/bash17:58
gtristanor, perhaps a fake bash which points to a non existing file17:58
gtristanin any case, it's all due to not depending on fhs-dirs, because with user merge, fhs-dirs *must* absolutely be staged first17:59
gtristanin any scenario where you want a symlink to a directory, you need the symlink created before anything puts stuff in that directory17:59
gtristanotherwise, what is staging to do ? actually move content around to new locations ?17:59
gtristanhttps://gitlab.com/baserock/definitions/builds/15104426 looks like that issue I previously had a link for and posted on some bug report but dont recall exactly what the url or related bug is18:02
gtristani.e. you need to build gcc with `CXX="gcc -std={something-else}"18:03
gtristanif you want to build the gcc 5 we have using a host gcc 618:03
gtristanor `CXX="g++ -std={something}"` rather18:03
gtristanSame error with https://gitlab.com/baserock/definitions/builds/1510442518:04
gtristanjjardon, before you end up reproducing with https://gitlab.com/baserock/definitions/pipelines/7887683, first you need to complete build essential by fixing that gcc issue18:05
* gtristan wonders where he wrote that down, it certainly was written down somewhere18:05
* gtristan even had the link included to the upstream gcc bug report18:08
*** ssam2 has quit IRC18:14
*** paulw has joined #baserock19:13
*** gtristan has quit IRC20:33
*** rdale has quit IRC23:56

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!