IRC logs for #baserock for Sunday, 2016-03-20

*** _longines has quit IRC05:42
*** gtristan has quit IRC06:15
*** gtristan has joined #baserock06:40
*** locallycompact has joined #baserock07:59
*** gtristan has quit IRC10:13
*** gtristan has joined #baserock11:09
*** locallycompact has quit IRC11:31
*** gtristan has quit IRC11:35
*** gtristan has joined #baserock11:36
gtristanSomeone has a build log of procps-ng ?11:50
gtristanah, I think no worry11:52
gtristanummm... artifact splitting only happens at system creation time ...rrrright ?11:57
gtristanWe dont go selectively picking and choosing which parts of an artifact needs to be staged for a build that depends on that chunk... right ?11:57
* gtristan clearly has an ncurses artifact with /usr/include/curses.h... procps-ng fails to find curses.h, and it's indeed not staged, however the /usr/lib/libncurses stuff is there11:59
gtristanhmmm, oooohhh12:01
gtristanI know what happened12:01
*** cosm has quit IRC13:01
gtristanthis is odd13:32
gtristan2 16-03-20 00:09:22 [4/16/86] [TOTAL] Elapsed time 00:09:2213:32
gtristandont I have 12 more steps to do ?13:32
paulsherwoodyes, you should have 12 more steps. but i think you're not using latest ybd... and the counter has been borked at times14:13
paulsherwoodi think master now gets the count right, but it still needs more work for multi-instance
gtristanAh, ok, I think I rebased this weekend against master, maybe thats why14:21
paulsherwoodcould you paste the full log for that, please?14:22
gtristanHere is a subsequent paste, it's from ../ybd/ strata/tools.morph armv5l14:24
gtristanit ends on [0/12/86]14:25
gtristanand I still have the log from the one ending on [4/16/86] for the same invocation:14:26
paulsherwoodand this is rebased on master/16.11 ?14:26
gtristanummm, I may be a bit behind14:27
gtristanit's based on 646ada1515a216b46ad46a3d6916ac56a337efe814:27
gtristanMarch 14th: Add bootstrap-commands to defaults.conf14:28
gtristanno thats the tip14:28
gtristanthe others are merges of my pull request also present in my branch :)14:28
gtristanIt *could* have something to do with something stoopid I did though :)14:29
paulsherwoodi'm not sure14:29
gtristanI had an embarrassing thought... wouldnt it be nice if 'artifacts' started with the letter 'z' instead ?14:29
gtristanthen, when you accidentally do rm -rf ~/ybd instead of rm -rf ~/ybd/tmp... you have a better chance of still having the artifacts when you hit CNTL-C14:30
paulsherwoodanyway... hah14:30
gtristanThat was the cause of my ncurses missing header file... the artifact was there, but part of the .unpacked/ dir was removed14:31
paulsherwoodgtristan: you should never have to rm on ybd/tmp... ybd does that at the start of every run14:31
gtristanYes but I do it14:31
gtristanat times14:31
gtristanI have this habit of hacking function to not cleanup the old tmpdir14:31
paulsherwoodah, ok14:32
gtristanso when I hit an exception, it doesnt nuke the directory where it happened14:32
gtristanbut I still shouldnt need to, just a nasty habit, I learned the hard way14:32
paulsherwoodthe default shouldn't nuke anything until the next run?14:32
gtristan(and renaming 'artifacts' to start with 'z' was just a joke :))14:32
paulsherwoodgtristan: well, you could easily set artifacts-dir: zartifacts :)14:33
gtristanThe default nukes the tmpdir after every build, except when a build fails and app.exit() is called with an error message, then the "debris" remains14:33
gtristanbut if there is an unhandled exception, it's gone14:33
paulsherwoodthat's a bug, thne14:34
gtristanyeah I could hehe14:34
gtristanyeah probably a bug indeed, that removing of the directory shouldnt be in finally:14:34
gtristanit should be in only-when-no-exception-happened: (not sure python has that)14:35
gtristanor some retyping of that try/catch anyway14:35
* gtristan actually pushed some arm patches to gerrit14:36
gtristan, and
gtristannot entirely sure of the correctness of them14:38
gtristanthat perl thing (for parted.morph) only works accidentally on armv7lhf14:38
paulsherwoodare you in a position to try/test
gtristananyway, I have a *feeling* that gnome will build the rest of the way14:40
paulsherwoodthat will be quite a result14:40
gtristanthat looks right, but not really... I have 3 builds going on (glib/systemd and my third instance is building smaller chunks)14:41
* gtristan would rather not interrupt systemd... it uses libtool14:41
paulsherwoodi'll try it myself14:41
gtristanfwiw, just when you thought you had blamed libtool for everything under the sun, you find something new to blame it for !14:41
gtristanbuild times of a .lo are like... 90% running libtool, 10% distcc14:42
gtristanadd preprocessing to the mix somewhere in there, but it's not much of the equation14:43
paulsherwoodtested and pushed14:50
paulsherwoodback to your mis-counts... i have no idea what's happening there14:52
paulsherwoodthe last fix i did around this was for situations where user typed a bad arch... which would lead to lots of cache-keys, but no builds...14:53
*** gtristan has quit IRC15:29
*** gtristan has joined #baserock15:39
*** edcragg has joined #baserock17:36
*** locallycompact has joined #baserock18:00
*** gary_perkins has quit IRC21:24
*** brlogger has joined #baserock21:27
*** paulsherwood has quit IRC21:28
*** gary_perkins has joined #baserock21:28
*** paulsherwood has joined #baserock21:29
*** Zara has joined #baserock21:32
*** bwh has joined #baserock21:33
*** SotK has joined #baserock21:33
*** paulsherwood has quit IRC21:33
*** benbrown_ has joined #baserock21:33
*** paulsherwood has joined #baserock21:34
*** brlogger has joined #baserock21:40
*** edcragg has quit IRC22:59
*** edcragg has joined #baserock23:31

Generated by 2.15.3 by Marius Gedminas - find it at!