IRC logs for #buildstream for Monday, 2017-03-06

*** tristan has joined #buildstream04:43
*** ChanServ sets mode: +o tristan04:44
*** fox has joined #buildstream05:08
*** fox has quit IRC05:27
*** tristan has quit IRC08:36
*** tristan has joined #buildstream09:35
*** ChanServ sets mode: +o tristan09:35
*** tiagogomes has joined #buildstream09:52
*** ssam2 has joined #buildstream10:14
*** tristan has quit IRC14:41
*** tristan has joined #buildstream14:51
*** ChanServ sets mode: +o tristan14:51
tristanJust fyi, buildstream has gained some better job control14:52
paulsher1oodcool :)14:52
tristannot entirely hooked up everwhere I want... but for right now it means14:53
tristan- When hitting cntl-C (this can be changed if not desirable), we suspend all tasks and ask the user if they want to quit, second cntl-C will quit (and cleanup any stale state on disk for ongoing force terminated builds)14:54
tristan- Could be asked at that point of they want to keep processing only whats currently processing and then quit14:54
tristan- When hitting ^Z, we go to sleep and put child processes to sleep (note this is all especially tricky because of bwrap and all)14:54
ironfootall of this sounds great14:55
tristan- ^Z from the terminal works as one would expect, regardless of whether you are currently in an interactive state with suspended jobs or not14:55
tristanWhat I plan to use this for next, is in interactive mode, put those 5 webkits hammering your cores to sleep while you debug one build in a shell (again with some questions and stuff)14:56
tristani.e. when a build fails, automatically put things to sleep, ask user what they wanna do14:56
tristanIf anyone ever told you... just dont play with unix signals... well, I have to concur :)14:56
tristanOh and in the meantime, I have toyed with blessings and came up with this: https://paste.pound-python.org/show/XACCVWzrJ90wC9dPFPgg/14:59
tristanWith that approach, I hope to keep the "N last lines" of logging (again only in interactive mode connected to terminal), reserved for showing stuff about ongoing tasks15:00
tristanso you know what's going on, perhaps ticking timers for each element and what it's doing (but small as possible), but still keep the logging flowing naturally above15:00
ironfootbtw, the paste that you posted (https://paste.gnome.org/powkej7sl) expired before I could have a look at it15:04
ironfootjjardon[m]: ^15:04
tristanoh looks like I missed that too, whatever it was about15:04
tristanaha, well, I'm not sure everything is pushed to the baserock conversion script ( jjardon[m] )... and it needs some options I think depending on what you're converting15:06
tristanI think what is in defs2bst right now is baserock specific15:06
* tristan still curious what the error is heh15:07
jjardon[m]ironfoot: tristan sorry, I have the log in my personal laptop; I will try to post again when I arrive home15:14
tristanjjardon[m], ok... keep in mind that the script expects a ybd <target>.yml dump, it doesnt crawl through a definitions repo15:18
jjardon[m]ah, that is probably the problem15:21
tristanjjardon[m], this week I'll get back to completing the conversion and I'll add a README to that repo so everything is more clear16:19
tristanstill right now I doubt it's obvious how you have to base it in something16:20
tristani.e. for the baserock defs, it's expected to be an output dir holding the build-essential elements already16:21
*** xjuan has joined #buildstream17:02
*** ssam2 has quit IRC17:21
*** ssam2 has joined #buildstream17:38
*** tristan has quit IRC18:45
*** ssam2 has quit IRC18:55
*** jjardon[m] has quit IRC21:28
*** jjardon[m] has joined #buildstream21:43
*** jjardon[m] has quit IRC22:13
*** jjardon[m] has joined #buildstream22:22

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