IRC logs for #baserock for Thursday, 2015-07-30

*** edcragg has quit IRC07:32
*** jonathanmaw has joined #baserock07:59
*** bashrc has joined #baserock08:02
*** edcragg has joined #baserock08:14
*** mariaderidder has joined #baserock08:14
*** mdunford has joined #baserock08:43
*** ssam2 has joined #baserock08:51
*** ChanServ sets mode: +v ssam208:51
tiagogomes_ /me ponders why 'sh: foo: not found' is going to stdout on http://paste.baserock.org/surucikomi09:16
ssam2tiagogomes_: as in, you think it should be on stderr ?09:27
tiagogomes_ssam2, yes. It is an error after all09:28
ssam2what is the context? is this Morph, YBD, something else?09:29
tiagogomes_ssam2 morph09:29
ssam2tiagogomes: I guess it's because the build log is being written to stdout ?09:35
ssam2if that's running as part of a build process, it uses this code path: http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/morph.git/tree/morphlib/stagingarea.py#n29909:35
ssam2which clearly raises an exception on error, so you'd see an 'ERROR: In staging area xx: ..' message on stderr09:36
tiagogomes_no, it is not because of that. I took out the tee command09:38
tiagogomes_also doing that, would cause `exit, out, err = self._app.runcmd_unchecked(cmdline, **kwargs)` to not have anything in `err`09:39
tiagogomes_that's why when a build fails, it fails like this http://paste.baserock.org/wapawajoji09:40
ssam2ah, I see09:47
ssam2so the question is why 'sh' writes its error message to stdout09:48
ssam2i have no idea, in that case!09:48
*** locallycompact has joined #baserock10:07
*** locallycompact has quit IRC10:40
*** mdunford has quit IRC10:45
*** locallycompact has joined #baserock10:56
*** mdunford has joined #baserock11:00
*** mariaderidder has quit IRC11:09
*** lachlanmackenzie has joined #baserock11:10
*** mdunford has quit IRC11:11
*** mariaderidder has joined #baserock11:25
*** nowster has quit IRC11:31
*** mdunford has joined #baserock11:43
*** nowster has joined #baserock11:46
*** mdunford has quit IRC11:53
*** mdunford has joined #baserock11:57
*** petefoth has joined #baserock13:33
*** petefoth_ has joined #baserock13:39
*** petefoth has quit IRC13:40
*** petefoth_ is now known as petefoth13:40
*** petefoth has quit IRC13:43
*** locallycompact has quit IRC13:53
*** CTtpollard has quit IRC14:05
*** locallycompact has joined #baserock14:08
*** CTtpollard has joined #baserock14:12
*** CTtpollard has quit IRC14:24
*** CTtpollard has joined #baserock14:25
tiagogomes_What's the purpose of the build-log-on-stdout option?14:31
SotKI think it is used by distbuild workers, but I may be wrong14:31
tiagogomes_that's what the docs say, still I am puzzled by it14:32
pedroalvarezheh, I guess is an implementation detail14:33
tiagogomes_whatever it is, I am pretty confident that was a bad name the option14:33
tiagogomes_*bad name for the option14:33
pedroalvarezbad name? I know what it does and I only know its name!14:34
tiagogomes_pedroalvarez except the fact that if you don't pass that option, you can *still* see the build output out stdout14:35
SotKo.O14:35
ssam2it's used by distbuild, indeed14:35
ssam2there's a --verbose option that was added later which also lets you view the build log on stdout14:35
ssam2(and the old --verbose option is now --debug)14:38
*** locallycompact has quit IRC14:41
*** mdunford has quit IRC14:44
*** mdunford has joined #baserock14:44
*** jjardon has quit IRC15:29
*** drnic has quit IRC15:29
*** jjardon has joined #baserock15:35
*** drnic has joined #baserock15:41
*** jonathanmaw has quit IRC16:11
paulsherwoodnowster: your proposed split soln would also be helpful for https://github.com/devcurmudgeon/ybd/issues/9616:14
nowsterpaulsherwood: agreed, but that would be as a side-effect16:18
paulsherwoodyup. side-effects are useful sometimes :)16:18
nowsterpaulsherwood: am I right in thinking that ybd treats meta files as write-only? (I can only see it making them and checking for their presence.)16:23
paulsherwoodyes16:24
paulsherwood(so far)... but i expect that other tooling may want to read meta files at some point16:24
nowsterAnd we have a difference in formats... chunk metafiles are yaml; deployment metafiles are json16:25
paulsherwood(eg manifest generation, maybe license checking, checksumming etc)16:25
ssam2system-version-manager actually reads them already in order to carry certain settings across deployments16:25
ssam2which is needed for Jetson upgrades to work16:25
nowsterright! We don't want to break that.16:26
paulsherwoodah... that must fail for ybd upgrades then?16:26
ssam2not sure if it supports the YBD ones or not.16:26
paulsherwoodSotK: ^^?16:26
ssam2i've never tried, presumably if you deploy a system to Jetson with YBD, then try to upgrade it, it fails16:26
ssam2although actually if you specify DTB_PATH, KERNEL_ARGS, ROOT_DISK and BOOT_DISK in the cluster .morph when you upgrade then I guess won't fail16:26
ssam2I think it only needs to read the /baserock/deployment.meta file if those weren't specified explicitly16:27
nowstermorph metafiles are all json16:27
*** mariaderidder has quit IRC16:40
*** mdunford has quit IRC16:45
*** bashrc has quit IRC16:49
*** mdunford has joined #baserock17:01
*** edcragg has quit IRC17:17
paulsherwoodnowster: technically that means that morph metafiles are already YAML17:24
paulsherwoodi think :)17:24
* paulsherwood would prefer yaml, since it's easier to read17:24
*** mdunford has quit IRC17:26
*** lachlanmackenzie has quit IRC17:57
*** jjardon has quit IRC18:34
*** jjardon_ has joined #baserock18:35
*** ssam2 has quit IRC18:36
*** jjardon_ is now known as jjardon18:38
*** edcragg has joined #baserock19:59
*** jmacs has quit IRC20:51
*** jmacs has joined #baserock20:52
*** brlogger` has joined #baserock21:04
*** brlogger has quit IRC21:05
*** ratmice___ has quit IRC21:05
*** ratmice___ has joined #baserock21:05
*** edcragg has quit IRC23:42

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