IRC logs for #baserock for Monday, 2015-04-06

*** kejiahu has quit IRC00:37
*** kejiahu has joined #baserock00:37
*** edcragg has joined #baserock10:16
*** edcragg has quit IRC10:54
*** edcragg has joined #baserock10:56
*** edcragg has quit IRC11:05
*** edcragg has joined #baserock11:12
jjardonHi, was there any technical reason to hardcode the use of /src instead use the user home directory with the directorirs specify in the XDG Base Directory Specification?11:21
straycathardcode the use of /src for storing morph cache etc?11:22
paulsherwoodjjardon: one thing was/is so that /src can be mounted as a separate volume, i believe11:41
paulsherwoodwhat would be the advantage of changing?11:41
* paulsherwood googles, and notices ryan lortie in the thick of xdg specification11:42
jjardonIt would make easier to run morph outside chroot/vm11:57
straycatiirc the only hardcoded part is the location of the morph.conf file11:59
jjardonpaulsherwood: also, you can still mount your local folder as a separate volume12:11
paulsherwoodjjardon: ok12:18
paulsherwoodjjardon: would you accept foo-tools or foo-set rather than foo-stratum ?12:18
paulsherwoodor foo-things :)12:20
paulsherwoodor foo-group12:20
jjardon:) From those options I would prefer foo-group12:20
paulsherwoodstraycat: can i interest you in the above?12:21
paulsherwood(if so i'll send patches...)12:22
straycathrm, cat of little brain today, what is this in relation too?12:22
paulsherwoodmy RFC on the list... trying to disambiguate ruby chunk and ruby stratum etc12:22
straycatoh, i'd forgotten about the ml, hang on12:23
*** gary_perkins has joined #baserock12:24
straycatoh, i don't personally find that a problem12:25
jjardonpaulsherwood: Id say go ahead and write the patch, your proposal makes sense doesn't matter the name12:26
paulsherwoodjjardon: ok, i'll do that12:27
straycatwhy is it confusing?12:28
paulsherwoodtwo main reasons i think...12:29
paulsherwood1) for users, a name should mean one thing (eg ruby should mean the ruby program, not the collection of things required to build ruby imo)12:30
paulsherwood2) from a code point of view, it would be simpler if names were unique12:31
straycati don't think so, right now the names add no information, you could remove them if you wanted12:32
jjardonpaulsherwood: Do you have any opinnions about ?12:32
paulsherwoodstraycat: they get reported to users all the time12:32
paulsherwood(in logs)12:32
straycatokay i must be mistaken12:33
paulsherwoodjjardon: that appears to provoke very strong reactions from others, in both directions! :)12:35
straycatjust have an alias for --build-log-on-stdout and be done12:35
jjardonpaulsherwood: yeah, thats why I wait to commit even it already have 2 +1 for long time12:35
jjardonstraycat: sure, any suggestions?12:36
paulsherwoodjjardon: what's wrong with straycat's suggestion? (although i admit it took me a *long* time to realize that this option existed)12:36
paulsherwoodjjardon: i think he means user can set an alias12:37
straycatwell, it was only there for distbuild >.> i still think a nicer interface for accessing the existing build logs might be nice, but we can do both12:37
jjardonMy point in the comment that its going to be more likely that a baserock user is interesed in the build log than in the log from morph, and probably the intuitive option would be to use -v for this. But anything would be better then --build-log-on-stdout :)12:37
straycatjjardon, i don't know what it should be called really12:37
paulsherwoodjjardon: i do agree with you. personally i find morph's actual log to be of approximately zero usefulness12:38
paulsherwood(which leads me normally to hack morph and make it print what i need)12:39
straycatit's useful for developers, less useful for users12:39
jjardonexactly, I think the option shoulw be --morph-log-on-stdout for the morph log and -v for the actual build12:40
paulsherwoodjjardon: i agree, i'm reluctant to get drawn into a religious debate :)12:41
paulsherwoodjjardon: wouldn't it be better if morph just clearly told the user where the build log went, rather than spewing it onscreen, though?12:42
straycatalright, so use the verbose log for build output etc12:43
straycatand have a --debug option12:43
straycator something12:43
jjardonpaulsherwood: sure, but I think thats a different bug12:43
paulsherwoodjjardon: but if it were fixed, would you still see a real need for that patch?12:44
straycatI don't know why we don't just drop the build logs into the definitions repo like we do for distbuild really, perhaps because then that doesn't really give you a good way to look at older logs12:45
jjardonpaulsherwood: yes, one step less if I can see the output instead check the output file (I can simply scroll in my terminal). For example, jhbuild shows everything on the terminal; maybe Im used to that12:46
straycatiirc nix also prints everything when it builds12:47
straycatwell, at least that can be configured12:47
paulsherwoodybd just displays the build commands it runs, at the moment, while putting the log in the artifacts directory12:47
paulsherwood(which i think is a reasonable soln)12:48
paulsherwood(since the log is an artifact)12:48
straycatthat's where the build logs are right now in morph12:49
paulsherwoodstraycat: only on success, iirc? last time i looked in detail morph was putting them in some failed tmpdir?12:53
paulsherwoodalso, morph doesn't feature the name of the component in the name of the build-log, which is... .annoying :)12:54
jjardonpaulsherwood: Did you submit a bug about that? :)12:56
paulsherwoodno. i could submit a patch, though :)12:58
straycatpaulsherwood, possibly, i don't remember13:00
straycatjjardon, btw i did comment on your response to
straycati'm really hoping this versioning stuff gets sorted this week because it's already holding stuff up13:02
jjardonpaulsherwood: thats even better :)13:04
straycatindeed patches welcome!13:04
*** straycat is now known as undeadcat13:04
* jjardon starts to think that maybe the jmacs suggestion to set in the definitions the needed version of morph would be better13:06
undeadcatwhat's wrong with what we have?13:08
undeadcati mean, i'd like to make it a set rather than a list as richard_maw suggested, but other than it works fine13:10
undeadcatalso, having a version of definitions gives us more control in morph, i think13:10
undeadcati mean, morph is where the decisions are made13:10
* paulsherwood looks at the backlog of changes waiting to go into morph, and decides not to increase it by sending patches13:25
jjardonpaulsherwood: send it anyway. While some patches require more time for review because needs the approval of a more experience morph developer , others can be easily reviewed by anyone13:34
undeadcatpaulsherwood, should be okay? unless your change changes the definitions format :)13:34
paulsherwoodundeadcat: :)13:35
jjardonundeadcat: returning 0 is exactly what I'm suggesting: "I think maybe its better in _parse_version_file to return "0" instead None if we are not able to parse the file"13:42
undeadcati don't get that, the parse function should just parse13:45
undeadcatthen if you get None, return 013:45
undeadcatso functionally equivalent, but without putting the policy into the parser13:45
jjardonalso, no need to change that whn we stop supporting version 0: In that case we will detect we are version 0 and we will fail (which is the correct thing to do)13:53
*** pedroalvarez has quit IRC13:53
*** flatmush has quit IRC13:53
*** kejiahu has quit IRC13:53
*** jjardon has quit IRC13:53
*** simonh_ has quit IRC13:53
*** inara` has quit IRC13:53
*** Zara has quit IRC13:53
*** bwh has quit IRC13:53
*** petefotheringham has quit IRC13:53
*** rjek has quit IRC13:53
*** petefoth has quit IRC13:53
*** sambishop has quit IRC13:53
*** paulsherwood has quit IRC13:53
*** ctgriffiths has quit IRC13:53
*** JPohlmann has quit IRC13:53
undeadcatokay yes Kinnison said that too13:54
*** kejiahu has joined #baserock13:55
*** jjardon has joined #baserock13:55
undeadcatjjardon, okay yes Kinnison said that too13:55
*** simonh_ has joined #baserock13:56
*** inara` has joined #baserock13:56
*** Zara has joined #baserock13:56
*** petefotheringham has joined #baserock13:56
*** bwh has joined #baserock13:56
*** rjek has joined #baserock13:56
*** pedroalvarez has joined #baserock13:58
*** flatmush has joined #baserock13:58
*** sets mode: +v pedroalvarez13:58
*** petefoth has joined #baserock13:58
*** sambishop has joined #baserock13:58
*** paulsherwood has joined #baserock13:58
*** ctgriffiths has joined #baserock13:58
*** JPohlmann has joined #baserock13:58
undeadcatjjardon, i'll rework it to not raise that exception when i get a moment13:59
undeadcathrm, i'm tempted to submit something as a draft14:36
*** edcragg has quit IRC14:37
undeadcatgerrit > 2.3 has this iirc?14:37
*** edcragg has joined #baserock14:38
undeadcatpaulsherwood, i wonder whether that might require some changes to the cache fetch stuff14:38
jjardonundeadcat: thanks15:08
jjardonpaulsherwood: nice, reviewed :)15:08
paulsherwoodjjardon: i'm not sure whether my reply to your comment is sent or not... it says 'draft' on my screen15:12
* paulsherwood finds a different way to reply in gerrit15:15
undeadcatpaulsherwood, yes you need to go back up to the change and then click reply for all your draft comments to be submitted15:16
undeadcati think this confuses everyone new to gerrit15:17
paulsherwoodundeadcat: :-)15:19
* undeadcat discovers the show-dependencies command15:26
jjardonundeadcat: ty for the review, I will reword and send back15:31
undeadcatnp :)15:43
*** edcragg has quit IRC16:25
jjardonDo anyone know if the port to armv8b64 has been completed?17:04
paulsherwoodi believe it has... kejiahu ?17:11
richard_mawlast I saw there were still patches on gerrit, and I'm not at all fond of the approach taken to allow a separate boot partition17:13
paulsherwoodah, ok17:36
jjardonDo you know if there is a place somewhere (storyboard?) when I can take a look to the remaining items?17:37
paulsherwoodjjardon: i don't understand your question, sorry17:49
jjardonDo you know if there is a place somewhere (maybe storyboard?) where I can take a look to the remaining items of the armv8b64 port?17:50
*** undeadcat is now known as straycat17:58
straycatrichard_maw, is a draft of the change i was working on yesterday18:59
straycatbah, i should have added you as a reviewer before i wrote the "cover letter" comment19:01
richard_mawstraycat: remind me to take a look at it tomorrow, I'm not capable today19:36
straycathehehe :)19:44
straycatokay thanks :)19:44
*** gary_perkins has quit IRC22:02

Generated by 2.15.3 by Marius Gedminas - find it at!