IRC logs for #baserock for Thursday, 2016-01-28

*** Lachlan1975 has joined #baserock07:55
*** Lachlan1975 has quit IRC08:05
*** toscalix has joined #baserock08:38
*** gtristan has quit IRC08:42
*** ctbruce has joined #baserock08:57
*** bashrc_ has joined #baserock09:07
*** rdale has joined #baserock09:22
*** jonathanmaw has joined #baserock09:37
*** franred has joined #baserock10:00
*** ctbruce has quit IRC10:04
*** gtristan has joined #baserock10:04
*** ctbruce has joined #baserock10:11
*** edcragg has joined #baserock10:19
*** tiagogomes__ has quit IRC10:22
*** tiagogomes__ has joined #baserock10:37
*** locallycompact has joined #baserock11:00
*** jonathanmaw has quit IRC11:23
*** jonathanmaw has joined #baserock11:23
*** jonathanmaw has quit IRC11:31
*** jonathanmaw has joined #baserock11:33
*** gtristan has quit IRC12:14
*** jonathanmaw_ has joined #baserock12:41
*** jonathanmaw__ has joined #baserock12:41
*** jonathanmaw has quit IRC12:45
*** gtristan has joined #baserock12:47
locallycompacthow much space do I need for a g.b.o mirror these days?13:05
*** radiofree is now known as radiofree_13:05
*** radiofree_ is now known as radiofree__13:05
*** radiofree__ is now known as radiofree___13:05
*** radiofree___ is now known as radiofree____13:05
*** radiofree____ is now known as radiofree13:06
*** tiagogomes__ is now known as tiagogomes13:07
benbrown_Huh? radiofree stopped evolving!13:07
pedroalvarezlocallycompact: only for gits? let me check13:07
pedroalvarezlocallycompact: question: including the copy that lorry has? or without it?13:08
pedroalvarezso "space of all the gits" vs "space that a trove needs to mirror all the gits"13:09
locallycompactpedroalvarez, I guess with, a full downstream trove13:09
pedroalvarezok give `du` a moment13:10
*** ctbruce has quit IRC13:19
locallycompactpedroalvarez, this is a long moment13:29
pedroalvarezlocallycompact: 52.6Glorry/13:32
pedroalvarezit's still with git/13:32
pedroalvarezgit/ should be around 1/2 times lorry/ IIRC13:36
*** toscalix_ has joined #baserock13:47
*** toscalix has quit IRC13:51
*** toscalix_ has quit IRC14:19
pedroalvarezbenbrown_: thanks for spotting that! Now merged14:29
benbrown_pedroalvarez: Thanks!14:30
mwilliams_ct\o/ concourse can build a bit more then :D14:30
pedroalvarezlet's see how many more errors you find14:30
mwilliams_ctI'm pretty pleased that the CI worked exactly as it should and spotted something was wrong14:32
mwilliams_ctless pleased that I spent literally seconds blaming myself before I decided to blame you instead14:32
pedroalvarezonly seconds? that's good14:34
pedroalvarezlocallycompact: 52.6Ggit/14:52
locallycompactpedroalvarez, that was a a very long long long.14:53
locallycompactpedroalvarez, thanks very much14:53
pedroalvarezI know, sorry14:53
pedroalvarezmaybe there was a faster way to do it14:54
persiaIf you have dedicated mount points for specific data paths, filesystem metadata can provide that information in real-time.16:00
paulsherwoodanyone seen gtristan lately?16:02
* paulsherwood is wondering how the aboriginal exploration is going16:02
*** jonathanmaw__ is now known as jonathanmaw16:14
gtristanpaulsherwood, I'm here at the hackfest...16:23
gtristanI read the email about naming of everything yesterday...16:24
gtristanI think... or I want, personally... groups to not be layers first of all, but what I mostly want understood is that groups are not mutually exclusive16:25
gtristanI thought maybe "assembly" may be a good word for ... a way one can assemble a bunch of components16:26
gtristanbut I'm basically ambivalent about naming and dont care so much16:26
gtristanwhats not nice about 'layers' is that it seems to imply, that if a component is in one layer, it cannot be repeated in another in the same system16:26
gtristanand thats what I think is part of the problem right now, stratum are containers of their chunks, instead of groups which simply refer to chunks which may appear more than once in the groups which are included in a system, but are only built once16:27
gtristanon aboriginal, I have not gotten to running ybd on the image again yet... what I did last time to build python and all the requirements to run ybd on aboriginal was... just a try and I dont want to waste time doing it again16:28
rdalewhat if the build-depends are different for the same chunk in different groups?16:28
SotKif a chunk is in more than one group, what do you do if each group asks for a different version of the chunk?16:28
gtristanabout aboriginal, currently landley is looking into fixing the 'lfs control image' which should present a good way of extending aboriginal from within aboriginal, to reach a point (repeatably) at which ybd can run again16:29
gtristanTo answer that, if they are parallel installable, they should be named differently16:30
gtristanlike say, webkit-1 and webkit-2 is a good example16:30
gtristanif they are not parallel installable, the build should early return and report an error16:30
Zarawhat if the chunks aren't versioned except via, say, sha1?16:30
gtristanideally, we have the ambivalences which my proposed variants provide, and there is a voting system16:31
gtristanZara, this would be anyway impossible16:31
gtristanbecause you would have a directory full of all the chunk definitions16:31
gtristanand each one's file name would be named after the chunk16:31
gtristanyou could never have 2 chunks with the same name16:32
gtristanthe problem now is that they are contained in strata, so there is room for this error where we actually build 2 things and install them on top of eachother, and dont even have a clear story of which one "wins"16:32
SotKah, so you would put the repo, ref, and so forth data in the chunk morphology?16:33
gtristanAnyway, I know this is not the case *right now*, but, if we're examining the renaming of everything, I would like to take into consideration for that renaming festival, that chunks are in fact sharable entities across stratum, and not contained inside of stratum16:33
gtristanSotK, yes, I would have one file for each chunk16:34
*** toscalix has joined #baserock16:34
gtristanand in fact the ref could potentially be conditionally different in a "variant"/"flavor", in which case other depending chunks could be ambivalent about the flavor, or prefer a specific one16:35
gtristanif we get to the point of implementing the variants approach, which... I think is of interest...16:35
*** toscalix has quit IRC16:47
*** toscalix has joined #baserock16:47
*** gtristan has quit IRC16:50
paulsherwoodit is to me, for sure :)16:51
*** jonathanmaw has quit IRC17:28
paulsherwoodanyways happy /win 317:59
paulsherwoodbah... wrong window, wrong keys, wrong brain17:59
*** bashrc_ has quit IRC18:01
*** toscalix has quit IRC18:08
*** CTtpollard has quit IRC18:17
*** edcragg has quit IRC18:30
*** locallycompact has quit IRC18:37
* persia thought we weren't using the word "morphology" anymore18:44
rdaleit's difficult to fix that when the main build tool is still called 'morph' and the definitions end in .morph18:50
paulsherwoodheh. ybd handles .def files18:53
rdale.def is ok, or .baserock or .rock might be good too18:55
rdale.rock is probably 'too surprising'18:56
paulsherwoodwe settled on 'definitions' a long time ago. i didn't propose this change because there have been other more pressing matters18:58
*** locallycompact has joined #baserock19:28
*** edcragg has joined #baserock19:56
*** edcragg has quit IRC20:05
*** gtristan has joined #baserock23:06
*** rdale has quit IRC23:11
*** gtristan has quit IRC23:34

Generated by 2.15.3 by Marius Gedminas - find it at!