IRC logs for #baserock for Tuesday, 2016-02-16

*** zoli_ has joined #baserock00:01
*** tiagogomes_ has quit IRC00:15
*** edcragg has quit IRC00:21
*** tiagogomes_ has joined #baserock00:32
*** locallycompact has quit IRC00:34
*** gtristan has joined #baserock02:12
*** CTtpollard has quit IRC08:10
*** CTtpollard has joined #baserock08:23
*** bruce_ has joined #baserock08:26
*** gtristan has quit IRC08:34
paulsherwoodactually, this exposes an interesting problem in the concourse pipeline we are testing08:54
paulsherwoodthe definitions changes have made the pipeline itself obsolete :)08:55 was complaining too09:01
paulsherwoodis it just that lorry hadn't done that repo yet?09:02
paulsherwoodit seems to be there now09:02
jjardonOh, submodules; patch coming09:03
*** bashrc has joined #baserock09:06
*** edcragg has joined #baserock09:24
*** gtristan has joined #baserock09:25
paulsherwoodjjardon: it seems to be working now09:27
paulsherwoodbut yes, there are submodules to fix09:27
*** jonathanmaw has joined #baserock09:30
*** franred has joined #baserock09:36
*** rdale has joined #baserock09:37
*** toscalix has joined #baserock09:43
pedroalvarezwow, that DEFAULT's change will rebuild the world :)09:51
pedroalvarezI wonder if it has been tested though09:51
tiagogomes_sounds like another bug, unless the split rules changed09:56
*** ssam2 has joined #baserock10:09
*** ChanServ sets mode: +v ssam210:09
*** rdale has quit IRC10:15
*** rdale has joined #baserock10:16
*** edcragg has quit IRC10:19
jjardonpedroalvarez: I was about to ask that: it didnt rebuild the world here, so not sure that was a bug or not10:19
pedroalvarezbut it should almost rebuild everything, right?10:20
franredIMO, it should rebuild all the packages which use autotools10:20
*** locallycompact has joined #baserock10:21
*** CTtpollard has quit IRC10:24
*** CTtpollard has joined #baserock10:28
*** jonathanmaw has quit IRC10:30
*** paulw has joined #baserock10:30
*** jonathanmaw has joined #baserock10:32
*** edcragg has joined #baserock10:36
pedroalvarezfranred: yup, and since zlib has that build system, and is in build-essentials..10:46
pedroalvarezrebuild the world it is10:46
paulsherwoodis it merged?10:46
jjardonHow I tell morph/ybd to ignore previous caches and do a full build?10:50
paulsherwoodjjardon: mv artifacts dir?10:51
ssam2if you change a build-system in DEFAULTS then it should automatically rebuild anything that uses that build-system10:56
pedroalvarezjjardon: does your question mean that either morph/ybd are not rebuilding anything with your patch in?10:56
ssam2because the build commands are part of the cache key10:56
ssam2so if the build commands change, the cache key changes10:57
* ssam2 congratulates jjardon for misspelling the name of the 'spell-checking' stratum in
jjardonpedroalvarez: no, they are not rebuilding the stuff11:00
paulsherwoodbug, then11:01
paulsherwoodwhere is your branch, jjardon?11:01
pedroalvarez`git fetch refs/changes/18/1818/1 && git checkout FETCH_HEAD`11:01
pedroalvarezthere are checkout, cherry-pick, etc buttons in gerrit11:01
jjardonpaulsherwood: very useful to see what the commands are actually doing, btw11:02
pedroalvarezI can say that Morph doesn't have that bug11:04
pedroalvarezand that the patch doesn't work11:04
ssam2hmm.. good catch11:05
ssam2however I think the problem is that ZLib isn't autotools11:05
ssam2so we should probably stop pretending that it is11:05
* richard_maw pointed this out in his review of the patch11:05
*** tiagogomes_ has quit IRC11:06
*** jonathanmaw has quit IRC11:09
pedroalvarezyou did :)11:09
jjardonok, lets fix the incorrect chunks then!11:09
paulsherwoodand ybd :/11:09
jjardonpedroalvarez: sorry, I though I tried this with morph but seems I only used ybd11:10
pedroalvarezno worries, I just wanted to double check11:11
*** jonathanmaw has joined #baserock11:12
*** CTtpollard has quit IRC11:18
*** cosm has joined #baserock11:20
*** tiagogomes_ has joined #baserock11:21
locallycompactI'm looking at morph and I can't figure out how it handles nested submodules. update_submodules(app, repo_dir) doesn't return anything or call itself on its submodules.11:21
*** CTtpollard has joined #baserock11:21
ssam2maybe it doesn't11:23
richard_mawlocallycompact: the Submodules.load() function does the depth walking11:24
ssam2oh yeah11:24
locallycompactdoes it? how?11:25
richard_mawah, I misparsed Submodule as Submodules, hmm11:26
richard_mawah, you need morphlib.builder.extract_sources11:27
*** cosm has quit IRC11:33
ssam2as usual Morph has two implementations of the same thing :-)11:37
*** paulw has quit IRC11:38
locallycompactupdate_submodules is only called from plugins11:43
locallycompactwhat do those do?11:43
pedroalvarez"": you give it the name of a chunk and a path, and it will clone the repo in that path11:44
locallycompactextract_sources doesn't seem to make any calls to RepoAliasResolver though.11:47
*** cosm has joined #baserock11:47
richard_mawlocallycompact: it gets the repo object from the LocalRepoCache.get_repo() method11:52
richard_mawwhich uses the resolver to fill in the url11:52
ssam2my favourite thing about the get_repo_plugin is that there is an expand_repo_plugin that does the same thing :-)11:56
ssam2which is largely my fault11:56
tiagogomes_Both need to die12:00
jjardonok to merge the zlib fix?
* richard_maw reviews12:03
locallycompactI'm unsure how to make this change without breaking something12:05
locallycompactDoes app carry a copy of definitions I can query?12:07
richard_mawit didn't the last time I looked12:08
richard_maw(partially because we don't have a neat data structure describing definitions)12:08
richard_maw(also because eww, singletons)12:08
franredjjardon, looks ok to me, I haven't tested, though12:13
ssam2locallycompact: there are two code paths that are affected by this I think12:16
*** cosm has quit IRC12:16
ssam2locallycompact: one codepath is fairly easy, it uses the DefinitionsRepo class12:16
ssam2actually I'm not sure that has anything to do with submodules anyway12:17
ssam2the other code path goes through the SourceResolver module12:18
ssam2which creates a bunch of Source objects12:18
ssam2which can carry any data, e.g. what submodules were specified in the stratum .morph file12:18
*** cosm has joined #baserock12:28
*** edcragg has quit IRC12:38
*** edcragg has joined #baserock12:40
*** cosm has quit IRC13:07
jjardonHey ssam2 , seems the '>-' trick doesn't work as expected; I still have to use '\'13:25
*** gtristan has quit IRC13:28
richard_mawjjardon: what's your code?13:30
richard_mawjjardon: weird, you need to align your --prefix and --sysconfdir to the same column as the ./configure13:36
jjardonrichard_maw: uh, is this the intended behaviour in the yalm spec? or is it a bug in the parser?13:37
richard_mawdon't know13:37
richard_mawjjardon: added a comment of the other style13:39
jjardonrichard_maw: ah, great, thanks13:41
*** ctbruce has joined #baserock13:51
*** bruce_ has quit IRC13:51
jjardonrichard_maw: btw, if you have a moment, I think this is what you wanted to happen?
*** ctbruce has quit IRC14:06
ssam2jjardon: that's annoying!14:09
ssam2jjardon: seems the normal mode works though, good14:09
jjardonmmm, I have a problem here; I need to add libva-intel-driver to be able to use the libva in intel platforms: Id put this in the bsp stratum, but this will make that stratum depend on libdrm, wayland and x11. O ther solution is to put it in multimedia-gstreamer; but this will break the build in ARM; any other suggestion?14:10
jjardonI guess I can always create a completely different stratum for this ...14:10
paulsherwoodnew stratum, arch-specific14:11
pedroalvarez+1 to new stratum, arch-specific14:14
jjardonwhat do you mean by arch-specific?14:17
paulsherwoodonly to be used in systems where arch: x86_64, so by convention include x86_64 in the name14:18
pedroalvarezexactly, that would be enougg14:19
jjardonah, yeah sure; Im suing x86 only actually14:19
* paulsherwood is surprised that bsp*.morph don't actually specify arch:14:19
pedroalvarezalmost all of them14:20
*** ctbruce has joined #baserock14:20
pedroalvarezwell, in the name, if that's what you mean14:21
paulsherwoodi mean surprising they don't have an arch: field, as systems do14:21
*** toscalix has quit IRC14:27
*** toscalix has joined #baserock14:29
jjardonvulkan is here:
jjardonpaulsherwood: pedroalvarez I think I found a better approach in the end:
*** bashrc has quit IRC15:03
pedroalvarezthat's what paulsherwood suggested I believe15:04
*** bashrc has joined #baserock15:05
jjardonpedroalvarez: I've moved all the libva stuff there15:05
pedroalvarezjjardon: the approach works for me15:10
*** cosm has joined #baserock15:17
*** gtristan has joined #baserock16:18
*** cosm has quit IRC16:30
paulsherwoodnode is broken in master for ybd?16:30
pedroalvarez"configure: error: no such option: --sysconfdir"16:31
pedroalvarezjjardon: ^16:31
pedroalvarezs/ for ybd?//16:32
paulsherwoodsame in morph?16:32
jjardonpaulsherwood: node doesn't use autotools, the node morph file needs to be fixed16:35
jjardonpaulsherwood: can you try the patch from: git fetch refs/changes/35/1835/1 && git checkout FETCH_HEAD16:42
*** cosm has joined #baserock16:44
* pedroalvarez +1s
* richard_maw wonders if we need an autotools-like build-system which just takes --prefix16:47
*** ctbruce has quit IRC16:54
paulsherwoodjjardon: +116:55
paulsherwoodworks, merged16:55
ssam2richard_maw: we could maybe consider 'autotools-like' things to use the 'build-api'  (
ssam2interestingly the build API says "The configure script MUST either ignore unknown options that start with --enable- and --disable-, or accept an argument --help which prints valid options."17:05
ssam2which means that a configure script that blows up on unknown arguments is still valid according to the build-api. not sure if that's useful or not17:06
* richard_maw isn't sure we can guarantee that --libdir is accepted, so wouldn't use build-api for node or zlib17:07
jjardonany reason why gdp is not using stock weston?17:14
jjardonpedroalvarez: maybe you know?17:14
paulsherwoodradiofree probably knows more?17:14
richard_mawwouldn't surprise me to discover that it's locked to an old version because they haven't updated the IVI-shell plugins17:14
jjardonat least agl uses stock weston; maybe our definitions are old, as the gdp specific patches has been already upstreamed?17:15
ratmiceno mouse around, so hopefully these urls work, but they are the origin of things like libdir...17:15
pedroalvarezjjardon: gdp needs to update some components before they can move to weston 1.9.0 without patches17:17
ratmiceso, to be compatible with gnu coding standards configure *should* accept libdir17:17
pedroalvarezjjardon: I've heard that will happen soon17:17
pedroalvarezI tried to implement the changes needed but failed17:18
jjardonpedroalvarez: ah rigth, thanks for the heads up17:18
pedroalvarezI  have already tested genivi baseline with weston 1.9.0 and works nice17:19
pedroalvarezbut didn't want to break gdp17:19
pedroalvarezso I'll upstream the changes once it's possible17:19
jjardonpedroalvarez: what weston version is being used in gdp?17:20
ratmiceso, its weird that build-api and gnu coding standards differ here between should and must...17:21
jjardonuh, 1.6+patches; agl is using 1.817:24
paulsherwoodjjardon: plase go ahead and update gdp if you can17:25
jjardonI guess the layer_manager is the component that is holding the upgrade?17:26
pedroalvarezjjardon: afaict, gdp-hmi is not compatible with the required version of wayland-ivi-extension17:28
pedroalvarezjjardon: I tried this changes, and managed to compile it
pedroalvarezbut everything in the GUI was wrong when running it17:30
jjardonmmm, that repo looks quite dead; last commit in master almost a year ago17:33
* locallycompact identifies a grand total of one jillion thing still pulling submodules from github17:33
jjardonpedroalvarez: what is this component suppose to do?17:33
richard_mawjjardon: I think that's the example GUI that jonathanmaw was showing up17:33
*** ssam2 has quit IRC17:41
*** ssam2 has joined #baserock17:42
*** ChanServ sets mode: +v ssam217:42
pedroalvarezyup, that's the shiny launcher of GDP :P17:46
pedroalvarezTIL: hmi stands for human machine interface17:54
richard_mawbecause automotive appears to want its own acronyms for everything17:56
jmacs"HMI" is broader than automotive17:57
*** jonathanmaw has quit IRC17:59
*** bashrc has quit IRC18:05
*** edcragg has quit IRC18:20
gtristanfwiw... the bug reported here:
gtristanis actually a bug discussed here:
gtristanand can be worked around by using the gmp library on the gcc infrastructure site (instead of bleeding edge 6.1)18:22
*** rdale has quit IRC18:23
*** tiagogomes_ has quit IRC18:30
jjardonHi, is it ok to introduce a generic weston system with qt (and qtwayland)?
*** toscalix has quit IRC18:44
*** ssam2 has quit IRC18:56
pedroalvarezgtristan: good to know! Id like to remove our workaround19:10
*** bruce_ has joined #baserock19:59
*** toscalix has joined #baserock20:02
*** toscalix has quit IRC20:06
*** bruce_ has quit IRC21:28
*** edcragg has joined #baserock23:07
*** trn has quit IRC23:37
*** trn has joined #baserock23:39

Generated by 2.15.3 by Marius Gedminas - find it at!