IRC logs for #baserock for Wednesday, 2014-10-01

paulsherwoodpedroalvarez: yes. that's fine, i gave +1 on the list06:05
paulsherwoodjjardon: yes, that's what i did. it starts, but screen is black (VM on MacBook)06:05
paulsherwoodbut maybe i did something weird, checking.06:06
paulsherwoodsame result06:18
SotKis this using a recent version of weston ooi?07:01
SotK(I only ask because I tried installing a recent version of weston over the top of what I had installed already and got the same outcome)#07:02
radiofreeadd --log=weston.log, then check the output of that07:07
*** dutch [] has joined #baserock07:15
SotKI guess that "[07:14:05.117] Starting with no config file." has something to do with it?07:21
radiofreeno that's fine07:23
radiofreeodd, everything looks ok, you just get a blank screen with this?07:23
radiofreewhen you boot do you see tux?07:23
radiofreedo you use vga=788 on your command line?07:28
radiofreecat /proc/cmdline07:28
SotKI do07:28
SotKhmm... it works now I've rebooted07:29
radiofreeout of interest did you get the log from it?07:29
SotKyep, one sec07:29
radiofreehow odd... it's the same07:31
radiofreecould be an issue with the vm then, is this virtualbox?07:32
SotKvery strange07:32
SotKI'm using kvm/virt-manager07:32
jjardonpaulsherwood: mmm, odd. I will try again here. Ive only tested in x86, are you building for other architecture?08:12
*** tiagogomes [] has joined #baserock08:14
*** franred [] has joined #baserock08:17
*** fay_ [] has quit [Ping timeout: 258 seconds]08:25
*** jonathanmaw [] has joined #baserock08:34
*** fay_ [] has joined #baserock08:36
paulsherwoodnope, x8608:42
paulsherwoodi can try on Jetson later, though08:43
paulsherwoodi get blank screen too. maybe something similar between my issue and SotKs?08:43
paulsherwoodi'm using virtualbox08:44
* straycat shudders08:49
pedroalvarezpatch to fix fstab when doing an upgrade sent08:54
*** ssam2 [] has joined #baserock08:54
radiofreejjardon: (VM on MacBook)08:54
radiofreepaulsherwood: jetson should work with fbdev backend08:55
radiofreevioleta is currently working on getting it working with the drm backend though08:55
jjardonpaulsherwood: just rebuild and tested again: it works, but Im using qemu here09:04
pedroalvarezNormaly when I attach a disk to a system running linux, it appears under /dev, normally /dev/sda, /dev/vda09:23
pedroalvarezI wonder if this regex is enough: "^/dev/([a-zA-Z0-9]+)$"09:24
pedroalvarezor the new disk can  appear under another folder in /dev09:24
pedroalvareze.g. /dev/foo/sda09:24
richard_mawyou can sometimes see /dev/block/..., but I don't think that is a standard location, I just recall seeing that on Android09:27
jjardonpaulsherwood: FYI, seems to work here in virtualbox as well09:27
pedroalvarezrichard_maw: ta09:28
richard_mawthat regular expression will match things which certainly aren't disk devices though09:29
*** liw-orc [] has quit [Quit: Coyote finally caught me]09:30
richard_mawI'd be tempted to just use ^/dev/((sd|vd|mmcblk)[a-z0-9]+)09:31
pedroalvarezthat also makes sense09:32
*** fay_ [] has quit [Ping timeout: 244 seconds]09:33
richard_mawthe only exceptions to ones matching those which are proper devices I know of are /dev/loop* and device mapper09:37
richard_mawI used to see /dev/hd*, but I think that may have been pre-udev09:37
richard_mawan /dev/sg* may be possible, but I've only seen them be CDs09:38
*** fay_ [] has joined #baserock09:39
straycathd* is for IDE devices is it not?09:40
rjekhd* is very old.09:41
rjekEven IDE devices will come up as sd* these days, due to the unified ATAPI layer.09:41
rjekLVM and crypto will appear basically anywhere.09:42
rjekpedroalvarez: What problem are you trying to solve?09:42
rjekIf you're trying to find a list of valid block devices, cat /proc/partitions and then search /dev for the appropriate block device major:minor numbers09:43
pedroalvarezrecognise when a variable is a device09:43
rjekpedroalvarez: Hmm; see if the device is in /dev, if it is extract major and minor device numbers, and see if they're mentioned in /proc/partitions09:44
pedroalvarezrjek: yeah, I'll do something like that. Thanks!09:45
*** franred [] has quit [Quit: Leaving]09:46
*** franred [] has joined #baserock09:49
paulsherwoodjjardon: must be me :)09:51
*** violeta [] has quit [Read error: Connection reset by peer]10:27
*** violeta [] has joined #baserock10:27
*** violeta [] has quit [Client Quit]10:28
ridgerunnerI'm writing a critique of the "Trove Reference Manual" trying to pick up on the things in it that I find confusing/ambiguous/hard-to-understand. I'm writing it in Wiki format, is there a place I can put it for consideration by the Baserock team?10:28
*** violeta [] has joined #baserock10:28
petefothridgerunner: on the wiki? (
franredor the mailing list10:34
*** jonathanmaw [] has quit [Ping timeout: 258 seconds]10:34
ssam2ridgerunner: that may be useful, please keep in mind that said document was originally written as if Trove was a commercial offering and so the tone etc. might seem a bit strange10:35
ssam2but I don't think we have the resources to rewrite the whole thing from scratch right onw, unless you're volunteering :)10:35
ridgerunnerActuall ssam2 the tone is pretty good. I don't think it needs re-writing, just tweaking. I'd be happy to do it but the problem is that I'm picking it up on the things that I feel it doesn't explain, which means that I don't understand it so I couldn't re-write it, IYSWIM.10:37
ridgerunnerWhy don't I bung out what I've written so far and y'all can decide if it's worth continuing with?10:38
pedroalvarezridgerunner: pleae, go ahead :)10:39
ridgerunnerI could hang it in as a link at the top of the page?10:41
*** fay_ [] has quit [Read error: Connection reset by peer]10:42
*** fay_ [] has joined #baserock10:43
petefothridgerunner: or upload, but son't create any links on other pages. Then just email the url for us to look at10:43
ridgerunnerHm. How do I create a page "blind"?10:44
petefothcreate it as a .mdwn file in the root directory (e.g. foo.mdwn). git add, git commit and git push, the the url is
petefothridgerunner: in your clone of the w.b.o repo10:47
ridgerunnerWhich I don't have :-)10:47
paulsherwoodridgerunner: just go to the desired nonexistent url, then click on the ? to create your page10:48
ridgerunnerBrilliant paulsherwood, that works, thanks.10:48
ridgerunnerPage at
ridgerunnerLooking at it, I should probably have copied the page and inlined the comments.10:51
paulsherwoodit's fine as is. thanks10:52
paulsherwoodpedroalvarez: are you going to merge ca-certs? is it sufficiently +1 enabled?10:54
pedroalvarezI got a +1 from you and another from Richard Maw, but I'm not sure if I can take both of them10:55
paulsherwoodwhy not?10:55
paulsherwooddo you fear that either richard_maw or paulsherwood is not a competent reviewer?10:56
pedroalvarezI think both of the reviews are competent reviews, but I would like to have a policy for this10:57
pedroalvarezwhat makes you a competent reviewer?10:57
pedroalvarezwho already is?10:57
pedroalvarezwhen can we merge a patch?10:58
pedroalvarezis anybody allowed to +2 instead of +1?10:58
paulsherwoodthat has been happening, for certain patches where the reviewer felt especially empowered10:58
pedroalvarezIMO I think we have to write something down regarding this10:59
KinnisonTypically I'd only expect to see a +2 from someone who was an upstream contributor and considered by the community to have a strong grip on the codebase10:59
KinnisonHowever others have said +2 in the past11:00
KinnisonI think it's a question of using common sense on this.11:00
KinnisonAt least until upstream can decide a policy11:00
paulsherwoodwe could write something down, but i'd prefer to discuss after we've established gerrit11:00
*** Guest17402 [] has joined #baserock11:00
pedroalvarezI agree11:00
*** mwilliams_ct [] has joined #baserock11:00
paulsherwoodfor the moment, i don't think the current process is dangerous, just there's variable delay11:01
pedroalvarezI'm looking forward to have gerrit11:01
pedroalvarezis not dangerous because we are not too many11:01
paulsherwoodsure. i look forward to the time when we're inundated with patches and reviewers11:01
Guest17402 is now known as gary11:02
*** mike [] has joined #baserock11:02
*** mike [] has quit [Client Quit]11:02
paulsherwoodmeantime, please merge :)11:02
gary is now known as Guest4732511:02
pedroalvarezca-certificates, welcome to baserock!11:02
*** Guest47325 [] has quit ["Sayonara"]11:03
ssam2seems a commenter on LWN has invented Baserock :)
jjardonssam2: hey, about the NetworkManager bracnh, should I send another branch with the patch to fix the issue included or its not needed as we are using a more modern linux-api-headers now?11:16
ssam2jjardon: if the patch is upstream them maybe best to include the new upstream SHA111:21
ssam2if it's not merged to upstream master then maybe don't bother for now, since it should work fine anyway11:21
*** mSher [] has joined #baserock11:23
jjardonssam2: its upstream master and in the latest stable branch nm-9-1011:23
petefothridgerunner: a lot of good comments in your critique of the Trove Manual. Before we make any changes to the document however, I think we need to work out -and explicitly record - exactly the target readership is for the document. (regular readers will knwo this is a regular hobby horse of mine - the 'spec' for a document is a: Who are the intended readers and b: what should those readers get from the document.) It is almost certain t11:27
paulsherwoodregular readers will also note that things can get bogged down in discussion. i expect that someone can tweak the document relatively quickly. maybe i'll do it if nobody beats me to it :)11:28
ridgerunnerpetefoth: I certainly would not argue with that. Knowing your target audience and pitching it at the right level is pretty damned important.11:29
paulsherwoodridgerunner: it used to be. less so these days. mainly we need less words, shorter words, and more pictures/videos.11:30
* petefoth goes for an edit11:30
ridgerunnerI kind of assumed I might be a good candidate. Lots of experience, has used different IDEs, knows little about Baserock.11:30
ridgerunnerpaulsherwood: I would agree with that to get people in, but not for a reference manual.11:30
ssam2jjardon: oh, maybe stick with stable then11:31
paulsherwoodridgerunner: nobody reads the manuals any more. there are no manuals any more11:31
paulsherwoodwe're all on the interwebs11:31
ridgerunnerpaulsherwood: maybe that's why there is so much shit software out there.11:31
petefothridgerunner: probably not - the Trove reference manual was aimed at SysAdmins who would be responsible for maintaining one or morte Trove systems. Such users woule be expected to have areasonable knowledge of what BAserock is, and how itsa moving parts hang together11:32
paulsherwoodridgerunner: perhaps11:32
* petefoth decides *not* to go for an edit after all. Not until we have abetter view of what documents BAserock needs and what should be in them11:33
paulsherwoodpetefoth: why not just fix the common-sense things? before ridgerunner's feedback gets lost in the mists of time?11:34
*** jonathanmaw [] has joined #baserock11:36
petefothpaulsherwood: because 'common sense' is quite hard to tie down. To ensure ridgerunner's feedback doesn't get lost, we could just add a link to his critique in the manual page11:37
petefothpaulsherwood: 'Why not just fix the obvious errors in the code?' 'Because without a spec you don't know that they are errors.'11:38
paulsherwoodpetefoth: some errors are obvious, irrespective of spec11:47
petefothpaulsherwood: indeed. I'm not sure that ridgerunner's critique identifies any such errors in the Trove manual though.11:49
paulsherwoodspecs typically don't say "no race conditions allowed. no off-by-one errors please"11:49
rjekI've seen a contract from a customer at $PREVIOUSEMPLOYER that demanded bug-free code on the first delivery, and that no open-source software was to be used.  The contract's title was "Linux-based network router"11:52
jjardonIs there any way to tell morph not to rebuild the world if I upgrade a component in say, foundation?12:01
richard_mawnot currently, because it uses the sha1 of the git commit's tree in dependency calculation12:02
richard_mawan idea we had a while back was to change it to work on an ABI manifest12:02
jjardonso only rebuild if the ABI changes? That would be _extremely_ useful12:03
* jjardon doesnt want to rebuild everything because there is a new translation in the git module, for example12:04
richard_mawthere's a tool for C and C++ programs that does that based on symbols and headers, but we'd need something that works for everything12:05
paulsherwoodjjardon: for the avoidance of doubt, would you mind pushing a new libdrm branch to gbo, and sticking with the naming convention baserock/jjardon/your-branch ?12:09
* paulsherwood would like to be sure he's building what jjardon intends12:09
jjardonpaulsherwood: sure12:09
jjardonbut Im building f26dd579bf8ef44f4c0878bb01708c4671142b2d, if that helps12:10
paulsherwoodyup. me too12:10
straycatssam2, Would be cool if gentoo folks will started showing up here now12:21
paulsherwoodunlikely, though :)12:23
paulsherwoodradiofree: any ideas what this means?12:23
radiofreepaulsherwood: looks like weston-desktop-shell is segfaulting12:26
radiofreepaulsherwood: do you have a /usr/share/doc/weston/weston.ini file?12:26
jjardonpaulsherwood: seems its my fault: I get the same error as you when compiling with X11 support :/12:28
paulsherwoodi have only ivi-shell-weston.ini12:28
paulsherwoodyay. the bugmagnet strikes again :-)12:29
radiofreepaulsherwood: copy that to /root/.config/weston.ini12:29
radiofreethen try to run again12:29
radiofree(with the log option)12:29
jjardonpaulsherwood: at least if you build a only-wayland system it works ;)12:30
* jjardon investigating12:30
radiofreejjardon: i suspect you're not using the genivi weston version?12:30
jjardonradiofree: I am12:34
paulsherwoodradiofree: same result12:34
paulsherwoodjjardon: seems like this has some residual x11 linkage?12:37
paulsherwooddo you have a newer branch for me to try?12:38
pedroalvarezanybody knows a git:// equivalent for this url?
jjardonpaulsherwood: yeah, seems I forgot to add x-common to the genivi-baseline system, one minute12:39
paulsherwoodi thought the plan here was to drop x?12:40
paulsherwoodpedroalvarez: is x11 still required for genivi baseline in any way?12:40
pedroalvarezthey moved to wayland12:41
Kinnisonpedroalvarez: IIRC bitbucket don't expose git:// but they do do https:// --
jjardonnote that x-common is not x-generic12:43
pedroalvarezKinnison: but is that useful from the lorry-controller point of view?12:43
Kinnisonpedroalvarez: In principle it should work, in practice it may need the ca-certificates first12:44
pedroalvarezbtw, with the next g.b.o upgrade we will have less problems like this12:44
paulsherwoodjjardon: regarding the move for dbus-glib, you have +2, but neither reviewer has tested it, i fear12:44
radiofreejjardon: how are you building weston?12:44
radiofreeare we not doing -disable-xcb?12:45
jjardonradiofree: no changes from what is currently in baserock12:46
franredpedroalvarez, if we need the ca-certificates that may answer our questions12:46
pedroalvarezwe have more https urls that are working in te lorry controller like:
pedroalvarezbut I can't clone them with git clone, even in g.b.o.12:50
ssam2pedroalvarez: GIT_SSL_NO_VERIFY=yes12:50
ssam2in the environment12:50
pedroalvarezssam2: I guess the lorry controller is doing something like that to handle it12:50
SotKpaulsherwood: I just had the exact same log output after (yet again) reinstalling weston-ivi-shell, but rebooting my VM fixed it...12:52
paulsherwoodSotK: rebooting didn't help for me12:53
SotKmost strange12:53
paulsherwoodno, the strange one is yours... at least mine definitely doesn't work.12:54
pedroalvarezquick review to change some lorries from ssh:// to https://
paulsherwoodpedroalvarez: why, though?12:56
paulsherwoodif it works, don't fix it?12:56
pedroalvarezbecause they don't work :(12:57
Kinnisonpedroalvarez: I take it the ssh URLs were not working?12:57
Kinnisonpedroalvarez: the patch looks good to me.  I don't see any JSON errors12:57
straycatIs bitbucket https not ?12:58
Kinnisonstraycat: only for pushing12:59
* straycat nods12:59
radiofreejjardon (and me) is (are) pondering the total anhiliation of x11 from baserock13:13
jjardonso, seems the culcrip is cairo: it build the xcb backend if its in the system. Two possible solutions: build mesa with only wayland support or disable the xcb bakend in cairo unconditinally, any preferences?13:14
paulsherwoodflatmush: anyt thoughts?13:14
paulsherwoodiirc flatmush (and others) wrestled x11 in in the first place13:15
* Kinnison is worried that we'd be shooting ourselves quite hard if we drop X11, OTOH we don't currently build anything much which depends on it13:15
jjardonpaulsherwood: I did it ;)13:15
paulsherwoodah. jjardon always the unsung hero13:15
jjardonyeah, we will need X11 to run XWayland for example13:16
paulsherwoodbut not ncessarily for genivi13:16
radiofreeX11 would be necessary for steamos13:16
jjardonthe propper solution is the ability of morph to change dependencies between chunk/strata depending on configure options13:16
paulsherwoodlet's keep it, it works... .whereas wayland... :)13:17
jjardonok, I will disable the xcb backend from cairo then13:17
jjardonmmm, this will not work; if we disable xcb and xlib backend cairo build not work in X11 ... Is there any plans to add the  ability to morph to set different dependencies depending on configure options?13:28
ssam2jjardon: not plans that will be here any time soon13:29
jjardonwe could go for the wayland-only route for now, as seems genivi is the only system oficially supported by baserock13:29
ssam2am I right in thinking the other option is just supporting Mesa in Wayland ?13:30
ssam2then we'd still have X, but just no 3D in X13:30
ssam2which I think would be better than no X at all13:30
jjardonssam2: yeah, that could work13:41
ridgerunnerDoes th team think I should continue with my critique of the Trove Reference Manual? N.b. I will bear in mind the intended target audience :-)13:43
pedroalvarezA couple of python-packages needed for OpenStack:
franredpedroalvarez, they look ok to me14:25
franred! = 114:25
richard_mawI thought we had everything for OpenStack already. What happened?14:26
pedroalvarezridgerunner: doffm reported that he was missing some repositories due lorry-controller errors14:30
pedroalvarezbut this 2 were never lorried14:31
pedroalvarezI'm waiting to heard from him "yes I need them"14:31
pedroalvarezridgerunner: sorry, that wasn't for you14:32
pedroalvarezrichard_maw: ^^14:32
richard_maw+1 to merge14:33
jonathanmawrichard_maw: hunter214:53
richard_mawwell, that's the distbuild unit tests fixed, such as there are any15:13
richard_mawnow to integrate some yarns to do a proper integration test15:13
ssam2richard_maw: have you fixed the actual problem yet ?15:17
ssam2it's rather bad that master of Morph is still broken15:17
violetawhere are all the variables like "DESTDIR", "PREFIX", etc, defined?15:18
richard_mawssam2: I've made changes relevant to the new morphology structure, but I won't know if that works until I have an integration test15:18
richard_mawvioleta: do you need to know in code where they are defined, or just what they all are?15:18
ssam2richard_maw: OK. might be quicker to just deploy yourself a Mason instance and build something, though15:18
ssam2in fact, just running `morph serialise-artifact` might be enough to check15:19
violetarichard_maw, what they are, how can I define others, etc15:19
richard_mawssam2: serialise-artifact won't help me much beyond what I've already got from the unit tests15:19
richard_mawvioleta: there's no generic way to define extra variables15:20
richard_mawPREFIX was a special case for us wanting to be able to use the same morphologies in the bootstrap, which puts stuff in /tools, and elsewhere15:20
violetarichard_maw, more specifically I want to know what version of the kernel I'm building, or the name is giving to it...15:20
richard_mawwhat do you need the kernel version for?15:21
ssam2richard_maw: as we have learned last week, the unit tests passing does not prove that the code works15:22
violetaI have a problem with "depmod -a" when I'm building a different kernel version than the one my system has15:22
richard_mawssam2: yes, which is why I'm improving the tests >:-|15:22
violetait tries to access /lib/modules/3.10* that is the kernel my system has, but I need it to access the folder that is in /lib/modules, that is 3.17*15:23
violetathe error I'm getting:
juergbivioleta: you can use 'depmod'15:23
violetajuergbi, yes, but that name changes every time I build15:24
juergbiyou somehow need to get that version into a shell variable. i'm not familiar with the baserock kernel build, though15:25
violetarichard_maw, ^^15:25
richard_maw(cd /lib/modules && for version in *; do depmod "$version"; done) perhaps?15:25
richard_mawor `for basedir in /lib/modules/*; do depmod -b "$basedir"; done`15:26
violetait has to be -a, but I'm going to try!15:27
richard_mawthough after that's run you'll probably run into it complaining that it can't write the modules list file where it wants to15:27
richard_mawactually, no, integrations should be fine15:27
*** genii [~quassel@ubuntu/member/genii] has joined #baserock15:39
straycatNo test can prove the code works.15:56
violeta2014-10-01 15:38:41 Build ends successfully16:01
violetathanks richard_maw :'-)16:01
pedroalvarezvioleta: great! 16:04
violetapedroalvarez, :-D16:04
pedroalvarezI think we should fix the other chunk morphology with depmod in the system integration commands16:09
*** pedroalvarez [] has quit [No Ping reply in 180 seconds.]16:16
*** pedroalvarez [] has joined #baserock16:17
Mode #baserock +cnt by wolfe.freenode.net16:17
richard_mawit's a bit weird that we've merged the distbuild code into morph.git, but the cache server stuff hasn't been16:21
richard_mawand you can't run a distbuild without the cache server16:21
ssam2richard_maw: might make sense to move it into morph.git since it's only one file, yeah16:25
ssam2I'd like to have far fewer baserock/baserock/ repos16:25
richard_mawwell, 3 important file16:25
ssam2we have loads at the moment, most of which are useless, and it makes the project look way more complex than it actually is16:25
*** ssam2 [] has quit [Remote host closed the connection]16:32
*** jonathanmaw [] has quit [Quit: Leaving]16:35
*** CTtpollard [] has joined #baserock16:47
*** tiagogomes [] has quit [Ping timeout: 246 seconds]16:50
CTtpollardHello! got a quick question, if I've updated my branch in workspace with morph pull from master from g.b.o , and then go to rebuild a system I've built previously, shouldn't most things be buildable from cache aprt from anychanges to the system-moprh or upstream repo for specific chunks?16:50
richard_mawwe updated a component very low in the dependency graph recently, so everything above it needed rebuilding16:51
CTtpollardah that could be it indeed, well at least It does seem reasonable16:54
CTtpollardty richard_maw16:55
*** dutch_ [~william@] has joined #baserock20:02
*** dutch_ [~william@] has quit [Client Quit]20:03
jjardonHi, can I ask what is the reason to have cmake in core? just rebuilding the kernel now because cmake got updated...20:24
jjardonok, just found it: "The GENIVI components layer_management, DLT_daemon and AudioManager use cmake to build"20:29
jjardonNot sure its a enough strong reason, maybe it would be a good idea to put it in a separate stratum?20:30
richard_mawjjardon: the decision to put it there was at a time when we thought a small number of strata would be sufficient20:33
richard_mawwe could split it out, but until we can specify runtime-depends (my preference is by allowing you to embed a stratum in another stratum)20:34
richard_mawit'll mean the system morphologies will have a _long_ list of strata20:34
jjardonId be more interessed in the ABI checker you mentioned the other day. Or do not rebuild at all but that its more a radical decission20:36
richard_mawit's also something I don't believe to be generally solvable20:38
richard_maw if you're interested20:39
jjardonrichard_maw: you mean rebuilds?20:39
richard_mawjjardon: not directly, I mean creating ABI manifests and depending on those rather than the actual built source20:40
jjardonrichard_maw: what about not rebuild at all? Or at least have the option20:44
richard_mawhm, we could add a mode that doesn't include the dependencies in the cache keys, which ought to do that, but I can't see it working too well20:51
jjardonif not the only possible solution I imagine to avoid rebuilds is that in the end all the strata would have only one chunk20:57
richard_mawa small number perhaps21:15
jjardonrichard_maw: why small? any component can get updates frequently21:43
* pedroalvarez manages to get mosh working in baserock21:46
richard_mawpedroalvarez: woo!21:53
richard_mawin other news of a woo-ish nature, I've got some distbuilds working in the test-suite21:54
richard_mawhm, deploy doesn't currently work, as there are un-cached repositories22:02
richard_mawno, silly me, I was building a different system to the one I tested22:10
richard_mawor maybe I was right in the first place, it failed again22:27
richard_mawit's late, perhaps I forgot I'd commented it out or something22:27
*** genii [~quassel@ubuntu/member/genii] has quit [Remote host closed the connection]23:29

Generated by 2.15.3 by Marius Gedminas - find it at!