IRC logs for #baserock for Thursday, 2016-02-25

*** gtristan has joined #baserock06:41
*** toscalix has joined #baserock08:54
*** fay_ has joined #baserock08:59
*** locallycompact has joined #baserock09:14
*** ctbruce has joined #baserock09:40
*** jonathanmaw has joined #baserock09:45
*** ctbruce has quit IRC09:53
*** ssam2 has joined #baserock10:01
*** ChanServ sets mode: +v ssam210:01
*** franred has joined #baserock10:02
*** edcragg has joined #baserock10:12
franredssam2, is the only patch of this series which stops for the series to be merged. Are you still against it after pedroalvarez explanations? AFAICR, this patch-series fixes the installation in jetson boards.10:16
ssam2franred: i'm looking at it now10:17
ssam2i'm still reluctant to merge code to the deployment process that i can't understand10:18
ssam2am trying to add some comments to make it clear what's going on10:18
ssam2pedro *believes* to understand it, and nobody else has commented at all about how they think it works, so i'm not confident it's maintainable as-is10:18
pedroalvarezyes, and I've never tested it10:19
ssam2will try to submit a new version with more comments and docstrings10:19
pedroalvarezand I'd like to before it's merged10:19
paulsherwoodradiofree: ^^ ?10:19
radiofreeeven i don't understand how it works10:27
radiofreeto be serious, i'll comment on it later10:27
radiofreebasically you maintain btrfs filesystem structure on the boot partition to help system-version-manager out10:27
radiofreesystem-version-manager expects kernel and dtb to be in /systems/version/kernel and /system/version/dtb10:28
paulsherwoodif radiofree doesn't understand it, what hope is there for the rest of us?10:28
radiofreeso the layout doesn't magically change, you're maintaining the layout on the boot partition10:28
radiofreesystem-version-manager already supports this, the jetson flashing script does the same thing (though manually using cp....), this just allows you to build a single image you can dd directly10:30
radiofreetested on arm and x86 btw10:31
radiofreex86 was "normal" single partition image, deploy/upgrade still works, and deploy/upgrade to x86 partitioned system10:31
* radiofree will respond in gerrit at some point10:33
radiofreevery busy atm10:33
paulsherwoodradiofree: thanks a lot for the info10:33
ssam2ok, ta. will try and sprinkle some comments and docstrings into the patch10:36
*** fay_ is now known as faybrocklebank10:38
*** gtristan has quit IRC10:41
*** ctbruce has joined #baserock10:54
*** gtristan has joined #baserock11:09
pedroalvarezwooo! genivi systems are being upgraded to weston1.9 !11:12
radiofreeso have they ported the gdp-hmi-controller to use the new input api?11:26
pedroalvarezradiofree: jonathanmaw has already provided a version that works11:29
pedroalvarezI'll upgrade to the new gdp-hmi as soon as he thinks it's finished11:30
paulsherwoodhmmm .... as of 227b22d91893417b1c3633a8372f63a4489a68d8 i'm seeing
paulsherwoodjjardon: ^^12:49
pedroalvarezthat's probably caused by 84d3b08d9caf6183a2d1b677c335b00123638a4612:51
pedroalvarezI thought the patch was tested anyway..12:52
paulsherwoodwell, i'm happy to be part of testing :)12:53
paulsherwoodam hoping for pre-merge ci soon12:54
pedroalvarezcould we add the word "fast" to your hopes? :)12:55
* SotK wonders what the plan for pre-merge ci is these days12:55
* richard_maw assumes concourse12:55
jjardonpaulsherwood: sorry, feel free to revert it (this is one of the reason I wanted a way to build the gnome strata without python2 at all)12:56
jjardonpaulsherwood: or wait to the afternoon and you will get a fix12:57
jjardonpedroalvarez: 1.9 is old already :) I will send patches to upgrade stock wayland/weston to 1.10 today12:59
pedroalvarezI don't think I will try to upgrade just now13:02
pedroalvarezI was planning to merge weston and weston-genivi somehow.. but it looks like weston will be always ahead13:02
jjardonyeah, there are some new dependencies as well: wayland-protocols and now libinput depend on libwacom13:03
*** richard_maw has left #baserock13:03
jjardonpaulsherwood: if you upload you current cache you will get a fix sooner  :)13:09
paulsherwoodjjardon: i'll wait for fixes :)13:12
paulsherwoodjjardon: i always do :)13:25
paulsherwood(upload cache)13:26
radiofreeCan't we just merge Weston and weston-genivi now13:44
*** gtristan has quit IRC13:44
pedroalvarezradiofree:  right now yes, but if we upgrade weston I don't know if genivi sytems will continue to work13:50
paulsherwoodpedroalvarez: better to ask for forgiveness than wait for permission :)13:53
pedroalvarezoh well :)13:55
radiofreepedroalvarez: wayland-ivi-extension is pretty pro-upstream weston, i don't see it being a problem13:55
radiofreenicer to merge both and upgrade to 1.1013:55
pedroalvarezI could test a patch for that today13:56
*** gtristan has joined #baserock13:56
pedroalvareztomorrow I'll have less hardware for testing :(13:56
* pedroalvarez continues investigating:13:58
pedroalvarez  File "/src/definitions/extensions/", line 475, in create_versioned_layout13:58
pedroalvarez    version_label, os.path.join(mountpoint, 'systems', 'default'))13:58
pedroalvarezOSError: [Errno 1] Operation not permitted13:58
pedroalvarezos.symlink failing13:59
pedroalvarezcontext: testing  extlinux-partition-support patch14:00
*** faybrocklebank has quit IRC14:04
radiofreeoperation not permitted14:12
pedroalvarezI can't see the problem14:12
radiofreeis this with morph?14:13
pedroalvarezit shouldn't matter, i think14:13
pedroalvareztrying this partitioning:
* radiofree only tested this with ybd, but it shouldnt matter?14:14
pedroalvarezthe code used for the deployments should be the same14:17
*** faybrocklebank has joined #baserock14:19
pedroalvarezI don't understand... version_root folder is created just before trying to create the default symlink14:21
pedroalvarezbut creating the symlink fails14:22
franredpedroalvarez, permission problems?14:23
*** brlogger has joined #baserock14:38
*** bwh has joined #baserock14:39
pedroalvarezI don't understand what's going on... :/14:48
paulsherwoodpedroalvarez: welcome to my world :)14:55
tiagogomes_can I reproduce it easily?14:57
locallycompactpaulsherwood, please give me a clue here.15:06
*** ctbruce has quit IRC15:06
*** ctbruce has joined #baserock15:06
radiofreedoes it work in ybd?15:06
locallycompactgives back15:06
radiofreepedroalvarez: ^15:06
radiofreepaulsherwood: can you try it with ybd? I don't have an x86 vm to hand15:07
paulsherwoodlocallycompact: about what?15:07
paulsherwoodradiofree: what am i trying precisely? does it need btrfs?15:07
locallycompactwhere the side effect is15:08
pedroalvareztiagogomes_: branch baserock/pedroalvarez/testing-partitioning15:08
radiofreewas that for paulsherwood?15:09
pedroalvareza base system should be fast to build using cache15:09
radiofreethere's your issue15:09
pedroalvarezradiofree: haven't tested in ybd15:09
radiofreewhy vfat?15:09
pedroalvarezwhy not!?!?!15:10
paulsherwoodlocallycompact: is str(path) the same as path[2:] ? what about leading / ?15:10
* pedroalvarez <- facepalms15:10
pedroalvarezradiofree: I copied-pasted from another example, missed that bit15:10
locallycompactos.walk gives back paths prepended with ./. fs.walkfiles gives back unicode paths without, that's why the 2: is not there and the str() is15:11
paulsherwoodlocallycompact: i don't like the raise, by the way15:12
* paulsherwood continues to think15:12
locallycompactIt was hiding errors from me15:13
paulsherwoodwhy not sort() ?15:13
locallycompactno reason15:15
locallycompactit doesn't quite have the same meaning on that generator though15:15
paulsherwoodi'd be surprised if there wasn't a reason for it originally15:15
paulsherwoodjjardon: still getting the zeitgeist error15:16
*** tiagogomes___ has joined #baserock15:25
*** Zara_ has joined #baserock15:26
*** tiagogomes_ has quit IRC15:26
*** Zara has quit IRC15:27
*** gary_perkins has quit IRC15:27
*** jonathanmaw has quit IRC15:27
*** ssam2 has quit IRC15:27
*** jonathanmaw has joined #baserock15:27
*** gary_perkins has joined #baserock15:27
*** ssam2 has joined #baserock15:27
*** ChanServ sets mode: +v ssam215:27
*** Zara_ is now known as Zara15:30
*** richard_maw has joined #baserock15:38
*** vgrade has joined #baserock15:43
jjardonpaulsherwood: ^ sounds familiar?15:44
rdaleit looks like the docutils stratum is called 'docutils.morph'15:45
jjardonpaulsherwood: it fails only in master, latest tag builds15:46
* paulsherwood blames locallycompact, then :)15:47
paulsherwoodjjardon: you mean master ybd, or definitions?15:47
jjardonpaulsherwood: ybd15:47
jjardon(Im using master of definitions as well)15:47
radiofreepedroalvarez: it would be possible to support vfat i guess15:50
paulsherwoodjjardon: ack, please stick with tag15:50
radiofreeyou'd have to copy the kernel and dtb out of "factory" into a folder called default, rather than symlink15:51
pedroalvarezradiofree: to be honest I'm not worried about that right now15:51
radiofreeand then make sure system-version-manager did that (instead of updating the symlink, copy out the kernel and dtb from VERSION)15:51
* radiofree also doesn't care about supporting fat3215:51
pedroalvarezmaybe just raising an error there saying "symlinks in fat? u crazy?" is enough15:52
radiofreeit's kind of pointless without that mechanism though15:53
radiofreethe whole point of this (in my opinion) is to allow upgrades to systems where the bootloader can't read from btrfs15:53
pedroalvarezbut can read from ext15:53
radiofreeor fat15:54
radiofreeif we're not supporting fat, we should error if /boot is fat15:54
radiofreethe only benefit of using fat32, that i can think of, is that you could mount the boot partition on a windows laptop and fix things (and/or deploy testing kernels manually)... but i don't care about that (i can mount ext fine on linux)15:56
jjardonpaulsherwood: My understanding is that this should not be needed because transitive dependencies but, can you try this?15:56
*** ctbruce has quit IRC15:58
*** ctbruce has joined #baserock15:59
pedroalvarezpartitoned image upgrade failed, and i don't have logs16:01
pedroalvarezand no brain left after debugging why symlinks didn't work on fat...16:02
pedroalvarezbrain is back, base system doesn't have python 2, system-version-manager fails16:03
edcraggradiofree: hmm, i agree, except i have seen one case where the primary bootloader expects a fat partition to read uboot and kernel from16:03
radiofreedamn you whoever removed python2 from base system!16:05
richard_mawbut python3 is the future!!1116:06
*** richard_maw has left #baserock16:06
CTtpollardI have a feeling the minnowboard expects boot to be fat3216:07
radiofreeok so it sounds like fat32 support is needed16:07
rjekis it EFI?16:09
*** richard_maw has joined #baserock16:10
CTtpollardI'm just double checking now16:10
ssam2it should be simple to make system-version-manager work with either python16:11
ssam2are you laughing as a way of motivating me to do it? it's really not so hard for simple programs16:12
ssam2system-version-manager doesn't depend on cliapp or any such thing16:12
pedroalvarezIt's my fault for writting that code..16:12
richard_mawssam2: I spent a couple of hours trying before I had to get on with project work, our assumption that strings are sequences of bytes (and they are, since we only really care about file paths) is pretty ingrained16:13
ssam2cool, do you have a branch somewhere?16:13
* pedroalvarez upgrades again, hoping that the upgrade uses system-version-manager from the new system16:13
richard_mawno, it was coding by proxy with radiofree16:13
richard_mawchanging our strings to be b'foo' got much of the way there, but the last bit before I gave up was mkdtemp, which I have since learned how to make it give a proper bytes string16:14
radiofreethere were multiple other failures depending on what you were using in system-version-manager16:15
radiofreee.g "list" or "set-default" etc..16:15
richard_mawtd = tempfile.mkdtemp(dir=tempfile.gettempdirb())16:17
richard_mawnote: gettempdirb() is a comparatively modern invention (only python 3.5!!!1)16:18
CTtpollardradiofree, rjek: getting conflicting info, but from my local image I have here boot is fat16 EFI16:21
CTtpollardI've seen OpenWRT images for it use fat32 though16:22
CTtpollardso yeh, variants of vfat for minnowboard boot partitions16:25
ssam2actually, if updating system-version-manager to work with Python3 is non-trivial, rewriting it in C might be a better use of effort16:27
ssam2the Python implementation was never intended to be permanent16:27
radiofreewhat is the bootloader on the minnowboard?16:28
CTtpollardradiofree: uefi & grub16:30
pedroalvarezno luck with python2 either16:38
pedroalvarezand I'm really tired of testing this patch  :/16:38
*** ctbruce has quit IRC16:51
* ssam2 creates!/story/7816:51
ssam2for rewriting system-version-manager in C16:51
ssam2rdale: the more I use json-ld the more I like it17:14
rdalehave you been using it then?17:14
ssam2I thought this blog post by the author was very good:
ssam2yes, although not directly for Baserock stuff17:15
rdaleright, he does excellent videos17:15
rdaleyour owl grammar converts to json-ld with the rdflib json-ld plugin17:15
ssam2i've been writing json-ld documents as yaml though :-)17:16
ssam2which isn't ideal because you have to quote keys that start with @17:16
ssam2but it's more ideal than writing JSON and having to remember where I should and shouldn't put a comma17:16
pedroalvarezargh! BOOT_DEVICE wasn't set!17:16
rdaleyou can alias keywords with '@'17:16
ssam2using YAML aliases? is that easier than quoting them?17:17
ssam2pedroalvarez: sounds like there's another warning missing !17:17
rdalejson-ld alias's in the context17:17
ssam2oh, right, that's cool17:17
ssam2every time I try and do something I end up thinking "oh, cool, they thought of that"17:18
ssam2and I especially like: "Many W3C specifications are so cryptic that they require the sacrifice of your sanity and a secret W3C decoder ring to read. I never understood why these documents were so difficult to read, and after years of study on the matter, I think I found the answer. It turns out that most specification editors are just crap at writing. "17:18
rdaleyes, it's simple but allows you to do complex things which is difficult to design17:20
pedroalvarezI think, if BOOT_DEVICE is not set, used to not fail17:21
pedroalvarezbut yes, this needs a warning or something17:21
pedroalvarezbut, yay! patches work17:23
franredpedroalvarez, :)17:25
*** locallycompact has quit IRC17:26
*** franred has quit IRC17:26
*** locallycompact has joined #baserock17:29
*** locallycompact has quit IRC17:35
* pedroalvarez merges things17:43
*** jonathanmaw has quit IRC17:48
pedroalvarezjjardon: thanks for reviewing
pedroalvarezit raises an error now17:52
*** CTtpollard has quit IRC17:56
*** edcragg has quit IRC18:15
*** rdale has quit IRC18:29
*** ssam2 has quit IRC18:29
*** toscalix has quit IRC19:11

Generated by 2.15.3 by Marius Gedminas - find it at!