IRC logs for #baserock for Thursday, 2014-09-18

persiaI thought that could be set with a configuration extension at deploy time.05:11
* petefoth believes that docs are just another form of software: to make good docs you need good tools and processes06:43
petefothBaserock doesn't yet have those tools and process for documents06:43
persiaWhat tools do you seek?06:45
*** ctgriffiths [] has quit [Quit: No Ping reply in 180 seconds.]06:54
*** ctgriffiths [] has joined #baserock06:56
petefothI was wrong - we have lots of tools. We just don't have a good *process* for doing docs. The starting point is that we don't capture or specify any *requirements* for docs, so if we do write any, we don't have a spec for them, so we don't know what to write - see Commandment #1707:02
*** violeta [] has joined #baserock07:03
petefothThis applies not only to BAserock, but to many software development projects in many organisations07:03
* petefoth dismounts from his 'docs are software too' soapbox07:04
*** tiagogomes [~tiagogome@] has joined #baserock07:23
*** tiagogomes [~tiagogome@] has quit [Client Quit]07:25
*** tiagogomes [~tiagogome@] has joined #baserock07:25
*** franred [] has joined #baserock07:51
*** liw-orc [] has quit [Ping timeout: 246 seconds]07:52
*** liw-orc [] has joined #baserock07:52
*** tiagogomes [~tiagogome@] has quit [Remote host closed the connection]08:01
*** tiagogomes [~tiagogome@] has joined #baserock08:07
*** jonathanmaw [] has joined #baserock08:37
*** franred [] has quit [Remote host closed the connection]08:55
*** dutch [] has joined #baserock08:56
*** franred [] has joined #baserock09:01
*** ssam2 [] has joined #baserock09:07
petefothTo whom would I need to send my ssh key if I wanted to make edits to w.b.o locally and push them, rather than by edit in place?09:18
rjekpetefoth: I suspect liw-orc09:19
petefothrjek: you suspect him of what? (and thanks)09:19
pedroalvarezpetefoth: I think rjek's answer is related to your question09:20
persiaI thought there was no ACL for w.b.o09:30
liw-orcpetefoth, communicate your ssh public key to me via protected work channels, and I'll add it09:34
*** ssam2 [] has quit [Read error: Connection reset by peer]09:39
*** ssam2 [] has joined #baserock09:40
paulsher1oodquick mason question... where does the code that decides what appears onscreen live?09:44
paulsher1oodin other words if i wanted to tweak the ui, what do i need to hack?09:45
richard_mawdefinitions.git/mason I think09:45
paulsher1oodaha :)09:46
paulsher1oodinteresting approach... a shell file cat-ing html :-)09:47
paulsher1oodwhy do we do that so much?09:48
richard_mawshell's a useful language for knocking together prototypes09:48
persialow depdendency count09:48
paulsher1oodi'm not against shell, just wondering why not have the cat-ed content as a file and be done?09:49
persiaSimplicity of editing: it's just one file to change.09:51
liw-orcsometimes also shell variable expansion in the HTML text09:52
paulsher1oodok, thanks, that makes sense09:52
persialiw-orc: At least for mason, it seems to be done with sed, rather then shell variables in several cases09:52
paulsher1oodthere are templating engines for that, but clearly they lead to more dependencies09:53
persiaRight, and dependencies cause integration pain because of cycle time with current baserock tooling09:54
paulsher1oodi think we're mostly through that... i reckon i could use ssam2's tool to put a ruby thing in place in under a day :)09:55
paulsher1oodbut please don't challenge me to prove it :-)09:55
persiaIt's more a matter of needing to pre-define everything, and that we don't have a good development environment, really.09:57
persiaYes, one could get e.g. HAML+Rails working quickly, but not necessarily with the development environment one wants to develop in HAML.09:57
paulsher1oodpersia: i believe a high percentage of the HAML+Rails folk use Mac + Textmate. I use those, and can work on Baserock using them :)09:59
paulsher1oodbut i may be wrong about the 'high' part :)09:59
ssam2I don't see the problem with using 'gem install', 'rvm', 'bundler' etc. in a Baserock VM to achieve a development environment, either10:00
persiaThat's what I used when I was working with HAML+Rails, so you may not be wrong.10:00
paulsher1oodso my trick is to export my src tree via sshfs. it's tolerable10:01
ssam2that's what I do to so I can develop in Fedora10:01
* paulsher1ood feels a few uptodate video tutorials coming on10:01
persiassam2: Except I want vim compiled with more stuff in, tig, non-busybox utilities, and frequently some rendering tools when I do development, which is less easy to achieve.10:01
ssam2persia: yes, to achieve that in Baserock requires a bit more work10:02
persiassam2: Yep.  Hence "current" in my original criticism :)10:02
jmacs will be shut down NOW to make a backup image. This should take a maximum of 30 minutes.12:53
persiaExcellent!  I presume this is a precusor to g.b.o not being down unexpectedly quite so often?13:08
jmacsYes, me and liw-orc wil be upgrading it if this test is successful.13:10
jmacsrsync is currently waving between 30 and 60 minutes to complete, so this may take up to a further hour. Sorry.13:11
straycatAm I supposed to be able to force push to our branchable?13:32
liw-orcstraycat, afaik branchnable doesn't prevent force pushes13:34
petefothstraycat: which branchable? (i.e.
straycatpetefoth, yes wiki.baserock.org13:36
straycatI can delete the remote and create a new one, so it's not really much of a problem, I just expected to be able to force push.13:37
ssam2Kinnison, SotK: do you mind if I merge jjardon's graphics cleanup patch series to master ? I ask in case it would cause pain for the updated genivi baseline work that you are doing13:40
pedroalvarezssam2: You have to wait to do that, g.b.o, is down13:41
ssam2good point :)13:41
ssam2i'll merge it when it is back, unless anyone asks me to wait13:42
liw-orcrsync says 14:10... 11:50... 20:14... time remaining13:46
Kinnisonssam2: I'm okay to rebase on top of jjardon's work afterwards13:47
Kinnisonssam2: I already need to clean up what I did before it's mergeable anyway13:47
SotKssam2: as am I13:48
liw-orc7:20... 11:08... 12:23... 12:38... 08:08... 12:27...13:52
* liw-orc wishes rsync could make up its mind :)13:52
liw-orcthere's a strong trend towards 00:00, luckily13:52
liw-orcalmost there... about 90s remaining14:01 is now back up and running14:04
jmacsssam2: ^14:04
ssam2cool, thanks for notifying me14:05
paulsher1oodKinnison: how did you get on updating kernel etc on jetson?14:07
petefothliw-orc: Kinnison: anyone: If I were interested in being informed - by email or ...? - when any changes were made to w.b.o. would that be starightforward to achieve?14:10
richard_mawI think you can get an RSS feed14:10
petefothrichard_maw: thank you - I will investigate14:11
* paulsher1ood thought readers were dead now14:12
richard_mawpaulsher1ood: I can't see why they would be. There's still servers serving RSS, and clients capable of reading it14:12
paulsher1oodrichard_maw: fair enough14:13
richard_mawI use theoldreader.com14:13
paulsher1oodrichard_maw: in other news, is there a branch for your mega series anywhere?14:13
paulsher1oodi'd like to try it before it lands, if you don't mind?14:14
richard_mawno, as I was on my personal laptop and hadn't gotten around to putting my key for that on git.baserock.org14:14
richard_mawI can set that up and push now though14:14
paulsher1oodmaybe push to somewhere else? gh, gitorious etc?14:14
* paulsher1ood was quite pleased to see how easy it was to try out jjardon's stuff on gitorious14:15
Kinnisonpaulsher1ood: I am currently unwell and away from my Jetson, but I did get a patched 3.17-rc5 with appropriate DTB combo up and it seems to have DRM nodes in /dev14:17
paulsher1oodsrry to hear about the unwellness - but the result sounds exciting :)14:17
KinnisonI'm excited too14:18
KinnisonI also just tried patching smartdevicelink version 3's build system to be more Baserock compatible14:18
* Kinnison has ssh access to his build cluster but without fingers on boards I can't do anything for my test target :-)14:18
paulsher1oodcool. i'm sure the upstream smartdevicelink guys would be happy to adopt anything that's sensible14:19
paulsher1ood(but note they have to build for nonlinux too)14:19
*** genii [~quassel@ubuntu/member/genii] has joined #baserock14:20
KinnisonMostly they use some arguments to find which busybox can't do14:21
Kinnisonso I'm experimenting with what might work14:21
KinnisonI'm getting some confusing results for cmake files, so I wonder if I'm missing something14:27
KinnisonI think they're trying to be over-clever with their Qt finding stuff14:29
* Kinnison reads some web pages14:29
paulsher1oodrichard_maw: is it me?
richard_mawno, but I'm damned surprised that made its way past my testing14:44
paulsher1oodthe bugmagnet strikes again :)14:45
richard_mawah, we must be missing bootstrap build mode coverage, since we have to build in testing mode14:46
* paulsher1ood rises from his chair, arms aloft and does a small victory dance14:46
*** Krin [] has joined #baserock14:49
richard_mawpaulsher1ood: I've pushed baserock/richardmaw-os/tidy-build-logic-v5, and am going to do some test-builds of my own14:50
richard_mawmy only defense of not doing build testing of definitions.git earlier is that my brain was fried by the time I'd finished splitting the patch up14:51
richard_mawand that's more of an explanation than a justification14:51
* Kinnison is undergoing an effort in determining replacement find instructions :-(14:52
paulsher1oodrichard_maw: ?14:53
richard_mawrunning it myself now, thanks14:54
richard_mawKinnison: which command is it using that is missing?14:55
straycatCan we make http the default upstream trove protocol so people don't have to bother creating an upstream trove user if they don't really need one?14:56
Kinnisonrichard_maw: find's -L and -executable14:56
Kinnisonrichard_maw: at least14:57
Kinnisonstraycat: I thought we already had, and if we haven't then yes we should14:57
straycatNot from what I see in trove-setup14:57
richard_mawKinnison: -follow should be equivalent to -L14:57
richard_mawhm, slightly different semantics14:58
straycatansible/roles/trove-setup/tasks/check.yml:- set_fact: UPSTREAM_TROVE_PROTOCOL=ssh14:58
straycatansible/roles/trove-setup/tasks/check.yml:  when: UPSTREAM_TROVE_PROTOCOL is not defined14:58
Kinnisonrichard_maw: slightly different semantics, and busybox doesn't support it in the same place -L is used14:58
pedroalvarezstraycat: yeah, by default is ssh, but yuu can set it to http 14:58
Kinnisonstraycat: I suggest you chat with pedroalvarez about that, he did the ansible trove stuff and he can probably help you rework that14:59
richard_maw`-executable` = `-exec test -x {} \;`14:59
liw-orcstraycat, +2 for making http the default protocol for LC14:59
* straycat nods15:00
straycatKinnison, it's a trivial change15:00
Kinnisonrichard_maw: tbf, for what they're using it for, -perm +0111 is enough15:00
liw-orcKinnison, you need -perm for find, I think, for -executable15:00
Kinnisonliw-orc: they're subtly different :-(15:00
Kinnisonliw-orc: but in this instance, yes -perm +011115:00
pedroalvarezstraycat: but I dont understand why we need the change15:00
straycatalso i have a lot of lag15:00
straycatpedroalvarez, So that people don't need to bother adding a user on the upstreamtrove if they don't need access to private content15:01
pedroalvarezbut is not needed15:01
straycatSeems like the sane default to me15:01
pedroalvarezyou can specify  "UPSTREAM_TROVE_PROTOCOL: http"15:01
liw-orcpedroalvarez, you can, but it's one more thing you have to specify15:02
pedroalvarezwell, now that is possible to create troves without UPSTREAM_TROVE, I guess I don't mind15:03
persiaAnd the default should be freindly for someone starting from scratch: imagine the user has a laptop and is in an internet cafe, or similar.15:03
liw-orcpetefoth, see the "RecentChanges" link near the top of every page on; there are RSS and Atom buttons (orange background) if you want to subscribe (either format should work, I prefer atom)15:09
pedroalvarezI'm not opposed. I'm just saying that is not needed any change to deploy a trove using http to get the content from an upstream trove15:09
persiaIf nothing else, the documentation needs to change: deploying-a-trove talks about getting an upstream account before writing the cluster morphology.15:12
straycatpersia, Working on that :)15:12
* persia sets out a bowl of cream in deep appreciation15:13
* Kinnison gets super-confused by this script15:15
Kinnisonrichard_maw: I may need you to bring your immense shell experience to bear on this so I can work out wtf is going on :-(15:15
* straycat laps up cream and disappears15:18
*** franred [] has quit [Quit: Leaving]16:09
*** tiagogomes [~tiagogome@] has quit [Quit: Leaving]16:31
*** jonathanmaw [] has quit [Quit: Leaving]16:34
richard_mawnew version of the building per-source branch pushed to baserock/richardmaw-os/tidy-build-logic-v616:50
richard_mawIt's merrily up to 110/14916:50
paulsher1oodrichard_maw: weird: /src2/morph # git fetch rmaw baserock/richardmaw-os/tidy-build-logic-v616:52
paulsher1oodfatal: Couldn't find remote ref baserock/richardmaw-os/tidy-build-logic-v616:52
paulsher1oodUnexpected end of command stream16:52
richard_mawthat's… interesting16:54
* richard_maw facepalms16:54
paulsher1oodi'm on a roll :)16:54
richard_mawI had --dry-run set16:54
paulsher1oodha :)16:54
rjekpaulsher1ood: Fig roll?16:54
rjekThat'd explain why we've not had any in a while.16:54
*** genii [~quassel@ubuntu/member/genii] has quit [Read error: Connection reset by peer]16:56
paulsher1oodrichard_maw: looking much better now :)16:57
paulsher1oodrichard_maw: is it now reporting x/y where y is number of chunks to build, not total chunks?17:00
richard_mawI'm tempted to take ttystatus and the code ssam2 and I wrote for processing subprocess output asynchronously, and change the morph TUI to display the last 5 lines of build commands17:00
*** genii [~quassel@ubuntu/member/genii] has joined #baserock17:00
richard_mawpaulsher1ood: y is the number of morphologies in total*17:01
richard_maw*it's actually sources, but that is only different here for strata17:01
paulsher1oodrichard_maw: i killed it, restarted. first time y was 219, now it's 14917:01
richard_mawx is the number of sources currently built17:01
*** Krin [] has quit [Quit: Leaving]17:02
richard_mawwell, you're clearly on a roll today then. I have no idea why it should have shown you 219 for y17:02
paulsher1oodi'm building genivi.17:02
richard_mawI'm building devel17:03
richard_mawdevel shows 14917:03
paulsher1oodah. this will be me, being an idiot17:03
richard_mawless of an idiot than me for not checking whether that bootstrap-mode code did anything sensible17:04
paulsher1oodit's looking good, richard_maw. i'll let it finish and deploy17:04
richard_mawI'm on 149/149 here17:05
* paulsher1ood has a 6 line script that builds, deploys to self, in an idempotent stylee17:05
richard_mawso I'll be able to let you know whether it builds ok before I leave today17:05
paulsher1oodi expect it builds ok - have you touched anything that may affect deploy?17:05
richard_mawnot directly17:06
richard_mawthe build graphing code has changed17:06
richard_mawbut that was exercised during build too17:06
richard_mawand that bit _was_ covered by the test suite17:06
richard_mawI think my next upstream-time project will be to allow in-line morphologies17:09
* persia has been looking forward to that since first encountering the syntax17:10
richard_mawsince if I can combine that with null sources, I can cut the source extraction stage out of some of the bootstrap17:11
persiaThat cuts into the initial 2-minute build delay?17:11
persiaAh, right "bootstrap" meaning the porting bits for new boards, etc.17:12
richard_mawyep, but since that's always in the build graph as part of build-essential, we also do it every time cache-keys are incompatibly changed and we have to rebuild everything17:13
persiaThe other nice feature is that it would allow removing chunk morphologies entirely, reducing the set of apparent moving pieces to be significantly smaller.17:15
* richard_maw would like a build mode where morph generates a shell script that does everything, so you could do a Baserock build on a system that doesn't have morph, but that adds too much complexity to the codebase17:16
persiaI'd rather just be able to install morph anywhere, rather than being dependent on having it installed as part of the host system.17:17
persiaAs most of the time I'd want to do a baserock build on a non-morph-containing-system, `pip install morph` ought to work.17:17
*** violeta [] has quit [Ping timeout: 246 seconds]17:21
* Kinnison urghs17:22
persiaYou're allowed to urgh at me wanting to `pip install morph` when you have no devices that fail to run Baserock natively17:23
persiaUntil then, we're just debating the relative merits of different package managers.17:24
*** pdar [] has quit [Read error: Connection reset by peer]17:28
KinnisonAt least you didn't ask for easy_install17:30
*** violeta [] has joined #baserock17:41
*** pdar [] has joined #baserock17:44
*** dutch [] has quit [Quit: Quit]18:08
*** violeta [] has quit [Ping timeout: 258 seconds]18:16
straycat"Life is too short for package managers."18:29
*** ssam2 [] has quit [Read error: Connection reset by peer]19:03
paulsher1oodbrew install morph :)20:00
petefothIf the answer is "brew install", then the wrong question is being asked!20:12
jjardonpersia: Did I heard coreutils? ;)20:29
*** thecorconian [] has joined #baserock20:32
jjardonforgot to mention in the cover letters: Ive deployed a genivi image and run weston to check everything keep working after applying the patches from both series (coreutils and NetworkManager)20:34
*** mwilliams_ct [] has quit [Ping timeout: 240 seconds]21:27
*** mwilliams_ct [] has joined #baserock21:27
*** pdar [] has quit [Ping timeout: 245 seconds]21:27
*** pdar [] has joined #baserock21:27
*** genii [~quassel@ubuntu/member/genii] has quit [Remote host closed the connection]21:40

Generated by 2.15.3 by Marius Gedminas - find it at!