noodles775Hey people. I'm trying to follow the ec2 getting started guide, but ran into a bunch of problems. I've tried to document them at https://bugs.launchpad.net/developer-ubuntu-com/+bug/1456390 , but any help appreciated.01:16
ubottuUbuntu bug 1456390 in Ubuntu App Developer site "Snappy ec2 getting started - various issues" [Undecided,New]01:16
sergiusensChipaca: no worries01:46
dholbachgood morning06:38
mvohey dholbach, good morning!06:45
dholbachhey mvo06:46
davidcalleMorning all o/07:34
davidcallercj, hello, do you mind having a look at https://bugs.launchpad.net/developer-ubuntu-com/+bug/1456390 ?07:36
ubottuUbuntu bug 1456390 in Ubuntu App Developer site "Snappy ec2 getting started - various issues" [Undecided,New]07:36
mvohey sergiusens and Chipaca - good morning! reading https://code.launchpad.net/~stephen-stewart/snappy/docs-tidy-up-package-metadata/+merge/259050 I wonder if we should simply standardize on django markdown and put that in the docs? its seems like a good idea to standardize and we love django :) WDYT?07:39
=== erkules_ is now known as erkules
Chipacamvo: gooood morning!08:20
* Chipaca remembers he has a "beuno" script just for this08:21
Chipacamvo: goooooooooooooooooooooooooooooooooooooooooooooood morning08:21
mvoChipaca: hahaha08:22
JamesTaitGood morning all; happy May Ray Day! 😃08:31
JamesTaito/ beowulf08:41
beowulfdholbach: hey, bugs in the docs, should they be put into snappy rather than developer-ubuntu-com, or does it not matter? (see noodles775 comment ^^)08:52
dholbachbeowulf, I can't see noodles775's comment - I guess I was not online at the time.08:57
dholbachbeowulf, if the bug is in lp:snappy ./docs/ better fix it there08:57
dholbachwe still have to figure out how to keep both in sync - we didn't quite come to a conclusion during UOS08:58
beowulfdholbach: ah, it was a bug report (https://bugs.launchpad.net/developer-ubuntu-com/+bug/1456390) and i was curious about the best place for same, as the doc src is in lp:snappy09:01
ubottuUbuntu bug 1456390 in Ubuntu App Developer site "Snappy ec2 getting started - various issues" [Undecided,New]09:01
dholbachthere are two sets of docs09:02
dholbachthere are the snappy internal docs (in the branch)09:02
beowulfi did not know this09:02
dholbachon developer.u.c we have most of the internal docs plus other stuff (like installation instructions, etc.)09:02
dholbachas I said above: things grew quickly so we haven't quite figured out how to go about things09:03
mvoChipaca, sergiusens: just fyi, I look at the recent fbtfs, I can reproduce it here in a sbuild environment09:05
davidcalledholbach, rcj should be able to help, he provided the ec2 images info for duc, I've pinged him about that09:07
dholbachthanks davidcalle09:11
Chipacamvo: do we want the syslog logger to ignore errors?10:15
mvoChipaca: how do you mean? I think its ok if syslog can not be initialized, i.e. I don't think snappy should fail entirely in this case. or was the question about something else?10:17
Chipacamvo: sergiusens: I can't find a unit test for the "application" -> "app" type serialization, so am writing it. Holler if it does exist and I missed it somehow.10:17
Chipacamvo: yeah, that's what i meant10:17
Chipacamvo: that is: i too believe that not being able to connect to syslog should continue, perhaps trying to log to stdout instead? or just not logging10:18
mvoChipaca: I think its ok unless I miss a strong reason why its not10:18
mvoand +1 for the test, I don't know that this is tested yet10:18
Chipacamvo: on the *other* hand, if snappy is a long-lived daemon, i think it should not start without a working syslog connection (but maybe make it an option because tests?)10:18
ogra_did you consider to swithc off logging by default ?10:19
=== dholbach_ is now known as dholbach
mvoChipaca: daemon> that is a good reason, we *might* consider just writing to a logfile in this case and yes, for the tests we should be able to disable it syslog or write a mock10:19
mvoogra_: whats the use-case?10:20
mvoogra_: to avoid spaming the log?10:20
ogra_appliuances usually have limited diskspace10:20
ogra_i would turn off loging completely by default and only log if the user supplies a certain flag somehow10:20
mvoa good point, however I would be suprised if we are the worst offender when it comes to writing logs10:20
ogra_really depends ...10:21
Chipacaogra_: well...10:21
Chipacaogra_: that's an argument to turn off logging to syslog by default10:21
ogra_some BSP kernels throw out 50msgs per second10:21
mvoChipaca: actually mocking syslog is probably the best idea10:21
Chipacano need, really10:21
ogra_Chipaca, well, all logging preferably ... probably leaving auth.log in place10:21
Chipacadepends what you mean10:21
Chipacaogra_: oh, you're not talking just snappy here10:22
mvoogra_: there was a discussion about syslog vs journald (which uses a fixed log space AIUI) a while ago10:22
ogra_i'm talking "snappy defaults"10:22
ogra_in general that is ... not just the daemon10:22
ogra_android actually only uses ringbuffers for logging to get around this ... but ringbuffers are limited (and gone when you reboot) ...10:23
Chipacaogra_: so, currently (as of last night (!)) we have a logger that logs things to syslog, and to the user's face (e.g. stderr if they're on console, or something tbd browser-appropriate for web), the in-users-face does not take up disk space10:24
ogra_no, indeed not ... but the users face also means a console10:24
mvoogra_: right, I think we can get the ringbuffer behavior by just removing rsyslog from the image, then journald takes over and we have the behavior you suggest10:24
Chipacaogra_: my expectation is that webdm will provide a logger there that'll show the messages to the user10:24
ogra_that makes sensew10:25
Chipacaogra_: Notices to a popup, Debugs to the console log10:25
Chipacasergiusens: ^ :)10:25
Chipacaanyway, bottom line, we need to disable syslog because it's FTBFMFS10:26
mvowell, we need to disable it in the tests at least :)10:26
Chipacamvo: would journald pretend it's syslog?10:26
mvoChipaca: I think so, I haven't looked into the details though, but I think journald just provides /dev/log and forwards to the rsyslog rightnow10:31
mvoright now10:31
Chipacamvo: sergiusens: is it intentional that we write out "app" as the json of something we read in as "application"? beowulf: does webdm expect "app"?10:37
beowulfChipaca: not sure i understand, do you mean the type attr?10:43
Chipacabeowulf: yep10:43
beowulfChipaca: webdm would need to know type oem, app, framework to do filtering, that's all10:44
beowulfoh, and to prevent installation of types10:44
Chipacabeowulf: right. the click store knows "application", not "app"10:44
Chipacabeowulf: so we're translating in snappy10:44
Chipacabut that translation is currently one-way10:44
Chipacaand i was wondering if that was design or accident :)10:44
beowulfChipaca: fewer words to describe the same things sounds good to me10:45
sergiusensChipaca: it's because they didn't want to add type app on the store side10:58
Chipacawe *do* have unit tests for serialization!10:58
Chipacafound them fixing up the code after the move :)10:58
Chipacasergiusens: i know why it's there10:58
sergiusensChipaca: we will need to add the new names anyways10:58
Chipacasergiusens: my question is, is it intentional that we deserialize "application" to "app", but serialize "app" to "app"10:58
ogra_sergiusens, so after staring at http://bazaar.launchpad.net/~ubuntu-system-image/ubuntu-system-image/server/view/head:/lib/systemimage/generators.py#L525 for half my morning, i'm actually wondering why we dont create the respective files, dirs and links in the respecive initrds, what do you think ?10:59
sergiusensChipaca: the store marks anything it doesn't know as application10:59
ogra_there doesnt seem to be a single dir file or link we would actually need to pre-exist for the initrd ...10:59
ogra_(and additionally system/userdata could just be a link to system/writable on the phone ...)11:00
ogra_(or bind-mount)11:00
ogra_that way you wouldnt run into probs once the rootfs is a snap either11:01
Chipacasergiusens: and hydrogen is light, but that doesn't answer my question either11:01
Chipacasergiusens: :D11:01
sergiusensmvo: good morning! Welcome back. And I welcome django11:01
ogra_unchained ?11:01
sergiusensogra_: it's too early or I lack context to understand what you want to tell me :-P11:01
ogra_sergiusens, this is about not re-packing the rootfs tarball on system-image.u.c11:02
mvosergiusens: hey, good morning! ok, I will look into md and lint in a bit11:02
* sergiusens understood the unchained joke11:02
ogra_so you are at least a bit awake :)11:02
sergiusensogra_: so, what files, dirs and links are you talking about?11:03
ogra_sergiusens, line 525 to 565 in the above link11:04
ogra_(and later also 488 to 523 for the phone)11:04
ogra_sergiusens, it alters the rootfs ... rsalveti was suggesting to move these bits to the live-build config instead11:05
ogra_but i think it might make more sense to do it from the respective initrds11:05
sergiusensogra_: we already do a bunch of dir creating in initrd, I don't see a problem aside from the "do the least you can in initrd" mantra11:07
ogra_(talking to you about it because you spoke up when we discussed it yesterday)11:07
Chipacai've got a silly question about initrd11:07
ogra_there are no silly questions about initrds ...11:08
Chipacaare the contents of the initrd left mounted so we can use them once booted?11:08
Chipacaif not, would that not be a good idea?11:08
ogra_it never gets mounted11:08
Chipacaexplicitly you mean?11:08
ogra_the kernel unzips it and unpacks it ... once it is done with using it the memory region it lived in gets flushed11:08
Chipacathe kernel mounts it directly, and then pivots?11:08
Chipacaahh, ok11:09
ogra_it pivots but to /dev/root (which is a ram device)11:09
sergiusensogra_: you could create a kernel that wastes memory and keeps it somewhere accessible inside proc maybe :P11:09
Chipacabut we still have things in initrd duplicated? but gzipped, so binding probably not doable i guess11:09
ogra_or just mount --move / /myinitrd at the end before calling run-init11:10
ogra_(or some such)11:10
Chipacai may or may not have abused initrd back when kernel+initrd fitted on a single floppy11:10
ogra_Chipaca, what is duplicated ?11:10
Chipacaogra_: sh? some modules?11:10
ogra_oh, that11:10
Chipacanearly all of initrd i guess11:11
ogra_we could do the same as andrroid and use initrd as / :)11:11
ogra_it would grow quite a bit then though11:11
ogra_(android never pivots ... they just use it as / with a minimal install and then mount nearly everything else in subdirs)11:12
Chipacai think this ties in with what we tossed around with i think mvo, about having multiple /bins, so we could have kernel snap have its bin and an oem snap have its own, without having to merge things11:12
Chipaca(or was it kernel and os)11:12
* ogra_ actually likes that initrd as root approach .... but the press would rip us into pieces :P "Ubuntu goes android OMG !!!11one1"11:13
sergiusensogra_: but it is rather clever11:13
Chipacadude, we were doing that way before android even existed11:13
Chipacait's not invented by them :)11:14
ogra_Chipaca, using the initrd as default / ?11:14
Chipacawe == the real open source community11:14
Chipacanot we == ubuntu :)11:14
ogra_we only use it as jumpstart and pivot out of it11:14
Chipacareal as opposed to android, which is openish11:14
ogra_as a buffer for HW compatibility in fact ... if it comes to ubuntu11:15
ogra_nothing more ...11:15
Chipacai'm pretty sure knoppix did it, for a start11:15
ogra_i know there is a terminal server project where they do it ...11:15
ogra_TCOS ...11:15
Chipacaand it was the only way to do rescue diskettes, and single-diskette distros11:15
Chipacano need for nfs with that11:16
Chipacaie just blat the initrd over tftp11:16
ogra_heh, last commit 8 months ago ... so not completely dead11:16
ogra_lol, the page offers packages for dapper ... nice :)11:18
Chipacaanyway, i guess we're not as constrained for disk space as we were back then, when even a 6$ computer comes with zomg space11:18
Chipacacan i say "6$ computer" again?11:19
* Chipaca shakes his nostalgia off, like a rain-drenched overcoat11:20
dholbachdavidcalle, let me know when you're there - maybe we can have a quick chat to discuss docs :)11:33
davidcalledholbach, in ~15 min?11:34
dholbachsure, wfm11:36
davidcalledholbach, ok11:36
Chipacasergiusens: you around?11:40
sergiusensChipaca: I need to walk the dogs before meeting marathon starts, so in 10' maybe  ;-)11:42
dholbachdavidcalle, so I thought that it might make sense if we would create google docs (or something) for the articles we want to import as tutorials, test them and bring them up for review for the snappy team, so we can see what's missing, what's broken and what needs to be updated11:51
dholbachdavidcalle, we could also open them up for comments by the wider community11:51
dholbachand maybe do the same for a "tools overview" kind of article11:51
davidcalledholbach, I very much agree, except the Google Docs part, GDocs -> d.u.c is quite painful. On the other hand, I don't have a better idea :)11:52
ogra_mvo, do you have any idea why we still build vivid images for core ?11:53
dholbachdavidcalle, maybe we just copy in everything and don't do any markup11:53
* ogra_ notices he gets arm64 failure mails for vivid builds11:53
dholbachdavidcalle, and copy in the same from the gdoc into the html view , so we don't get broken/unnecessary injected markup?11:53
davidcalledholbach, sure, that works11:55
dholbachdavidcalle, ok... shall I create placeholder docs in a google folder which we can freely share and we then take it from there?11:55
davidcalledholbach, yes, I'll probably be head down in scopes tutorials until thursday, but that sounds like a plan11:57
mvoogra_: hi, we need to get a updated image at some point, but I guess we don't really need to have a cron job for that currently11:58
mvoogra_: there are a number of pending bugfixes11:58
sergiusensdavidcalle: dholbach git branches and people creating MPs which act as comments12:01
ogra_mvo, yeah, thats what i thought ... we should build them on demand12:01
dholbachdavidcalle, cool12:02
ogra_wow ...  http://tracker.geops.ch ... thats like the real world in sim city12:02
ogra_(shows all buses, trains and trams in realtime for nearly the whole world)12:03
dholbachsergiusens, ... right12:03
davidcallesergiusens, maaaaybe, if you take care of merge conflicts12:05
Chipacaraaaah, raaaah, i am the shredder of go files, fear me12:06
rickspencer3hi all ... what is the best way for me to start with a snap in Go?12:09
rickspencer3ideally, I'd love a QtCreator project, but barring that, any suggestions?12:09
sergiusensrickspencer3: you can look at webdm or the go example server maybe12:10
rickspencer3sergiusens, webdm sounds rather complext12:10
sergiusensI don't think we have qtcreator integration at all with snaps12:10
sergiusensrickspencer3: right, the go example server is a better start12:10
rickspencer3I need something where it is obvious where to write the code, and where the snap build command just work12:10
Chipacarickspencer3: http://bazaar.launchpad.net/~snappy-dev/snappy-hub/snappy-examples/files/head:/go-example-webserver/12:11
Chipacarickspencer3: even includes a friendly README12:11
Chipacawe're getting soft in our old age12:12
rickspencer3Chipaca, sergiusens thanks gents, I'll give it a try12:12
rickspencer3next stop: QtCreator integration12:12
Chipacasergiusens: we should probably make that GOARCH=arm GOARM=712:12
ogra_whats GOARM ?12:13
ogra_and why do you need to specify it12:13
Chipacaogra_: 5, 6, or 712:13
Chipacaogra_: because it defaults to 612:13
Chipacaogra_: and we don't really support 6 :)12:13
ogra_it shouldnt :)12:13
ogra_(on ubuntu that is) ...12:13
sergiusensChipaca: for webdm you mean?12:13
ogra_sounds like a bug to me12:13
ChipacaGOARCH=arm GOARM=7 GO7=path GOPATH=....12:13
Chipacasergiusens: in the README. But everywhere.12:13
Chipacaif you're setting GOARCH=arm without also setting GOARM=7, it's a bug12:14
Chipacasupport for ARMv8 is in progress, also, fwiw12:14
Chipacasergiusens: i fixed it in webdm way back, btw12:15
nothal_https://code.launchpad.net/~chipaca/ubuntu-core-launcher/unshare/+merge/258367 | Approve: 1, Needs Fixing: 1 (12 days old)12:19
nothal_https://code.launchpad.net/~sergiusens/webdm/queryPackageNames/+merge/258640 | No reviews (10 days old)12:19
nothal_https://code.launchpad.net/~mvo/snappy/snappy-docs-mdlint/+merge/259490 | No reviews (less than a day old)12:19
nothal_https://code.launchpad.net/~chipaca/snappy/pkg-types/+merge/259485 | No reviews (less than a day old)12:19
nothal_https://code.launchpad.net/~mvo/snappy/snappy-fix-ftbfs-sbuild/+merge/259479 | No reviews (less than a day old)12:19
Chipacaverterok: nothal seems to not find git MPs, btw12:19
Chipacaverterok: compare with code.launchpad.net/snappy/+activereviews12:19
verterokChipaca: probably becase those git MPs are against a new branch which isn't configured? :)12:25
Chipacaorly? ah ok then :)12:25
verterokChipaca: what's git master/trunk branch you want to keep track of?12:26
Chipacaverterok: i dunno :)12:27
Chipacaverterok: but there is a high chance of us moving to git in the near future, was just checking whether it'd work12:27
verterokah, ok12:27
=== alex-abreu|off is now known as alex-abreu
Chipacame, refactoring: http://i.imgur.com/t0XHtgJ.gif12:47
Chipacamvo: question about iterHooks: is there a reason this is hung onto the click manifest, instead of the package yaml?12:48
beunoChipaca, and then mvo is all like: https://31.media.tumblr.com/28c234e45e157b55f923e109bbb9aa75/tumblr_inline_no7omriCW31raprkq_500.gif12:52
* Chipaca reaches for the mind bleach12:53
verterokChipaca: FYI just checked launchpadlib/launchpad API, git branches aren't yet supported there (at least I couldn't find a way to get the MPs for a git repo/branch)12:53
Chipacaverterok: boo, etc12:53
Chipacaverterok: if only we knew somebody who could ask the launchpad guys what's up with that12:54
verterokChipaca: indeed...oh wait! beuno might know12:54
* verterok slowly steps back12:55
* Chipaca grins and gets back to work12:55
mvoChipaca: I don't think so really, probably a good cleanup opportunity too12:55
mvobeuno: *ugh* that was a scary gif, NSFW12:55
Chipacamvo: that's where these little branches are all popping out of; start reworking things to be the way i think, get stuck because something seems wrong, back everything out and change that, and start over12:56
Chipacait gets easier every iteration \o/12:56
beunoChipaca, verterok, http://38.media.tumblr.com/d0d80832b4eb32e06deb46d1d1d774f0/tumblr_inline_nmpwjlVg201raprkq_500.gif12:56
Chipacabeuno: http://i.imgur.com/eTic1wS.jpg12:57
mvoverterok: have you asked on #launchpad or #launchpad-dev about this?  cjwatson will certainly know I would be suprised if there is really no way to get that info12:57
sergiusensChipaca: did you raise those refactor MPs already or are you still changing the light bulb?12:57
* Chipaca checks his pipeline12:58
ogra_he is cleaning his face from mvo's kisses still :)12:58
Chipacasergiusens: debsig-verify-to-clickdeb and pkg-types12:58
Chipacasergiusens: now working on reducing click manifest's reach12:58
verterokmvo: the devel version of the apidocs shows there is a way to get it...just not yet available in edge/production12:58
nothal_https://code.launchpad.net/~chipaca/ubuntu-core-launcher/unshare/+merge/258367 | Approve: 1, Needs Fixing: 1 (12 days old)12:59
nothal_https://code.launchpad.net/~sergiusens/webdm/queryPackageNames/+merge/258640 | No reviews (10 days old)12:59
nothal_https://code.launchpad.net/~mvo/snappy/snappy-docs-mdlint/+merge/259490 | No reviews (less than a day old)12:59
nothal_https://code.launchpad.net/~chipaca/snappy/pkg-types/+merge/259485 | No reviews (less than a day old)12:59
nothal_https://code.launchpad.net/~mvo/snappy/snappy-fix-ftbfs-sbuild/+merge/259479 | No reviews (less than a day old)12:59
Chipacasergiusens: the end game is pkg/snap/ (and not sure whether e.g. pkg/snap/oem or pkg/oem)12:59
Chipacathat's what i'm incrementally getting at at least12:59
Chipacathen it's just a move/rename if i've chosen bad names :)13:00
sergiusensbeowulf: ey, you never reviewed the package queries branch ;-) ^13:00
kyrofasergiusens, ping13:07
beowulfsergiusens: oops, sorry, will do this afternoon13:07
sergiusenskyrofa: pong13:22
elopiohey team. This cable works to connect to the beagle serial port, right? http://www.amazon.com/Ftdi-TTL-232r-3v3-Serial-Converter-Cable/dp/B00M41OUYA/ref=sr_1_1?ie=UTF8&qid=1432041571&sr=8-1&keywords=FTDI+to+TTL+3.313:25
kyrofasergiusens, as discussed yesterday (with alecu), webdm does indeed handle broken icons, by using http://webdm.local:4200/public/images/default-package-icon.svg . However, that image is completely blank on my install. Can you confirm?13:25
ogra_elopio, hard to tell if the order in the plug is correct13:25
kyrofasergiusens, does this relate to the svg-png weirdness?13:25
ogra_elopio, i usually buy cables with "flying ends" so i can more easily re-order them for specific boards13:26
ogra_elopio, like http://www.amazon.de/gp/product/B00KVUSI30?psc=1&redirect=true&ref_=oh_aui_detailpage_o02_s0013:27
tbrogra_: if it's the "standard" FTDI order, then it will work13:27
sergiusenskyrofa: you want beowulf to assist there; but it is for missing icons, not really broken ones13:28
ogra_tbr, but not all boards use that order13:28
kyrofasergiusens, ah indeed missing, not broken-- sorry13:28
* ogra_ has at least two in the most recent batch of boards on his desk that dont 13:28
tbrogra_: BBB and MinnowboardMax do. I have two of those FTDI things13:28
noise][i have http://www.amazon.com/gp/product/B00IJXNE1W/ and works with BBB, but DONT connect the red power line13:29
ogra_yeah, you usually dont need the power line at all13:29
sergiusenselopio: it should, you might be better off with one that has all the pins loose so you can plug them in any order (e.g.; banana pro)13:29
noise][since it doesn't convert down to 3.3v13:29
elopiook, makes sense.13:30
sergiusensI cut the power line just in case (I bought the bad batch 5vcc 3.3v ttl one that they decided to sell anyways)13:30
* tbr has a bucket of jumper wires for cases where he needs to have random layouts13:30
ogra_well, you never know if you need it some day ... my power line us tape wrapped :)13:30
sergiusensogra_: you can always reconnect what you cut ;-)13:31
kyrofabeowulf, any idea why http://webdm.local:4200/public/images/default-package-icon.svg is blank?13:31
ogra_sure ... but ripping of tape is easier than soldering :)13:31
ogra_i havent needed it yet anyway ... could as well have cut it i guess :)13:32
ogra_RX/TX/GND are usually enough13:32
alecukyrofa: is it a blank svg, or is it an empty file?13:36
beowulfkyrofa: it's not blank, it's a white svg on a transparent background on a white body :)13:36
alecuah, that.13:36
kyrofabeowulf, Ah! That would do it!13:37
beowulfkyrofa: the default body background is a shade of grey :)13:37
kyrofaAh, so it paints through on the "home" page, but appears blank in the store13:38
alecukyrofa: to be useful on the scope, perhaps we can ask for the shade of gray to be part of the svg13:38
kyrofaalecu, good idea. beowulf how would you feel about that?13:40
ogra_make it 50 though ... thats modern :)13:40
beowulfkyrofa: what are you trying to do?13:41
alecukyrofa: beowulf: I think there should not be icons with transparency in the store (at least that was a requirement for click apps)13:41
alecubeowulf: we are building a unity8 scope to install snappy apps13:41
kyrofabeowulf, and we're using the webdm API as a starting point13:42
beowulfalecu: kyrofa: yeah, i can change this to be an actual icon and not a placeholder (the background is the ubuntu.com background which has all sorts of noise and gradients so picking 'grey' will likely look strange)13:43
beowulfif you see what i mean13:43
beowulfso what you want is a proper 'missing icon' icon image13:43
kyrofabeowulf, that makes sense13:43
kyrofabeowulf, correct, and using the same one between webdm and the scope makes sense13:44
beowulfkyrofa: does the scope overlay to make a "squircle"?13:44
beowulfor do you simply take the icon as you find it?13:44
kyrofabeowulf, it does make a squircle13:45
beowulfkyrofa: ack13:45
elopiodholbach: ping. The two first tables in here are almost the same: https://developer.ubuntu.com/en/snappy/guides/filesystem-layout/#partitions13:45
elopiowhy do we need the first table?13:45
dholbachelopio, I don't know... davidcalle: do you know where we inherited these tables from ^?13:46
kyrofabeowulf, want me to make an issue for it?13:46
davidcalledholbach, imported doc from trunk13:48
dholbachhum, do you know where it's from?13:49
dholbachI can't find it in the branch13:49
davidcalledholbach, https://bazaar.launchpad.net/~snappy-dev/snappy/snappy/view/head:/docs/system-updates.rst13:50
dholbachoh ok13:50
beowulfkyrofa: yes please, thanks13:50
dholbachelopio, ^13:51
elopiojodh: ^ Do we need the first table in there?13:52
=== vrruiz is now known as rvr
kyrofabeowulf, done: https://bugs.launchpad.net/webdm/+bug/145665414:26
ubottuUbuntu bug 1456654 in webdm "Default package icon can't be seen in Store page" [Undecided,New]14:26
elopiodholbach or davidcalle: where is the source code for https://developer.ubuntu.com/en/snappy/start/#try-beaglebone ?14:26
dholbachelopio, there is none14:26
beowulfkyrofa: thank you14:26
davidcalleelopio, it's not an imported doc, it's only here14:26
elopiodavidcalle, dholbach: I think this:14:26
elopiounxz -c ubuntu-15.04-snappy-armhf-bbb.img.xz | dd of=/dev/sdX bs=32M14:26
elopioshould be:14:26
elopiounxz -c ubuntu-15.04-snappy-armhf-bbb.img.xz | sudo dd of=/dev/sdX bs=32M14:26
dholbachI'll fix it14:26
dholbachthanks elopio14:26
jodhelopio: just the first, or tables in general? I think the information is best presented in tabular form, hence rst (since md has poor support for tables)14:26
elopiodholbach: thanks!14:26
mterryDo you folks tend to use one GOPATH for all your branches, and just switch which one is the current "src/launchpad.net/snappy"?  Or do you have a new GOPATH for each branch, which you put at "src/launchpad.net/snappy"?14:26
elopiojodh: I like tables. It's just that the first table is almost the same as the second table.14:26
sergiusensChipaca: the pkg branch looks fine, but what is the next step? Just wondering about the layout there14:29
elopiojodh: I think we can remove the first table, and just leave the uboot and grub ones.14:29
jodhelopio: yes, we could drop table 1 and add the FS type and description to tables 2+3, but that would make them a lot wider and duplicate the information.14:29
mvomterry: I use a single GOPATH and symlink my snappy branch14:29
jodhelopio: they contain different information if you look carefully.14:29
* elopio looks carefully.14:29
mvomterry: git checkout will fix that nicely its really a non-ideal soluation14:29
elopiojodh: ahh, I see. You are right.14:29
mterrymvo, hmm makes sense -- the deps in GOPATH aren't going to need to change between branches14:29
mvomterry: yep14:29
elopiomight not fit anymore.14:29
mvoelopio: this makes me wonder if we could make it more obvious why the tables are obvious, just a random thought while reading your conversation14:29
sergiusensmterry: I use te same GOPATH14:29
sergiusensmvo: what might fix it better is to use cheney's go get replacement and just vendor14:29
mvosergiusens: I still haven't looked at this yet :/14:29
mvoI was hoping for git :)14:29
sergiusensmvo: http://getgb.io/ solves a bunch of problems though ;-)14:29
elopioI hate this network :@ It disconnects my quassel every two minutes.14:29
elopiomvo: well, there's a note that says: The boot partition must be formatted as a vfat. We could keep that and add another note saying that the rest of the partitions are ext4.14:33
elopionot sure if that's better than the table.14:33
elopioI think I would prefer to duplicate the FS column than to have an extra table.14:34
sergiusensmvo: did you see anything about zfs making it into the linux kernel btw?14:34
mvoelopio: i have no strong opinion either way, but it seems like a good opportunity to make the doc better because you look at it with fresh eyes now :)14:35
mvosergiusens: I have not, that would be exciting, was there anything new about this recently? I thought there were license issues14:36
sergiusensmvo: I think I heard some lawyers saying that it's not technically a gpl violation to include it, but I can't find any citation about that when googling; but if it becomes so, we will be in an awesome position :-)14:37
sergiusensmvo: maybe I read too much into this https://lists.debian.org/debian-devel-announce/2015/04/msg00006.html14:38
elopioI tried to reinstall my beagle with the stable image, and now I get:14:39
elopiossh: Could not resolve hostname webdm.local: Name or service not known14:39
mvosergiusens: oh, I think thats about shipping it as a dkms module or something14:39
elopiois there anything I can check while my serial cable arrives?14:39
sergiusenselopio: you need to install webdm for that avahi name to be exposed14:40
sergiusenselopio: maybe u-d-f core 15.04 --output bbb.img --install webdm14:40
mvosergiusens: fwiw, https://code.launchpad.net/~mvo/ubuntu-cdimage/mvo/+merge/228214 was my attempt to make the image building indepentant of launchpad14:45
mvosergiusens: just fyi, we talked about it during the call14:45
sergiusensmvo: let's see if ogra_ is bold enough to land that ;-)14:46
* sergiusens reads14:46
mvosergiusens: heh, I guess it needs some polish first, but thats the direction14:46
sergiusensmvo: procmail is a requirement? wow14:46
mvosergiusens: thats what I mean with "polish", I think we could do something with python here to avoid this dependency14:47
ogra_haha "HACKING"14:47
mvoogra_: I would like to resurrect something like this, wdyt?14:47
ogra_mvo, looks fine at a first glance, but if you dont add the necessary changes tpo the selftest too i think cjwatson might hunt me down14:49
sergiusensmvo: it's not even that invasive14:49
mvosergiusens, ogra_: thanks, I will try to spend a bit of time on it this week then14:52
elopiosergiusens: no luck with the --install15:29
sergiusenselopio: what kind of no luck?15:29
sergiusensas in what errors15:29
elopiosame: ssh: Could not resolve hostname webdm.local: Name or service not known15:30
sergiusenselopio: how is the bbb connected to the network?15:33
mvoelopio: so "bash" in the selftest and the "why"? it was basicly a quick way for me to not have to run these tests manually, but I much agreee, shell is a bad language for this and sergiusens suggested replacing that with something besser. I wonder if we should go with gocheck for the sake of consistency or python-unittest15:33
sergiusensmvo: elopio I used to just use pythons default unittest module and have it run all when calling it with the if __main__ gimmick15:35
kyrofasergiusens, the icon in the snappy webcam-demo does actually seem to be invalid. Is this known?15:36
sergiusenskyrofa: yeah, we need lool or asac to setup a proper icon in there15:38
kyrofasergiusens, okay, as long as it's a known issue :)15:38
sergiusenskyrofa: it is ;-) and we expose it loud and clear on webdm :-)15:43
sergiusensChipaca: can we discuss 'pkg' now?15:43
kyrofasergiusens, Ha! I'm doing the same on the scope :)15:44
Chipacasergiusens: sure15:44
sergiusensChipaca: ho r irc?15:44
Chipacasergiusens: irc would be better right now15:44
Chipacasergiusens: or ho in 30'15:44
sergiusensChipaca: ho in 30'15:46
Chipacasergiusens: k15:46
sergiusensChipaca: I wrote something down on the review itself, but the direction would be nice15:46
sergiusensChipaca: maybe mvo wants in as well15:46
asacsergiusens: you can make one :)15:47
asaci think we have svgs15:47
asacfor the snappy log15:47
asacyou can put a nice camera overlayed over it15:48
sergiusensasac: I don't have the package sources though15:50
Chipacasergiusens: responded in the mp also15:51
Chipacasergiusens: question for you: how much do we care about returning "invalid part" when it doesn't have a namespace, in places that don't really care about it?15:52
Chipacae.g. in partsForGlobExpr15:52
sergiusensChipaca: I need to figure out the ripple effect for that15:52
Chipacasergiusens: i ask because i'm trying to move all these "if it's a framework or an oem" thing to one or two places at most :)15:53
Chipacathat's at the top of the stack, previously at the top of the stack was "make things not use the click manifest", which ended up with more namespace checks everywhere15:53
Chipacaso, new pipe, clean up namespace whatsits15:53
Chipacathen back to that one15:53
Chipacaand then not using the click manifest means one step closer to packageYaml being separable15:54
Chipacastill a ways off though15:54
Chipacaman, if i were still in córdoba people would be running around like crazy hiding their cars15:56
Chipacathe sky is that weird green you get before a big hailstorm15:56
Chipacazomg, hail :)15:56
sergiusensChipaca: hah, true :-)15:58
sergiusensChipaca: let's move that meet a bit, going to run some lunch errands15:58
elopiosergiusens: mvo: sorry, had a meeting here, and now lunch. I like the idea of writting the tests in the same language you are using, so go sounds nice. But then we would have to build the tests, or tell adt-run to build them on the board.16:26
elopioI know python better, and that's easy to write.16:27
Chipacaelopio: OTOH if you do it in python, then you'd have to get python onto the device16:32
Chipacaelopio: (we hope to remove python entirely at some point)16:32
beowulfsergiusens: Chipaca: mind you helping me with this, output of webdm ./build.sh http://pastebin.ubuntu.com/11227865/16:32
* Chipaca looks16:33
Chipacabeowulf: you're building with the system go, and you don't have the arm system go16:34
Chipacabeowulf: install golang-go-linux-arm16:34
Chipacabeowulf: and you should be set16:34
beowulfChipaca: thanks16:35
Chipacabeowulf: option b is you're building with your own go, and haven't done an "all" build with GOARCH=arm16:35
* beowulf makes note to fill out readme 16:35
Chipacabeowulf: if option b sounded like something almost but not quite entirely dissimilar to gibberish, then it's probably the first one16:36
Chipacawait, i16:36
Chipaca'm missing a negation in there16:36
* Chipaca adds more negations16:36
Chipacabeowulf: if option b failed to sound like something almost but not quite entirely dissimilar to gibberish, then it's probably the first one16:36
Chipacathere, now you can go16:36
beowulfChipaca: I...16:38
beowulfChipaca: Part II of your test http://pastebin.ubuntu.com/11228018/16:38
Chipacabeowulf: rm -rfv --wtf --bbq /home/sstewart/go/pkg16:39
Chipacabeowulf: or set it aside if you'd rather be cautious16:39
Chipacaalso maybe without --wtf and --bbq16:39
Chipacabeowulf: but basically you've got compiled pacakges for go 1.2.x and it's looking for 1.3.x16:40
* beowulf regrets the things he did to end up here16:40
beowulfChipaca: thanks, works now16:40
Chipacabeowulf: i'm sorry16:40
beowulfChipaca: whyfore?16:41
Chipacabeowulf: i accidentally got you unblocked16:42
mterrysergiusens, I see you added python to the ubuntu-core seed.  It never got released as part of the meta package in the archive though.  Should that still be there/be a part of the meta package?16:43
* mterry wants to release a new meta package removing gawk and original-awk16:45
sergiusensmterry: I think I proposed it against the wrong seed originally16:52
mterrysergiusens, you did, but it got put in the right seed.  But still, a new meta package was never spun up16:52
mterrysergiusens, do we want it in the image?16:52
sergiusensand either slangasek or mvo (gone) took care of it16:52
mterrysergiusens, oh16:53
mterrysergiusens, so we don't want it in the ubuntu-core seed at all?16:53
sergiusensmterry: I hope I described in the MP why it was suppposed to be there16:53
beowulfsergiusens: when i build lp:~sergiusens/webdm/queryPackageNames and query /api/v2/packages/?q=8nzc1x4iim2xj1g2ul64 i get too many results16:53
mterrysergiusens, yeah for "Adding python to system-image seeds to support walinuxagent"16:53
sergiusensbeowulf: to be fair, that just uses the search api from the store, what if you search for something more specific?16:53
mterrysergiusens, it just never got released as far as I can see, so I wanted to make sure it should still be there before I released a new meta16:54
sergiusensmterry: great, utlemming did the port to python3 ever get done?16:54
utlemmingsergiusens: yeah, I think we're just waiting on some final bits16:55
utlemmingsergiusens: and an sru for cloud-init16:55
sergiusensutlemming: great16:55
sergiusensutlemming: but it's in wily already, right?16:55
sergiusensmterry: if that port is done, we can get rid of the python dep (which I think was also manually setup in livecd-rootfs as well)16:56
beowulfsergiusens: actually, results are the same every time... (minor lol that you're asking me to search for somethign more specific than '8nzc1x4iim2xj1g2ul64')16:56
sergiusensbeowulf: :-)16:56
sergiusensbeowulf: weird, it was working fine for me when I tried it...16:57
beowulfsergiusens: i'll check my install again in a bit, going for food now16:57
sergiusensbeowulf: I'll give it a spin16:57
mterrysergiusens, well the python dep was never a part of the meta package, and if it will go away for wily, I won't add it now.  I'll drop it from the seed16:57
sergiusensChipaca: talk talk?16:57
Chipacasergiusens: natter natter16:58
sergiusensmterry: sounds good16:58
Chipacasergiusens: https://plus.google.com/hangouts/_/canonical.com/natter-natter?authuser=116:58
=== vrruiz is now known as rvr
Chipacaallright, tests pass, ftw, bbq, & etc19:22
Chipacasergiusens: https://code.launchpad.net/~chipaca/snappy/reduce-manifest/+merge/259539 for when you're able and willing19:25
* sergiusens doesn't mind a bbq sprint19:30
sergiusensChipaca: you have conflicts19:30
Chipacayeah, noticed that19:30
sergiusensChipaca: the lp preview really makes it hard to see why, but I guess it's obvious to you so I'll wait before doing a local pull19:32
Chipacasergiusens: i removed files that have changes in them on trunk19:33
Chipacasergiusens: there ya go19:34
Chipacatyhicks: nudge nudge20:04
tyhicksChipaca: ah! sorry! will do it before my eod20:09
Chipacatyhicks: wink wink20:09
Chipacatyhicks: say no more20:09
sergiusensrickspencer3: I reproduced your webdm always wants to upgrade problem on the rpi2 with lool's image; I'm guessing this is a framework problem20:10
sergiusensI'll research this20:10
rickspencer3hi sergiusens20:10
rickspencer3so, if someone builds robots and is interested in snappy, which is the best mailing list to recommend they join?20:12
rickspencer3sergiusens, did webdm actually work for you on the pi2?20:14
sergiusensrickspencer3: and it's because the img is bad20:14
sergiusensrickspencer3: it needs to be regenerated20:14
rickspencer3seems like we need some kind of CI that a community image developer could run20:14
rickspencer3sergiusens, ah, lool is on holiday until next week20:14
sergiusensrickspencer3: no, the oem package is bad; it has a namespace which means it probably was created even before the release20:15
rickspencer3sergiusens, oh20:15
sergiusensrickspencer3: he is back on thursday; I can generate one with u-d-f, it should work20:15
elopioChipaca: right, but adt-run can install the python deps just for the tests in a tmp dir, and then remove them.20:15
rickspencer3sergiusens, ok, if you are done soonish I can test it20:15
elopioI think, I haven't tried installing everything, just some packages.20:16
elopioit will be slower, while it downloads and installs everything.20:16
sergiusenselopio: Chipaca our tests could very well be written in go then20:16
sergiusensno need for dep baby sitting there20:17
Chipacaelopio: go has very good cross-compilation support, meaning that building it for arm on intel is a no-brainer most of the time20:17
sergiusensrickspencer3: wrt to the skynet creators, snappy-app-devel I think would be the best20:18
elopiogo it is then :)20:18
sergiusenselopio: you will love go, like it or not(?) :-P20:18
elopioI'll try to re-write one, to improve my go.20:18
rickspencer3thanks sergiusens20:18
elopiosergiusens: I like go. I just need to learn more.20:19
elopioso if you see on my branches something stupid, please don't be shy...20:19
sergiusenselopio: I'm never shy with reviews as I hope people are never shy to say I did something wrong ;-)20:20
Chipacasergiusens: we should all go have a sprint at elopio's, to help him get up to speed20:21
Chipacasergiusens: he just happens to live a stone throw's away from a rather nice bit of beach20:22
elopioI pay for the first round of beers!20:22
ChipacaI can see no holes in this plan20:23
sergiusensChipaca: sounds like a plan, rsalveti should set it up ;-)20:24
sergiusensgood team building20:24
sergiusensand... winter is coming (here)20:24
elopiowell, we don't have winter. But we are in the middle of the rainy season.20:29
elopioyou should bring a poncho, or wait a couple of months.20:29
Chipacai've got to wait a couple of months for a passport anyways20:32
Chipacaverterok: https://github.com/wyc/armbot20:37
Chipacaverterok: just sayin'20:37
Chipacaverterok: now all you need to do is rewrite launchpadlib in arm assembly, and you're set20:38
verterokChipaca: "motivation: lol"20:52
Chipacaverterok: :)20:52
rsalvetiyeah, wouldn't mind going there either, raining like hell here lately21:00
rsalvetiand, winder is coming21:00
sergiusensricmm: sudo ubuntu-device-flash core rolling --channel edge --oem generic-i386 --output i386.img21:14
=== j12t_ is now known as j12t
tyhicksChipaca: I got started on the review but need some more time to finish up - it'll be my top prio tomorrow morning22:38
Chipacatyhicks: ok, thanks!22:39
tyhicksChipaca: check out the MP - I found an issue22:58
Chipacayeah, just reading it22:58
tyhicksChipaca: I haven't had time to think of the proper fix yet and still have more review to do22:58
tyhicksI gotta run22:59
Chipacaeasiest fix is to not make tmpdir predictable22:59
Chipacawhich was a feature maybe, but maybe not23:00
Chipacawill chat with mvo tomorrow23:00
=== noodles775 is now known as michaeln
=== michaeln is now known as mnelson
=== mnelson is now known as noodles775
ricmmsergiusens: thanks23:49
=== noodles775 is now known as miken

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!