IRC logs for #baserock for Tuesday, 2015-10-06

*** Walkerdine has left #baserock00:42
*** De|ta has quit IRC00:50
*** gtristan has quit IRC03:06
*** gtristan has joined #baserock03:20
*** zoli___ has joined #baserock03:57
*** zoli__ has quit IRC03:57
*** zoli___ has quit IRC04:10
*** zoli__ has joined #baserock04:10
*** bfletcher_ has quit IRC05:03
*** gtristan has quit IRC05:03
*** benbrown1 has quit IRC05:04
*** benbrown_ has joined #baserock05:05
*** gtristan has joined #baserock05:08
*** bfletcher has joined #baserock05:19
*** zoli__ has quit IRC05:29
*** paulwaters_ has joined #baserock06:59
*** ssam2 has joined #baserock07:08
*** ChanServ sets mode: +v ssam207:08
*** zoli__ has joined #baserock07:11
*** zoli__ has quit IRC07:13
*** zoli__ has joined #baserock07:14
*** zoli__ has quit IRC07:15
*** zoli__ has joined #baserock07:15
*** zoli__ has quit IRC07:33
gtristanSo... autoconf-archive and gnulib07:38
gtristanI'm tried moving that over to core/, but perhaps it does indeed belong in gnome/ ... in any case... gnome.morph does not list these, yet there are gnulib.morph and autoconf-archive.morph in the strata/gnome/ subdir07:39
gtristanI suppose it's of little concern where they are, however building them produces some error... gnulib does it's thing and seems to install, yet autoconf-archive build calls ./bootstrap.sh... which fails to find gnulib-tool07:40
gtristanand build aborts...07:40
gtristanlooking at the 'debris'... there is indeed a usr/bin/gnulib-tool handy in there07:41
gtristanany clues ?07:41
ssam2gnulib is used as a submodule only07:41
wdutchSotK: nice :) it even looks good on my phone07:41
ssam2gtristan: unless jjardon did something different in strata/gnome.morph07:42
gtristanaha, interestingly... gnulib-tool is a symlink to /gnulib.inst/usr/share/gnulib/gnulib-tool... which doesnt apparently exist anywhere07:42
gtristanssam2, not sure what the definition of 'submodule' is in this case :-/07:43
ssam2gtristan: the normal Git definition :-) however, you have to do a bit of a hack to get submodules to work with Morph07:43
ssam2gtristan: well, you don't *have* to, but policy kind of dictates it07:44
gtristanssam2, what jjardon did in strata/gnome.morph, is not mention autoconf-archive or gnulib at all, but leave stale morph files in strata/gnome/07:44
gtristan:-S07:44
ssam2gtristan: ouch07:44
ssam2well, i did review it as well and didn't notice :)07:44
ssam2thing is, lots of GNU programs require the gnulib  Git submodule to build07:44
ssam2so we did discuss installing it in the system, for reasons which I forget07:44
ssam2it's not meant to be installed, hence the 'cp' stuff in gnulib.morph07:45
ssam2generally the gnulib 'bootstrap' script needs a few tweaks to get it to run inside a Baserock build, it has a crazy number of dependencies that it usually doesn't actually need07:46
ssam2see the instructions for Bison, for example: http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/tree/strata/core/bison.morph07:46
ssam2and when I talk about the submodules 'hack'... I mean this: http://git.baserock.org/cgi-bin/cgit.cgi/delta/bison.git/commit/?h=baserock/v3.0.2&id=2ab6d1daaccf32fc4314e4b2fe44da977f11a30807:47
*** toscalix has joined #baserock07:47
ssam2we do that so that you can mirror the submodule on git.baserock.org or wherever, instead of using the gnu.org URL07:47
ssam2but it'd be much nicer to overwrite the .gitmodules file in the .morph file somehow, instead of having to fork the actual repo07:47
gtristanI see... sort of07:48
ssam2none of that answers your original question, I think :)07:48
gtristanso basically, its going to take a lot of arm wrestling to get any gnome module which has ported away from gnome-common macros to actually build07:49
ssam2well... adding autoconf-archive to core seems like a no-brainer07:49
ssam2at a glance, the hiccough seems to be that instead of putting the url for gnulib.git in .gitmodules, they've hardcoded it in http://git.baserock.org/cgi-bin/cgit.cgi/delta/autoconf-archive.git/tree/bootstrap.sh07:50
gtristanexcept that it requires gnulib-tool, which... should *not* be installed ?07:50
ssam2so a 'normal' autoconf archive build would happily run bootstrap.sh and clone gnulib.git from that hardcoded URL07:50
ssam2(ignoring the fact that a 'normal' build would use tarballs, not Git, for a moment)07:51
ssam2but for us that's broken, because builds might not have access to the internet, maybe just one Git server with all the repos on it07:52
ssam2bootstrap.sh seems to check for existence of a 'gnulib' dir before doing the clone itself.. so perhaps adding a .gitmodules file similar to the Bison one would be enough07:52
gtristanat this point, it would seem that gnulib is a build tool which is used by more than one module, and... I dont see the story (yet) about the manifests for the final distributed image either07:53
gtristanso, considering that, there seems to still need to be a story for stripping away all the dev/doc from a built image (the gnome system actually uses > 4GB !)... would it not make more sense to just install it as a build tool ?07:54
ssam2there is a story for that, it's just a bit undocumented and weird07:54
gtristanheh :)07:54
ssam2both Morph and YBD support splitting artifacts into foo-bins, foo-docs, foo-libs, etc07:54
ssam2and do so by default07:54
gtristanok, I did not see any manifests in the definitions/ so thought there was nothing there07:55
ssam2currently it's hardcoded in Morph and YBD07:55
ssam2there's a change to the format in progress (version 7) which moves the rules into definitions.git07:55
ssam2https://gerrit.baserock.org/#/c/1083/4/DEFAULTS07:55
ssam2(split-rules are near the bottom)07:56
ssam2and then.. here's an example of a system that excludes some of the artifacts from the final image: http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/tree/systems/minimal-system-x86_64-generic.morph07:56
gtristanright, and I guess that artifact/strata morphs themselves are intended to describe what parts of the payload is -dev and -doc ?07:57
ssam2the default split rules do that in general07:58
ssam2a chunk .morph file can override them07:59
gtristanssam2, have a meeting with Agustin...07:59
ssam2but (this is one of the big problems with the system right now) it can only override, not extend..07:59
ssam2(i.e. needs to copy and paste the defaults in order to add to them)07:59
* gtristan will have to figure out what to do with autoconf-archive / gnulib after :-/07:59
ssam2good luck!07:59
gtristanfor gnulib... looks like I'll need it ;-)08:00
*** bashrc has joined #baserock08:07
*** tiagogomes has joined #baserock08:08
*** jonathanmaw has joined #baserock08:23
*** zoli__ has joined #baserock08:29
*** mariaderidder has joined #baserock08:43
jjardongtristan: what problem do you have with autoconf-archive and gnulib?08:44
jjardonit was not needed to use that hack in the end in the GNOME stratum: m4-common fixed the problem I had in the end08:45
*** toscalix has quit IRC08:48
*** toscalix__ has joined #baserock08:51
jjardonCan someone give me another +1 to lorry this? https://gerrit.baserock.org/#/c/1156/08:51
perrylhi all, i'm working on a few modifications to trove that will create links in definition morphologies between repo and ref fields and their locations on g.b.o. to test this out i'll need to spin up a trove instance; is this something anyone can do?08:52
ssam2perryl: yes, and it's actually documented! see http://wiki.baserock.org/Trove/08:58
perrylssam2: ahh brilliant! thanks :)08:58
*** toscalix__ is now known as toscalix09:01
gtristanjjardon, gnome-disk-utility switched away from gnome-common... provides the volume monitor for gvfs... which otherwise crashes without any volume monitors in place09:04
gtristanjjardon, however, rishi says not to use gdu now but to use udisks2, afc, gphoto2 and mtp instead09:04
gtristanjjardon, so I can scratch gdu off the list, which will at least temporarily get rid of the need of autoconf-archive09:05
gtristanhowever, I think the trend is going to be to move away from gnome-common macros, so I'm quite sure the need of autoconf-archive will arise again, as it's the preferred replacement09:05
SotKwdutch: good :)09:05
jjardongtristan: make it depends on m4-coomon an it should work09:06
gtristanjjardon, alright :)09:06
jjardon(m4-common will install autoconf-archive)09:06
gtristanah good09:06
jjardongtristan: yeah, I hope I have time to push to get completely rid of gnome-common this cycle09:07
*** zoli__ has quit IRC09:07
*** edcragg has joined #baserock09:08
*** gtristan has quit IRC09:29
*** ssam2 has quit IRC09:41
*** ssam2 has joined #baserock09:44
*** ChanServ sets mode: +v ssam209:44
*** gary_perkins has joined #baserock09:51
*** franred has joined #baserock09:55
jonathanmawI'd like to see how well I can fit yocto recipes into CIAT. Is there documentation for how to get an instance of CIAT running?10:11
KinnisonNope, it's currently very loose.  There're READMEs in the various repositories, but frankly you'd likely be better off bringing up your own buildbot+slave given that there is little in common between yocto and baserock definitions10:12
jonathanmawKinnison: ok, where do these repos live? looking at git.baserock.org, I see firehose and Trove, I'm not sure what repo Orchestration and CIAT-Tester would live in.10:15
Kinnisonjonathanmaw: all the CIAT prototype content is on cu010-trove.codethink.com currently10:16
jonathanmawta Kinnison10:16
*** zoli__ has joined #baserock10:33
*** zoli__ has quit IRC10:42
*** zoli__ has joined #baserock10:50
*** gtristan has joined #baserock11:08
*** gary_perkins has quit IRC11:13
*** fay_ has joined #baserock11:35
*** paulwaters_ has quit IRC11:41
*** paulwaters_ has joined #baserock11:41
*** gtristan has quit IRC11:48
* pedroalvarez thinks about git bisect-ing failures in CIAT whenever something fails 11:59
*** gtristan has joined #baserock12:08
*** fay_ has quit IRC12:49
jonathanmawhrm, the clone URLs on cu010-trove are wrong, they say to use cu010-trove.codethink.co.uk, instead of *.com13:01
Kinnisonyep13:02
Kinnisonit's a dodgy trove I hope to decommission as soon as we can13:02
*** gary_perkins has joined #baserock13:06
paulsherwoodssam2: ybd doesn't support splitting. it did, briefly, but i backed out the change since it broke something else13:13
ssam2oh, that's a shame13:14
paulsherwoodagreed. i need to get around to re-adding it13:14
paulsherwoodthe revert commit is c3567f4 if anyone is interested to look at it13:17
gtristanalright so let's say I have some patches against definitions/ to submit for review, I have logged into gerrit... what's next ?13:30
KinnisonYou need to sort out being able to push13:31
Kinnisonthen you can submit a topic to gerrit13:31
KinnisonI think that involves registering an ssh key with gerrit13:31
gtristanI see, I did register my public key there13:31
Kinnisongerrit should then have instructions on submitting topics13:31
Kinnisonbuilt into the UI13:31
gtristanAlso, there are a couple changes to gnome.morph, which currently point to repos in file:///home/tristan...13:32
* Kinnison has been told it's quite simple13:32
KinnisonThen the patches definitely won't be accepted13:32
gtristanand I'm not sure exactly how to lorry them into git.baserock.org/delta13:32
gtristanright I was expecting that :)13:32
* Kinnison isn't sure if we review lorry changes on gerrit or not13:32
KinnisonSotK: Do you remember?13:32
tiredcatwe do13:32
KinnisonOh tiredcat hey13:32
KinnisonOkay, so you should first submit lorry patches13:33
Kinnison(to baserock/local-config/lorries)13:33
gtristanok, so I should clone the lorry repo I guess, and once those get upstream... I can change my definitions/, test, and submit patches I'm more sure about ?13:33
Kinnisonyep13:33
gtristangotcha13:33
KinnisonYou may want to accompany your lorries patch when an indication of what you're working on, otherwise whoever reviews the lorries might decide they're pointless13:34
gtristanHmm, currently my definitions repo is parented to the url: http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git ... but that's strange... should it really be cgit ?13:35
Kinnisonyou can reset your origin to whatever you want, that's just one of the ways to clone (over http)13:36
gtristanright... I was wondering more specifically because I dont understand yet how gerrit patch submission works, and you said I need to sort out being able to push :)13:37
gtristanso I wonder if the origin probably matters for that13:37
Kinnisonoh you can just add another remote for gerrit13:37
gtristanhmmm, so in open-source-lorries... there is a little bit different setup than in definitions/... i.e. in definitions... the gnome.morph and associated morphs, list modules not necessarily part of gnome... while gnome-all.lorry seems to only mention software that actually comes from gnome, so... if I wanted to add udisks2 and libatasmart... those would go into separate lorry files correct ?13:44
Kinnisonmost likely, yes13:45
* gtristan looks at commit dc52f53d1f7fd0d3010bf3d88ad5c972be271eca... which seems to just only add a file, so no need to update anything else but to add a lorry file... alright13:48
* Kinnison wonders what gtristan is working on13:49
Kinnisonit sounds complex :-)13:49
Kinnisonmust be fun to be at it at nearly 11pm13:50
pedroalvarezgtristan: I normally send patches using git-review tool13:53
gtristanheh13:54
gtristanKinnison, nice of you to notice :)13:54
pedroalvarezYou should be able to do `git review -R - t topic-of-the-changes'13:54
pedroalvarez(IIRC)13:54
Kinnisongtristan: it's amazing what /whois and then proper whois can tell you about a person who doesn't ghost their IRCname13:54
gtristanI'm actually asking cause, I know when I wake up tomorrow, nobody will be around to answer me ;-)13:54
rjekwin 5213:55
rjekfail13:55
gtristanpedroalvarez, so does this imply I should create a local branch ?13:55
gtristanI've staged 2 changes to lorries here...13:55
pedroalvarezNope, no local branch needed13:56
*** gary_perkins has quit IRC13:56
*** zoli__ has quit IRC13:56
pedroalvarezNor fixing the remote13:56
pedroalvarezNor adding a new one (this tool does this for you)13:57
paulsherwoodgtristan: i assume you;ve seen the gerrit instructions at http://wiki.baserock.org/contributing/#index6h2 ?13:58
gtristanpaulsherwood, well, it doesnt say much about how to use it :)13:59
gtristanonly that there may be false merge conflicts13:59
gtristansetup a gerrit account... I should read though that though :)14:00
paulsherwoodsorry, i cop14:00
paulsherwoodied the wrong link14:00
pedroalvarezOh, no mention about git-review14:01
paulsherwoodtrue. hence i've never used it14:01
paulsherwoodif someone could document setting that up, i could try :)14:01
pedroalvarezpip install git review14:02
pedroalvarezI think14:02
pedroalvarezAnd then the other command I said to send something for review14:02
paulsherwoodnot here, on the wiki :)14:02
pedroalvarezHeh :)14:02
* pedroalvarez makes a note 14:03
gtristangit review... or git-review ?14:03
Kinnisonsame thing14:03
Kinnisongit foo ==> git-foo14:03
Kinnisonjust it looks in both the path and the git-core libexec stuff14:03
*** gary_perkins has joined #baserock14:03
gtristanKinnison, except that pip craps out with 'pip install git review' ;-)14:04
Kinnisonoh pip14:04
* Kinnison shrugs14:04
SotKit'll be git-review on pip I imagine14:04
KinnisonI don't use pip14:04
* Kinnison resists his diatribe about things which abuse system installation paths14:04
tiredcatpip/pypi is indeed a pile of crap14:05
* edcragg learnt everything he needed to know about git-review from the mediawiki wiki, iirc14:08
*** zoli__ has joined #baserock14:08
tiredcatedcragg, i guess the point is that all the git-review stuff is an oft enough repeated instruction that we could include/link to existing instructions for git-review in our patch submission process14:09
tiredcatanyhoo…14:09
edcraggindeed14:12
gtristanhmmm, ok git review almost worked like magick14:12
gtristanand then: http://paste.baserock.org/bunetumosu14:12
gtristanah14:13
gtristansorry14:13
* gtristan palmface14:13
gtristanalright, lorry changes submitted as https://gerrit.baserock.org/1157 and https://gerrit.baserock.org/1158 :)14:17
franredgtristan, what are you going to use these repos for? are they go to be in some system?14:25
paulsherwoodmerged :)14:26
tiredcatomg merge all the things!14:26
gtristanfranred, gnome-system-x86_6414:26
paulsherwoodtiredcat: indeed :)14:26
franredgtristan, cool :)14:27
gtristanfwiw, gvfs *currently* craps out if it does not build any volume monitor14:27
gtristanbut14:27
gtristanthat at least will be changed after today: https://bugzilla.gnome.org/show_bug.cgi?id=75610514:28
gtristanstill, I think it needs the volume monitors, and probably worth adding gphoto and others too14:28
gtristanbut lets get a running system first14:28
*** gtristan has quit IRC14:47
*** fay_ has joined #baserock14:48
*** gtristan has joined #baserock14:51
*** franred has quit IRC15:11
*** franred has joined #baserock15:11
*** zoli__ has quit IRC15:21
perryli'm currently setting up ssh keys for a trove system and i'm confused as to what $TROVE_HOST refers to and what value i should give it; can someone explain it?15:29
richard_mawI think it's the hostname that the trove will be reachable at15:31
* richard_maw checks15:31
perrylahh, i was looking in the example cluster for anything that had the same field15:32
richard_mawthere's TROVE_HOSTNAME in trove.configure which I think is exactly that15:33
richard_mawsome older definitions appear to call it TROVE_HOST, but yeah, it's generally the address that the trove will be reachable from15:34
perrylahh, so it defaults to the trove id? that makes sense15:35
perryllooking at trove.configure that is15:35
Kinnisontrove-id should be short, trove-hostname should be an FQDN unless you have good reason for it not to be (e.g. local VM)15:37
*** gary_perkins has quit IRC15:37
*** paulwaters_ has quit IRC15:40
*** zoli__ has joined #baserock15:52
*** franred has quit IRC15:56
*** ssam2 has quit IRC16:03
*** De|ta has joined #baserock16:04
*** zoli__ has quit IRC16:09
*** gary_perkins has joined #baserock16:15
*** gary_perkins has quit IRC16:35
*** gary_perkins has joined #baserock16:37
*** gary_perkins has joined #baserock16:38
*** zoli__ has joined #baserock16:43
*** mariaderidder has quit IRC17:02
*** flatmush has joined #baserock17:03
*** bashrc has quit IRC17:03
*** zoli__ has quit IRC17:03
*** jonathanmaw has quit IRC17:07
*** edcragg has quit IRC17:07
*** gary_perkins has quit IRC17:07
*** toscalix has quit IRC17:10
*** zoli__ has joined #baserock18:21
*** zoli__ has quit IRC18:25
*** zoli__ has joined #baserock18:25
pedroalvarezpaulsherwood: do you think is a good idea to mention ciat in our homepage of the wiki?20:05
pedroalvarezSome people will go there to find more information20:05
pedroalvarezMaybe we currently do that20:05
pedroalvarezNope20:06
*** zoli__ has quit IRC20:23
*** zoli__ has joined #baserock20:23
*** peeterkins has joined #baserock20:42
*** zoli__ has quit IRC20:43
*** zoli__ has joined #baserock20:46
*** peeterkins has quit IRC20:46
SotKARM builder's disk looks full :/21:11
*** zoli__ has quit IRC21:21
*** zoli__ has joined #baserock21:23
*** flatmush has quit IRC22:01
*** fay_ has quit IRC22:01
*** jjardon has quit IRC22:22
*** drnic has quit IRC22:22
pedroalvarez:/22:42
*** jjardon has joined #baserock23:17
*** drnic has joined #baserock23:17
*** fay_ has joined #baserock23:27
*** gary_perkins has joined #baserock23:32
jjardongtristan: btw, I think you will be interested on reviewing this to avoid the integration command in pango :) https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:pango_1_3823:42
*** fay_ has quit IRC23:57

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!