IRC logs for #baserock for Friday, 2015-10-30

*** jjardon has quit IRC03:06
*** jjardon has joined #baserock03:13
*** gtristan has quit IRC03:57
*** gtristan has joined #baserock04:36
*** gtristan has quit IRC04:48
*** gtristan has joined #baserock04:48
*** gtristan has quit IRC07:37
*** CTtpollard has joined #baserock08:02
*** gtristan has joined #baserock08:17
*** paulwaters_ has joined #baserock08:51
*** paulwaters_ has quit IRC08:54
*** bashrc has joined #baserock09:08
* pedroalvarez takes 5 minutes of his morning to deploy Gnome to a Jetson09:13
radiofreeThat isn't going to display anything09:14
radiofreeIt *might* display something using the fbdev X driver09:15
radiofreeBut it'll be slow09:15
radiofreemutter wayland won't work without patching either09:16
pedroalvarezbah09:16
radiofreeDoes the gnome system have gcc etc?09:17
radiofreepatch for mutter shouldn't take too long09:17
pedroalvarezI'd gess so09:17
*** toscalix__ has joined #baserock09:31
*** jonathanmaw has joined #baserock09:38
*** tiagogomes_ has joined #baserock09:42
perrylwhen upgrading a kvm trove, what form does the upgrade-location take? is it just the vm name, or qemu:///system/trove-name? does it require the host system ip address?09:44
Kinnisondon't upgrades tend to be done by sshing in?09:44
perrylssh-rsync09:45
SotKit takes the IP of the system being upgraded IIRC09:45
*** mariaderidder has joined #baserock09:46
pedroalvarezgtristan: I expect some things to be wrong, given that I've upgraded an existing system (not a clean deployment), but I'm getting this error: http://paste.baserock.org/ikonopakot09:49
jjardonHi, if someone has a jetson at hand, maybe you can give a try to this series I did some time ago when I had access to a jetson: https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:jetson_linux_4_209:54
* pedroalvarez can provide jetsons :)09:55
pedroalvarez(not to everyone though)09:56
* pedroalvarez tests them09:56
gtristanhmmm10:02
gtristanpedroalvarez, it worked before and now broken ?10:02
gtristanpedroalvarez, let me check my journal, however accountsservice should not need to be started10:03
gtristanas its enabled and starts already, not on demand10:03
*** franred has joined #baserock10:04
jjardonpedroalvarez: thanks, they are completely untested though; I was hopping to have time to play with them but then I got dragged to another project10:04
pedroalvarezjjardon: I'm going to test a genivi-demo-platform system10:04
jjardonpedroalvarez: if you have some free time, it would be great if you can take a look to this as well: https://gerrit.baserock.org/#/c/1129/10:07
jjardonpedroalvarez: are you using a jetson system with current master of definitions? is 3D acceleration working? If yes nice to know that upstream mesa/libdrm works out of the box with the jetson10:10
pedroalvarezI believe that it's working10:11
pedroalvarez(I believe, because maybe I don't know how to differentiate)10:11
gtristanpedroalvarez, I dont know whats up, my build doesnt do that10:15
gtristanit does say "activating via systemd: service name='org.freedesktop.Accounts' unit='accounts-daemon.service'10:17
gtristanpedroalvarez, this is a separate system for the jetson ? (i.e. != x86_64) ?10:17
pedroalvarezyes10:18
gtristandid you miss including something perhaps ? or maybe system integration hooks didnt run properly ?10:18
jjardonHi, can someone give another look to https://gerrit.baserock.org/#/c/1326/ ? Its blocking all the gtristan work :)10:18
pedroalvarezit might be the system-integration hooks, yes10:18
gtristancan you log into the system then I guess ? try systemctl enable accounts-daemon and restart10:19
paulsher1oodjjardon: merged10:20
pedroalvarezgtristan: I removed the system, to use the jetson for other things, I'll try that later10:21
gtristangot it10:21
*** toscalix__ has quit IRC10:21
pedroalvarezI did't think it would work after radiofree's comments10:21
jjardongtristan: did you see my comment in https://gerrit.baserock.org/#/c/1331/ ? maybe you want to add grilo-plugins as well10:21
gtristanpedroalvarez, I wont start thinking about this now, but I think we can get away with not having a per-arch system definition really10:22
gtristanjjardon, I'll take a note, thanks10:22
*** edcragg has joined #baserock10:23
gtristanthere needs to be another pass, probably we want tracker in there; some of the gnome packages so far desire cheese as well10:23
gtristanand have to take another look at wacom, and also deal with making geoclue work - I think there is also an EDS calendar crash10:23
gtristannautilus would also like to have some exotic parsers, one of which desires boost10:24
gtristanboost took hours to fail to compile here :-S10:24
pedroalvarezgtristan: it might need some conditionals, for bsps and other things arch specific10:25
gtristanpedroalvarez, indeed, I thought the bsp was pretty much all there is to it though10:26
gtristanmaybe some specific system integration extensions, but if those, themselves could be included in the bsp morphology, that covers it I think10:27
gtristanwe could just run ybd.py clusters/gnome-system.morph <arch>10:28
gtristanand the <arch> would define the flavor of bsp to include10:28
pedroalvarezthere might be various bsps for the same arch10:28
pedroalvarez<bsp> would make more sense there10:29
pedroalvarezbut I'm not sure about this10:29
gtristanah, ok... as I said; not thinking of that yet :)10:29
pedroalvarezyes yes, let's leave this for now10:29
gtristanif I keep adding things, wont get anywhere :)10:29
jjardonpaulsher1ood: thanks, this is for you: https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:gnome_jetson10:39
paulsher1oodheh10:40
radiofreemaybe test that it's usable with the x11 fbdev driver first?10:43
radiofreeor try http://cgit.freedesktop.org/nouveau/xf86-video-nouveau/ (no idea)10:44
pedroalvarezyes please, it makes no sense to have it on the ci, it won't build10:45
pedroalvarezit needs some fixes10:45
pedroalvarezand we don't know that it works yet10:45
pedroalvarezjjardon: your jetson_linux_4_2 doesn't build10:46
pedroalvarez*patch10:46
jjardonpedroalvarez: bah; dont worry, I will try to fix it myself next week10:47
*** locallycompact has joined #baserock10:47
gtristanthanks for the merge !10:47
jjardonpedroalvarez: what thing is not going to build?10:48
pedroalvarezi would expect you to know given that you sent the patch :)10:48
pedroalvarezjjardon: see "baserock/pedroalvarez/gnome-arm" in definitions.git10:49
*** toscalix__ has joined #baserock10:50
jjardonpedroalvarez: I only see a problem with mutter there: no a problem with the system definition itself10:51
pedroalvarezbut it won't build10:52
jjardongtristan: you forgot the build-system parameters again :) Fixed now10:52
gtristandid I ?10:52
gtristansorry :-/10:52
jjardonI make the same mistake often as well, maybe is not as user friendly as it should as it depends if you have a morph file or not10:53
jjardonpedroalvarez: thanks for the path, now it will build10:58
pedroalvareznote that my quick attempt to run it faileed11:09
pedroalvarezjjardon: re jetson linux 4.2 http://paste.baserock.org/inopoqufej11:12
pedroalvarezI'll post this in the patch11:12
pedroalvarezoh11:13
radiofreewell tested i see11:13
pedroalvarezmight be that the "sed -e 's/.*android\/sync.*/#ifdef CONFIG_SYNC\n&\n#endif/' -i drm/nouveau/nouveau_fence.c" is not needed anymore11:13
pedroalvarezradiofree: hm?11:14
radiofreethis upgrade to 4.2 patch11:14
pedroalvarez:)11:14
pedroalvarezthere is one comment in the patch saying that it hasn't been tested11:15
radiofreewhy submit it then?11:15
pedroalvarezI guess that question is not for me11:16
jjardonradiofree: because it was in my laptop and I want to share it in case someone fancy continue the work; I clearly state that it was not tested at all11:17
radiofreeit's just noise then, post the patches to the mailing list11:17
jjardon?11:17
jjardonwe use gerrit in this project, so I post the patches there11:19
SotKDo we need a "Workflow" voting column for people to vote -1 in if their patch is a WIP like OpenStack have?11:20
jjardonthat looks useful11:21
pedroalvarezno, I have no idea how to fix this nouveau error11:22
pedroalvarezintroduced here: http://git.baserock.org/cgi-bin/cgit.cgi/delta/nouveau.git/commit/?h=baserock/tegra/4.2&id=bb777b20b2a0bb90e487f203f3d618e0532f419511:23
pedroalvarezo wait, I think I was right, and we don't need that `sed` anymore11:25
pedroalvareznope, nouveau doesn't build11:48
* pedroalvarez leaves it11:49
*** edcragg has quit IRC12:11
*** edcragg has joined #baserock12:28
*** gtristan has quit IRC12:40
*** franred has quit IRC12:41
*** gtristan has joined #baserock12:53
*** JPohlmann has quit IRC13:04
*** toscalix__ has quit IRC13:17
*** gtristan has quit IRC13:24
*** locallycompact has quit IRC13:25
*** franred has joined #baserock14:04
*** locallycompact has joined #baserock14:41
*** toscalix__ has joined #baserock14:43
*** franred has quit IRC14:53
*** gtristan has joined #baserock14:53
*** JPohlmann has joined #baserock15:01
*** JPohlmann has joined #baserock15:01
*** franred has joined #baserock15:07
*** gtristan has quit IRC15:21
*** gtristan has joined #baserock15:23
*** franred has quit IRC15:35
*** toscalix__ is now known as toscalix15:36
toscalixHi guys, I have been asking questions about maintenance on the dev ML to make sure we provide an accurate message to GNOME15:37
toscalixI want to make sure that I can have a meeting at Codethink wehere we talk about that compromise to see how far we can get15:38
toscalixfeel free to provide me your opinions or estimations directly instead of the public mailing list15:38
toscalixif you feel more comfortable about it15:38
paulsher1oodtoscalix: s/compromise/commitment/15:38
toscalixI undertsnad that sometimes providing estimations in public channels can be challenging15:39
toscalixyes...commitments sorry15:39
toscalixcompromiso in Spanish....a common mistake for Spaniards, yes15:39
toscalixestimations out of context can be taken as commitments15:40
*** mariaderidder has quit IRC15:53
*** CTtpollard has quit IRC15:55
*** edcragg has quit IRC15:58
*** edcragg has joined #baserock15:59
*** paulwaters_ has joined #baserock15:59
*** mariaderidder has joined #baserock16:08
pedroalvarezwow, I can trigger  pipelines in gocd for changesets in gerrit16:29
pedroalvarezand it's possible to configure it to give some feedback to gerrit16:29
pedroalvarezsame for github, and stash16:30
pedroalvarezif only I could put elastic agents..16:31
jjardonHi pedroalvarez ; did you have some time to see if its possible to put more RAM in mason in the end?16:58
pedroalvarezjjardon: I put 10G of swap16:58
pedroalvarezfor now until I have time to put more ram16:58
pedroalvarezalthough adding only ram, and no swap might not be enough16:58
jjardonpedroalvarez: thanks!, let's see how it goes16:59
pedroalvarezyup16:59
pedroalvarezI'd like to try this gocd premerge check, but I think it will be really slow17:00
pedroalvarezunless we put a gocd agent in a huge vm17:01
pedroalvarezbut gocd doesn't have elastic agents support yet17:01
*** paulwaters_ has quit IRC17:03
*** jonathanmaw has quit IRC17:29
*** tiagogomes_ has quit IRC17:40
paulsher1oodpedroalvarez: pre-merge is bound to be slow for low-level changes, though?17:44
pedroalvarezpaulsher1ood: yes, although it might be possible to have different pipelines, one that builds something minimal, other that builds something bigger, and so on17:49
paulsher1oodpedroalvarez: yup, that would be preferable. settle for the bigger set being done slowly+cheaply17:51
pedroalvarezi'd prefer fast+cheap17:52
pedroalvarez:)17:52
paulsher1oodwell, if that's possible, so would i :)17:53
paulsher1oodassuming gocd can run python programs, couldn't we use boto to start a pre-existing vm, then run the workload on it, then shut it down at the end?17:57
pedroalvarezhm.. it would be better if the orchetration knew that there are no more jobs in the queue and then shut it down18:00
pedroalvarezabout boto and starting VM's I have no idea18:00
pedroalvarezbut there should be API's for the clouds I know (openstack and aws)18:00
paulsher1oodboto is the normal python way to drive aws api i think18:01
pedroalvarezit's nice that you can create dependencies between pipelines: http://www.go.cd/documentation/user/current/resources/images/dependent_build.png18:02
*** bashrc has quit IRC18:03
*** mariaderidder has quit IRC18:04
*** locallycompact has quit IRC18:05
*** toscalix has quit IRC18:07
*** edcragg has quit IRC18:18
*** lostduck is now known as straycat18:44
persiaOn pre-merge being slow: how slow?  Human review times are usually measured in days, so the price merge build only in eeds to be faster than that.22:03
persiaThe goal being that the automation does not significantly slow landing, rather than being about providing fast feedback to developers.22:04
persiaDevelopers wondering if things work ought be able to run a desired subset of tests without submitting to gate (ideally with cached builds so that subsequent submission is faster)22:05
persias/price merge only in eeds/pre-merge only needs/22:06

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