IRC logs for #baserock for Thursday, 2016-01-14

*** Stanto has quit IRC02:52
*** Stanto has joined #baserock02:55
*** gtristan has quit IRC06:25
*** gtristan has joined #baserock06:53
*** edcragg has joined #baserock08:12
*** bruce_ has joined #baserock08:28
*** CTtpollard has joined #baserock08:32
*** toscalix has joined #baserock08:42
*** bashrc_ has joined #baserock09:04
*** rdale has joined #baserock09:26
*** edcragg has quit IRC09:35
*** jonathanmaw has joined #baserock09:50
*** edcragg has joined #baserock09:51
*** gtristan has quit IRC09:51
*** Lachlan1975 has joined #baserock10:28
*** gtristan has joined #baserock10:34
mwilliams_ctthis is potentially a silly question, but I'm having trouble with it conceptually. Take the base system as an example for this: it consists of strata for build-essential, core, foundation and bsp10:42
mwilliams_ctcore depends on build-essential which is fine but foundation depends on coreutils-common as well10:42
mwilliams_ctwhy is coreutils-common not in the system morphology?10:42
richard_mawbuild time only dependency?10:43
mwilliams_ctrichard_maw: the significance being that it doesn't need to be in the final system, so it won't be even though it's built?10:44
tiagogomesmwilliams_ct to reduce the size of the deployed image?10:44
mwilliams_ctOK, I grok. Sorry that wasnt intuitive to me though makes sense now. Thanks :)10:45
tiagogomeshey, nobody said that it was supposed to be intuitive :)10:46
mwilliams_ct:)10:46
*** rdale has quit IRC11:34
*** rdale has joined #baserock11:45
radiofreeis it possible to tell morph/ybd *not* to cleanup in the event something goes wrong during an upgrade11:47
gtristanpaulsherwood, so I got passed all that chroot issues, and have arrived at my next roadblock, which is an annoying one12:19
gtristanfirst, the chroot issues are not that bad, but there needs to be some things populated in /bin at least for a build to happen, partly (mostly?) because of scripts which do hashbangs expecting #!/bin/sh12:20
gtristanthat's not a hard obstacle anyway, things build in the sandbox with a mounted/prioritized distcc nicely12:20
gtristanthe next obstacle worries me because it's a political standpoint of aboriginal, which is: reject all things GPLv312:21
gtristanwhich means: Dont provide a compiler new enough to build the glibc we want to build12:22
gtristanSo, that may mean that we will have no choice but to take ownership of that initial bootstrapping process after all12:23
gtristanat the very least, right now I have to build a recent cross compiler12:23
radiofreeis building "initramfs-test" and making sure it boots a good enough test for that?13:32
radiofreehas that type of image ever been upgradable?13:32
pedroalvarezImages with intramfs? There are also a couple in release.morph13:34
radiofreewell there's an "initramfs-test" cluster as well13:35
pedroalvarezI believe they are upgradeable, although I don't remember where the image goes in the FS13:35
radiofreedo you have an example of an upgrade cluster for an initramfs system?13:36
radiofreei guess just change rawdisk to ssh-rsync and try?13:36
pedroalvarezThat should be it13:37
pedroalvarezHmm.. Would it work if /boot is in another partition?13:39
radiofreesystem-version-manager has code to handle initramfs13:39
radiofreei'm not going to try a partitioned initramfs system....13:39
pedroalvarezHeh, yeah, I don't think it will work13:40
radiofreewhat i've tested so far, jetson partitioned (deploy + upgrade), x86 non partitioned (deploy + upgrade), x86 partitioned system (now works! deploy + upgrade)13:40
radiofreei'll just make sure initramfs system don't get broken with these changes13:40
pedroalvarezYup, that sounds good :)13:43
pedroalvarezAlso, if intramfs can be put in different places, if in a partition doesn't work13:44
pedroalvarezs/if//13:44
radiofreethe initramfs upgrade failed :\14:02
radiofreebut at the system-version-manager stage14:02
radiofreewhich i haven't touched...14:02
*** edcragg has quit IRC14:02
radiofreedebugging things at this stage is a total pain in the arse14:03
* radiofree does it anyway14:03
*** edcragg has joined #baserock14:04
radiofreepedroalvarez: the good news is it doesn't work without my changes anyway14:24
radiofreeso i guess i don't have to spend any time fixing this atm14:24
pedroalvarezI agree14:34
pedroalvarezCould you create an story for that bug in https://storyboard.baserock.org/#!/dashboard/stories ?14:35
*** gtristan has quit IRC14:37
*** gtristan has joined #baserock14:54
*** jonathanmaw has quit IRC15:26
*** jonathanmaw has joined #baserock15:27
*** jonathanmaw has quit IRC15:42
*** jonathanmaw has joined #baserock15:45
*** nowster has quit IRC15:52
*** nowster has joined #baserock15:53
*** franred has quit IRC16:03
* richard_maw sent a mail to the list about what's blocking being able to have lorry import gcc with a shallow clone16:04
richard_mawin general: shallow push is broken, there's ways we can work around it, but they are more complicated than fixing git and should only be done if we need to work with old versions of git16:09
richard_mawand I don't have time to do either16:09
richard_mawso I thought I'd write up what I could while I have the time16:09
richard_mawand be able to answer questions16:09
* richard_maw will continue implementing what he can, but it's likely the result will have the caveats: 1. breaks downstream troves, 2. can't unshallow, 3. can't detect fast-forwards16:29
richard_mawoh, and anyone else's shallow repositories won't be handled correctly either, but nobody else is doing them16:36
radiofreeif you have submodules in a repo, the only way to have the clone from gbo is to modify the branch right?16:39
richard_mawyes16:39
radiofreeok16:39
radiofreewhat about things like http://git.baserock.org/cgi-bin/cgit.cgi/delta/qt5/qtwebengine.git/tree/.gitmodules?h=0.916:40
radiofreewill that clone from gbo/delta/qt5/qtwebengine-chromium.git automatically then?16:40
pedroalvarezI don't like that approach :/16:41
radiofreethat isn't my approach, i haven't modified the repo16:42
pedroalvarezyeah, I know some upstreams do that16:42
radiofreeisn't that better than modifying the branch though?16:42
radiofreeit's going to be the same as us modifying it to to exactly the same thing?16:42
pedroalvarezyeh, it's better.. if you are lucky that we put the submodule-repo in the right path16:43
*** bruce_ has quit IRC17:01
*** jonathanmaw has quit IRC17:05
CTtpollardthere is currently discussion in #yocto around shallow cloning17:13
pedroalvarezradiofree: thanks for the patches17:36
pedroalvarezI will be able to continue the RPi work17:36
pedroalvarezradiofree: btw, I never managed to use extlinux there, even using the same uboot as in the jetson. Do you know what can be the issue there? or am I dreaming when thinking that given that it works in jetson it should work in RPi?17:37
radiofreefor some reason people disable sysboot in the default configs17:39
radiofree(that was the case for the wandboard for example)17:39
radiofreemight just need to enable it there17:39
radiofreeCONFIG_CMD_PXE needs to be enabled... surprised that it wasn't. did you manually try to "sysboot"?17:43
edcraggit probably needs to be a fairly recent version i imagine. should be just a question of whether the sysboot command is available or not17:45
pedroalvarezhaven't tried anything else than puting an extlinux.conf file and just try17:46
edcraggyou can easily check if it's available by typing the command into the u-boot shell17:47
radiofreeeach boards default boot commands are different17:49
radiofreeso on the rpi it might not bother to call sysboot17:49
radiofreecould probably solve that by placing a /boot.scr that does though (since i imagine they all will attempt to source that)17:49
radiofreeqt 5.5.1 upgrade was.... seamless17:50
edcraggindeed, this all completely depends on the uboot environment, though17:50
edcraggit will have to either load a script which does so, or have a boot command in the environment that explicitly tries to boot with sysboot17:51
radiofreeedcragg: a quick grep of the tree shows how ubiquitous boot.scr is17:51
edcraggyep, not universal though, i've seen all sorts of things, particularly in oem releases17:52
radiofreeheh, yeah17:52
radiofreeheh, yeah, pandora board prefers "autoboot.scr"17:53
radiofreeyou should be able to do that yourself though, i.e for your board support place a boot.scr/autoboot.scr whatever... in the correct location that calls sysboot17:53
edcraggyou can also compile a uboot environment to a binary and dd it to the right place on the boot flash to overwrite the environment17:54
pedroalvarezheheh, I will for sure ask you whenever I get to it17:55
pedroalvarezI know some people has managed to get it working, but not with upstream uboot17:55
*** CTtpollard has quit IRC17:57
edcraggthere's usually a massive amount of patches on oem u-boot releases, so much that i've never successfully got mainline u-boot running on anything that i've looked at :P17:58
*** bashrc_ has quit IRC18:02
radiofreelet's see if qtwebengine works18:07
radiofreegotta do http://git.baserock.org/cgi-bin/cgit.cgi/delta/qt5/qtwebengine.git/commit/?h=baserock/v5.5.118:07
radiofree"16-01-14 00:01:06 [1/17/307] [weston-system-armv7lhf-jetson] ERROR: unable to get cache for qtwebengine"18:11
* radiofree gives up for the evening18:14
*** Lachlan1975 has quit IRC19:31
*** edcragg has quit IRC19:41
*** rdale_ct has joined #baserock20:00
*** rdale has quit IRC20:00
*** rdale_ct has quit IRC20:04
*** rdale has joined #baserock20:05
*** toscalix has quit IRC22:30
*** edcragg has joined #baserock23:10

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