**** BEGIN LOGGING AT Fri Jul 22 02:59:58 2016 Jul 22 08:35:57 morning Jul 22 09:01:24 likewise Jul 22 10:34:36 Does a bb.error() in a Python task mark it as failed, or does it just print a message? Jul 22 10:34:48 I have the impression it's the latter, but if it is, how do I make a task fail? Jul 22 10:35:43 Nevermind, it's right there in the docs Jul 22 10:35:51 There's even a big fat Note right next to bb.error() Jul 22 10:42:44 what's the best way to stop a package in an inherited images packagegroup not being built? I can think of a few but want to do it the most sane way Jul 22 10:44:00 Not inheriting the image but defining your own? Jul 22 10:44:37 not possible here Jul 22 10:45:08 can you append to packagegroup bb? Jul 22 11:04:13 yeah, replace or append the packagegroup Jul 22 11:14:24 rburton: I'll probably go down that route, on a second point, best way to find out why a do_fetch is taking an age? Jul 22 11:14:52 verbose doesn't seem to do the trick Jul 22 11:16:09 use master and you get a nice progress meter! Jul 22 11:16:33 the obvious answer is because its fetching some which is either slow or giant, or both Jul 22 11:16:37 again, limited to krogoth right now Jul 22 11:16:52 I'm used to fetching the kernel taking long, but this seems OTTT Jul 22 11:17:04 if it ends up attempting to download a huge git clone tarball from the sources mirror then that occasionally rate-limits a bit aggressively Jul 22 11:21:06 I'll try fetching it manually outside of bitbake to see if it's throttling Jul 22 11:25:02 (linux-yocto-4.4 from git.yocto) Jul 22 11:29:16 remember I said yesterday, that my image for qemu just booted and then shut itself off after 10 seconds or so without any pre warnings? Well, the same thing happens on native intel, so now its culprit hunting-time. Jul 22 11:31:32 RP: i'm around if you want to discuss my patch... Jul 22 11:49:36 armpit: have you looked at the krogoth failures on the AB? Jul 22 12:08:01 What does the license on a image recipe imply? A image is no more than a collection of packages with their own licenses, right? Jul 22 12:12:41 sveinse: it's the source code of something, so it needs a license Jul 22 12:17:31 boucman_work: The patch seems reasonable, I just need to decide whether its the correct form of the syntax we want to support. I'm mildly worried that its an immediate operation, compared to some operations which are deferred but I think that should be ok Jul 22 12:20:30 My image inherit core-image. When I look at core-image-minimal.bb, I see that it depends packagegroup-core-boot, which has the description "The minimal set of packages required to boot the system". With this title, I'd expect images to want to depend on this recipe, yet the images /does/ boot without it. Why is that? Jul 22 12:20:45 yeah, I can understand that problem... when I tested the patch I used random variables and most of them broke yocto. Yocto sets a default value and then expect it to be defined... but OTOH defered deleting of variable makes little sense Jul 22 12:21:14 (and the version I sent to the ML had documentation. I think it's still blocked in the moderation queue, but i'd gladly have some proofreading of that too) Jul 22 12:32:33 hi guys Jul 22 12:37:37 neverpanic: is maybe bb.fatal() what you want? (I just rewrote some of my recipes to use it after reading your comment) Jul 22 12:37:41 i'm trying to update from jethro to krogoth but i'm having issue with my installation iso. actually it's no longer working. I used to have a PACKAGE_INSTALL_append = " live-install-m57" with a live-install-m57.bb recipe that contained a install-m57.sh script Jul 22 12:38:31 I also tried to run a core-image-minimal but it is only working if i leave the AUTO_SYSLINUXMENU to 1, if i set it to 0, it's not working as well Jul 22 12:42:47 sveinse: license in an image recipe is the license of the recipe Jul 22 12:43:11 m57n: does the image not boot anymore ? Jul 22 12:43:17 sveinse: packagegroups are just ways to group. its entirely reasonable to build an image that works without using them. Jul 22 12:44:35 boucman_work: no Jul 22 12:47:06 rburton: well, my intentions is "give me a minimal system" so I use ${CORE_IMAGE_BASE_INSTALL} which core-image provides. Which seems to be fine. Yet the logic in core-image-minimal.bb confuse me, as it throws something else into the mix apparently Jul 22 12:55:13 during a do_rootfs on kergoth 'error: Can't install packagegroup-core-tools-debug-1.0-r3@all: no package provides gdbserver' ? Jul 22 12:56:33 my image inherits tools-debug, never seen this error before Jul 22 12:56:58 boucman_work: I approved it yesterday, it made the mailing list Jul 22 12:58:07 oh ? hmm Jul 22 12:58:11 and gdb have definitely been built to provide it... any ideas? Jul 22 12:58:12 * boucman_work go check his mail client Jul 22 13:06:34 rburton: the maintainers file leads me to you I believe Jul 22 13:12:11 CTtpollard: oh fun. yeah gdb for reasons nobody has yet managed to figure out randomly decides to not build the server. bitbake gdb -C unpack will likely bring it back. Jul 22 13:12:20 (basically just force it to rebuild) Jul 22 13:12:46 CTtpollard: if you have the work directory for the build which didn't produce gdbserver then it would be good to copy it somewhere safe first! Jul 22 13:12:55 CTtpollard: then you can run diff on both and see what the difference was Jul 22 13:13:07 rburton: let me see Jul 22 13:14:02 i've seen it once on a build where i couldn't recover the build tree Jul 22 13:14:25 ok taken a backup of it Jul 22 13:14:51 boucman_work: have you verified that the unset syntax solves your problem? Jul 22 13:16:26 rburton: also TIL -'C' Jul 22 13:16:46 CTtpollard: -C is awesome Jul 22 13:19:24 rburton: so, in that sense is it just forcing the package to start the stages again starting at do_unpack? Jul 22 13:19:32 yeah Jul 22 13:19:45 well Jul 22 13:19:57 marks the task you specify as needing to happen, and then does a build Jul 22 13:20:25 so you can do -C install and it would re-run install / package / sysroot Jul 22 13:20:53 but i tend to use it for "rebuild this recipe" as its quicker than bitbake -ccleansstate foo && bitbake foo Jul 22 13:21:33 rburton: great, I thought at first it basically going to do clean and then do unpack, but then it carried onto compile etc Jul 22 13:24:49 rburton: that has indeed fixed it, thanks again Jul 22 13:25:08 CTtpollard: did you manage to save the work tree for the old build? Jul 22 13:26:46 rburton: yup Jul 22 13:27:57 rburton: I dumped the whole work dir for gdb, happy to provide you with some info from it Jul 22 13:28:40 CTtpollard: verify that the one you saved didn't produce a gdbserver package first, but if you can diff the log.do_configure files in broken and working then that would be appreciated Jul 22 13:29:03 non-deterministic builds upset me Jul 22 13:32:33 rburton: ok first siren, work dir for 'non-gdbserver gdb' doesn't have any logs for do_configure, the 'working' one does however Jul 22 13:33:24 under the temp subdir Jul 22 13:33:50 meta/recipves-bsp/alsa-state/alsa-state.bb is apparently built for being overriden by local BSP. Does anyone know about any (correct) examples of how to do so? Jul 22 13:35:23 CTtpollard: oh maybe it pulled gdb out from a sstate and the actual logs are lost. Jul 22 13:59:19 rburton: I'm sharing sstate :/ Jul 22 14:00:25 CTtpollard: if you ever see it again you now know the drill :) Jul 22 14:08:03 Are there any bb vars to access the work dir in a recipe? E.g. if I want to write out to a logfile in temp/ Jul 22 14:08:43 bitbake.conf is your friend Jul 22 14:08:49 ${WORKDIR} is the work dir Jul 22 14:08:55 and ${T} is temp/ Jul 22 14:09:00 I'll find my friendship, thanks Jul 22 14:15:39 rburton: yes, I've tested that... did I miss something ? Jul 22 14:31:57 RP / rburton : do any automated tests cover linux-yocto-dev? Asking 'cause 3e0137113e894eb / OE 5c21fd5eb8b68 from you two breaks linux-yocto-dev outright here for me. Jul 22 14:32:37 I just get the crappy ancient 3.7 commit ID hard coded as the fallback in the linux-yocto-dev.bb -- reverting the above and all is good. Jul 22 14:36:15 paulg_: no, they likely don't Jul 22 14:36:34 paulg_: Its possible I've messed that change up :/ Jul 22 14:39:50 armpit: shame about the errors on the krogoth build :( Jul 22 14:50:21 RP, fwiw only thing I've got here related afaik is... Jul 22 14:50:24 PREFERRED_PROVIDER_virtual/kernel ?= "linux-yocto-dev" Jul 22 14:50:55 no other magic, other than the basic switch to the dev kenrel. Jul 22 14:55:31 paulg_: I think I must have checked it was parsing each time but not that the revisions were working :/ Jul 22 14:56:28 Must package names always be prepended by the recipe name? Jul 22 14:56:37 sveinse: no Jul 22 15:02:22 Must newly added classes/ directory be added to conf/layer.conf? The BBFILES var? Jul 22 15:04:59 sveinse: its found through BBPATH Jul 22 15:05:16 paulg_: I can't seem to make this not always get parsed :( Jul 22 15:20:55 paulg_: I'll revert it, I have another way to fix this Jul 22 15:22:24 Note to self: Using MAKE_prepend to add a make filter was not a good idea, as it exports MAKE which in turn use this prepended command, which was supposed to be run once Jul 22 15:23:21 Pity there is no ${EXTRA_OEMAKE_PRE} variable Jul 22 15:23:58 paulg_: patches on list Jul 22 15:44:26 RP, yeah looking at it. I so far see the checkurl failures. Jul 22 15:45:29 armpit: some of the devtool/recipetool failures will be due to libmatchbox changing, there's patches in master you'll want to pick Jul 22 15:46:12 k. Matchbox.. oooh, fire Jul 22 15:47:18 * armpit logs bugs Jul 22 15:48:09 ooi, why do I get double line shifts on compile output when using bitbake -v ? Jul 22 15:48:52 they are in the logs as well, it seems Jul 22 15:56:45 hi there, what is the best approach to generate a bootable image with a rootfs greater than 4gb? someone could recommend some link to read? Jul 22 16:09:20 inherit pkgconfig and PACKAGECONFIG is two unrelated things, right? Jul 22 16:19:25 sveinse: correct Jul 22 16:25:48 armpit, RP: I saw two Fedora-24 fails on krogoth-next AB Jul 22 16:31:57 Can I control the order of the package-splitting? Is it following the order in ${PACKAGES}? Because it will have effects on how file globbing will slurp the files Jul 22 18:26:07 Sorry if this isnt allowed, but I was wondering if anyone could help me out. Im having some embedded system issues Jul 22 19:07:41 khem: http://errors.yoctoproject.org/Errors/Details/73012/ <— e2fsprogs in mut fails under musl Jul 22 19:09:58 oh hardening again Jul 22 19:10:05 i guess that needs to be glibc only? **** ENDING LOGGING AT Sat Jul 23 02:59:58 2016