**** BEGIN LOGGING AT Thu Jan 25 03:00:06 2018
Jan 25 03:39:32 <xthundermobilex> I could use some help with a recipe I'm working on for dbus-cxx which isn't currently available in TO.
Jan 25 03:39:58 <xthundermobilex> The recipe fetchs and builds fine.
Jan 25 03:40:28 <xthundermobilex> The library and headers are packaged fine and make it into the deployable image
Jan 25 03:40:41 <xthundermobilex> And the target sysroot
Jan 25 03:41:21 <xthundermobilex> But the headers are placed in a subdirectory of /usr/include
Jan 25 03:42:15 <xthundermobilex> As a result, when a new eclipse project can't find it.
Jan 25 03:42:56 <xthundermobilex> I've tried 6 way from Sunday to figure out how to fix that a
Jan 25 03:43:12 <xthundermobilex> To no avail
Jan 25 03:44:28 <xthundermobilex> Should I change where the header gets installed into the image?
Jan 25 03:45:33 <xthundermobilex> Would I do that with a c
Jan 25 03:46:03 <learningc> how do I completely restart bitbake my image?
Jan 25 03:47:52 <xthundermobilex> learningc: bitbake -c clean would do it I beleave
Jan 25 03:47:57 <xthundermobilex> Believe
Jan 25 03:48:47 <xthundermobilex> But I think that removes everything from the download directory
Jan 25 03:49:16 <xthundermobilex> So it ends up fetching all the packages.
Jan 25 03:49:24 <xthundermobilex> Again
Jan 25 03:52:30 <xthundermobilex> Any tips would be appreciated. Would really like to figure this out so I can use it myself and submit the recipe to YP
Jan 25 03:57:16 <learningc> Somehow it des not do it
Jan 25 03:57:39 <learningc> bitbake will read something from cache
Jan 25 04:05:45 <learningc> Does -c clean cleans the cache?
Jan 25 04:10:54 <learningc> Otherwise how can I clean all including the cache?
Jan 25 04:13:45 <xthundermobilex> Im pretty sure it cleans the whole nine yards.
Jan 25 04:14:42 <xthundermobilex> Give me a second and
Jan 25 04:15:24 <xthundermobilex> I'll loo, it up in sre
Jan 25 04:15:53 <xthundermobilex> I'll look it up in Streif
Jan 25 04:20:33 <xthundermobilex> Looks like bitbake -c cleanall flushes everything.
Jan 25 04:21:17 <xthundermobilex> Might try a cleanstate first.  Looks like it leaves the downloads in t
Jan 25 04:21:34 <xthundermobilex> Intact.
Jan 25 04:28:42 <learningc> Does it clean whatever is in build/tmp/cache as well?
Jan 25 04:28:48 <learningc> For me it does not
Jan 25 04:48:30 <xthundermobilex> So bitbake -c cleanall <image> doesn't clean yo
Jan 25 04:49:00 <xthundermobilex> Or image?
Jan 25 04:49:51 <xthundermobilex> Dood I'm a noob really so I can't answer defi i
Jan 25 04:49:58 <xthundermobilex> Nj
Jan 25 04:50:25 <xthundermobilex> Definitively.
Jan 25 04:50:35 <xthundermobilex> I
Jan 25 04:51:06 <xthundermobilex> As man ... I hate this Samsung touch it be.
Jan 25 04:51:21 <xthundermobilex> Arrrrrrggghhh!
Jan 25 04:51:31 <xthundermobilex> Thing stinks
Jan 25 04:52:08 <xthundermobilex> Man was not meant to type with his/her thumbs ...
Jan 25 04:53:14 <xthundermobilex> While working on a recipe recently, I did clean t
Jan 25 04:53:57 <xthundermobilex> It multiple times with -c cleanstate or -c clean wh
Jan 25 04:54:32 <xthundermobilex> Which seemed to work as expected.
Jan 25 08:51:02 <sveinse> kergoth: how do you do it when sources and layers are the same, e.g. in the case of mono-repo. Then there is no distinction between the layer and the source. LIC_FILES_CHECKSUM should then point to the layer license
Jan 25 08:53:03 <sveinse> But I suppose Yocto isn't really designed for a mono-repo type thinking, where layer and source go together
Jan 25 08:58:42 <guzu> kanavin, hello, using the direct boost.org download link worked for me
Jan 25 08:58:57 <guzu> is there a document describing how to contribute a patch?
Jan 25 09:01:37 <maxin> guzu: this could help: http://www.openembedded.org/wiki/How_to_submit_a_patch_to_OpenEmbedded
Jan 25 09:03:34 <guzu> thank you maxin , i was not sure that's the right way
Jan 25 09:41:24 <guzu> ok, how do i get access to upload changes for review?
Jan 25 09:43:56 <nayfe> guzu:  access to what?
Jan 25 09:44:20 <guzu> it's boost change
Jan 25 09:44:27 <guzu> nayfe, ^^^
Jan 25 09:47:17 <nayfe> guzu: i think it depends if your modification is Yocto integration or Boost modification
Jan 25 09:47:39 <guzu> nayfe, the change is in yocto recipe
Jan 25 09:48:00 <guzu> i change the SRC_URI from sf.net to boost.org
Jan 25 09:48:40 <guzu> i want this to be reviewed because it failed to me often recently, and on the other hand such a move will put some pressure on boost.org downloads server
Jan 25 09:48:49 <nayfe> guzu: boost recipe is in poky/meta so it's OE-core, you need to register mailing list
Jan 25 09:48:57 <guzu> ok
Jan 25 09:49:10 <nayfe> http://lists.openembedded.org/mailman/listinfo/openembedded-core
Jan 25 09:49:13 <guzu> thank you very mcuh nayfe
Jan 25 09:49:17 <guzu> much
Jan 25 09:49:56 <nayfe> guzu: np
Jan 25 09:51:29 <ttllkk> I have a recipe named as 'myrecipe-native.bb' that inherits native, and it is working fine. I would like to include it in the sdk. how should the recipe be naemd so that it can be built as -native and as nativesdk?
Jan 25 10:00:46 <ttllkk> In order words I want a recipe that can be build as -native or nativesdk- but not for the target
Jan 25 10:05:18 <nayfe> ttllkk: maybe you need to add it to TOOLCHAIN_HOST_TASK ?
Jan 25 10:35:06 <ttllkk> nayfe: I think that is it, thanks
Jan 25 10:35:28 <nayfe> <ttllkk> np :)
Jan 25 10:49:27 <ttllkk> Still not sure about how to define a recipe for a tool that can be compiled as -native and as nativesdk- but not for the target
Jan 25 10:58:50 <nayfe> ttllkk: BBCLASSEXTEND = "native nativesdk" ?
Jan 25 10:59:26 <ttllkk> and what would be the recipe name? just <recipe>.bb with no nativesdk-prefix or -native suffix?
Jan 25 11:11:48 <nayfe> ttllkk: i understand this this way (never tried): if your recipe is pure native, inherit native and name it recipe-native, if it is pure nativesdk, inherit nativesdk and name it nativesdk-recipe, if it can be both, use BBCLASSEXTEND and name it recipe
Jan 25 11:12:53 <ttllkk> thanks, that was my intention also
Jan 25 11:13:07 <ttllkk> the ''problem'' then is that if you 'bitbake recipe' it tries to compile it for the target and afils
Jan 25 11:13:23 <ttllkk> but I guess the fix for that is.. don't do 'bitbake recipe' xD
Jan 25 11:13:54 <ttllkk> (as opposed to the nativesdk- and -native renaming methods, where if you try to bitbake recipe you get a message telling you that the recipe does not exist)
Jan 25 11:19:49 <LetoThe2nd> ttllkk: if that puts you off, you can always pour the logic into an include file and provide the two renamed recipes as wrappers to call
Jan 25 11:20:21 <ttllkk> Thanks for the suggestion
Jan 25 11:20:41 <ttllkk> I was also trying to find some example of what I am doing in some package in poky.. but seems like no one needs to do what I am trying
Jan 25 11:20:48 <ttllkk> so maybe what I am tryign is just wrong
Jan 25 11:27:34 <RP> ttllkk: you can make it raise a SkipRecipe in the non/native/nativesdk case
Jan 25 11:31:55 <yocti> New news from stackoverflow: KBUILD_DEFCONFIG_KMACHINE ?= defconfig_file does not work as expected <https://stackoverflow.com/questions/48441494/kbuild-defconfig-kmachine-defconfig-file-does-not-work-as-expected>
Jan 25 11:33:04 <RP> ttllkk: http://git.yoctoproject.org/cgit.cgi/poky/commit/?h=master-next&id=6630d3a6caa66e8d92d2558c23cf35ad082d906d - realised we never did make the rename :/
Jan 25 11:50:42 <RP> kanavin: https://autobuilder.yocto.io/builders/nightly-qa-extras/builds/729/steps/BuildImages_2/logs/stdio :/
Jan 25 11:50:56 <RP> kanavin: that was with mut last night which should have all your patches
Jan 25 12:05:02 <kanavin> RP: thanks, I'll look
Jan 25 12:09:29 <hundeboll> I just noticed that our all-sources image created using `bitbake -c fetchall rootfs-image` doesn't contain the sources for debianutil(-native), which are needed by do_package_write_ipk when creating the rootfs
Jan 25 12:10:01 <hundeboll> is there any way to make do_fetchall include such dependencies ?
Jan 25 12:30:38 <ttllkk> RP: thanks for the reference
Jan 25 12:40:39 <RP> hundeboll: which version of the project is that?
Jan 25 12:45:40 <hundeboll> RP: pyro at the moment
Jan 25 12:45:47 <hundeboll> let me try with rocko
Jan 25 12:48:03 <RP> hundeboll: ah, what you said isn't quite right
Jan 25 12:48:26 <RP> hundeboll: debianutils comes from ca-certificates and PACKAGE_WRITE_DEPS
Jan 25 12:48:33 <hundeboll> yes
Jan 25 12:49:06 <hundeboll> running `bitbake -c fetchall ca-certificates` doesn't fetch debianutils, which I would expect
Jan 25 12:49:20 <RP> hundeboll: right. Hmm
Jan 25 12:49:33 <hundeboll> (same on rocko, btw)
Jan 25 12:49:42 <RP> hundeboll: I just checked master too
Jan 25 12:50:33 <hundeboll> I am quite clueless about where to inject the right dependency to make it fetch PACKAGE_WRITE_DEPS...
Jan 25 12:50:58 <hundeboll> maybe I should just create a bug for now...
Jan 25 12:52:58 <RP> hundeboll: utility-tasks.bbclass:do_fetchall[recideptask] = "do_${BB_DEFAULT_TASK}" is the code which should do it
Jan 25 12:54:09 <hundeboll> RP: yeah, did stare at that for a short while
Jan 25 12:54:50 <RP> hundeboll: I suspect this is a bug in bitbake itself :(
Jan 25 12:54:58 <hundeboll> I have never completely understood the rec{,r,i}deptask magic, so I mostly work with those on a trial'n'error basis
Jan 25 12:55:21 <RP> hundeboll: trying to make them do the right things that everyone wants/expects under all circumstances is hard
Jan 25 12:56:15 <RP> hundeboll: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=ross/mut&id=de1841d9c2a0ec1d7103335d450b5fbc2273804e - sounds familiar :/
Jan 25 12:56:38 <RP> hundeboll: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=ross/mut&id=5af68d42ba62a2a18d1b0cdacad8e32c2b119255
Jan 25 12:57:49 <hundeboll> RP: i see
Jan 25 12:58:18 <RP> hundeboll: paging this in from 4 years ago is hard :)
Jan 25 12:58:26 <hundeboll> RP: maybe adding do_${BB_DEFAULT_TASK} ro recidepstak for do_package_write_{ipk,rpm,...} then would fix this ?
Jan 25 12:58:31 <hundeboll> hehe
Jan 25 12:59:29 <RP> hundeboll: it may fix it but its a workaround, not a fix
Jan 25 12:59:54 <hundeboll> RP: doesn't even fix it here...
Jan 25 13:07:37 <RP> hundeboll: I'm suspecting there is a problem with recursive use of recrdepends
Jan 25 13:08:14 <guzu> hello
Jan 25 13:08:30 <guzu> what is the significanse of a file.lock in the downloads cache directory?
Jan 25 13:20:00 <andrey_> hi! Did somebody faces with this error "undefined reference to `__multi3'" during kernel compilation?
Jan 25 13:24:40 <tsarrazin> hi community,  did anyone ever integrate the log4cpp on krogoth library? because I'm having some troubles
Jan 25 13:42:04 <hundeboll> RP: I won't go deeper into the dependency resolving in runqueue.py, so let me know if should create a bug for this
Jan 25 13:42:12 <hundeboll> thanks for taking your time with this!
Jan 25 13:43:03 <RP> hundeboll: yes, please create a bug. I've been playing with the code and I think I can see the problem, just not sure of a good way to fix it
Jan 25 13:43:16 <hundeboll> RP: OK
Jan 25 13:48:54 <kanavin> RP: seebs: so if this can be reproduced somehow without building all of yocto, seebs would appreciate :)
Jan 25 13:49:35 <kanavin> I can also try and enable extra debugging to get down to it, but remember I'm not really understanding pseudo :)
Jan 25 13:50:08 <kanavin> seebs: also, if you can take a quick look at custom patches that still remain in the new version, and if they're correct?
Jan 25 13:52:34 <seebs> I can try. I thought I had most of them, but I may have missed some.
Jan 25 13:56:13 <hundeboll> RP: https://bugzilla.yoctoproject.org/show_bug.cgi?id=12510
Jan 25 13:56:14 <yocti> Bug 12510: normal, Undecided, ---, richard.purdie, NEW , fetchall doesn't fetch PACKAGE_WRITE_DEPS
Jan 25 14:07:14 <RP> hundeboll: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip9981&id=a1931e5535d42953e9858e2c0a63fbbd4e7711d1 is a fix which does mean the missing dependency shows up
Jan 25 14:07:21 <RP> hundeboll: it also kills performance even more :/
Jan 25 14:14:05 <jaeckel> does someone know of a best-practice article of how to work with a monorepo in yocto?
Jan 25 14:23:09 <john1987> hello
Jan 25 14:23:38 <john1987> I have a bitbake recipe to create multiple user however in the end only the last user gets generated
Jan 25 14:23:54 <john1987> should i place each user creation in it's own package?
Jan 25 14:28:21 <kanavin> john1987: it helps if you share the recipe
Jan 25 14:30:58 <john1987> i have to find and alternative to pastebin to be able to share
Jan 25 14:31:18 <kanavin> john1987: github :)
Jan 25 14:33:45 <john1987> kanavin: inherit useradd \ USERADD_PACKAGES = "${PN}"  \  USERADD_PARAM_${PN} x 7
Jan 25 14:34:15 <john1987> kanavin: i chcked the resulting .spec file and has only 1 user(the last)
Jan 25 14:34:33 <john1987> kanavin: also i am using an old yocto --- 1.7.3
Jan 25 15:00:58 <ttllkk> Anyone knows where is the code in poky that copies the kernel modules (,ko) int he image?
Jan 25 15:01:04 <ttllkk> I was wondering if there is a way to blacklist some..
Jan 25 15:14:17 <eduardas_m> jaeckel: I actually try to use git submodules whenever possible so that I can avoid checking out unnecessary source for a specific component... bitbake recipes can be written to do a recursive checkout of git submodules
Jan 25 15:16:08 <eduardas_m> jaeckel: are you in a situation where there is no way that you can split up your monorepo into multiple git submodule components?
Jan 25 15:22:13 <nayfe> ttllkk: maybe add this class https://git.phytec.de/meta-phytec/tree/classes/kernel-module-split-blacklist.bbclass?h=rocko ?
Jan 25 15:23:12 <ttllkk> that seems like exactly what I wanted
Jan 25 15:28:16 <ant_work> zeddii, welcome back in the ncurses/menuconfig horror park
Jan 25 15:29:24 <zeddii>  ;) :D
Jan 25 15:38:35 <rburton> ttllkk: don't install kernel-modules and then you can pick the ones you want to install yourself.  kernel-modules just installs *all* of them
Jan 25 15:39:10 <ttllkk> That would work for a whitelist but I'd like a blacklist instead
Jan 25 16:35:32 <noway96> yocto running out of disk space: what is the reason?
Jan 25 16:35:45 <noway96> I have a 100GB partition
Jan 25 16:35:57 <noway96> all for the build
Jan 25 16:36:18 <kanavin> noway96: how much do you have on /?
Jan 25 16:36:35 <kanavin> and how does the error look like?
Jan 25 16:39:06 <noway96> something like memory called "ABORT"
Jan 25 16:47:53 <eduardas_m> noway96: after some experience with Yocto builds, I now use a 3 TB HDD, with four partitions of which  each has more than 400 GB of space... running out of disk space is not at all surprising
Jan 25 16:48:34 <kanavin> noway96: my best guess is that you need to clear out old stuff from sstate cache
Jan 25 16:49:40 <noway96> yup I deleted the tmp folder
Jan 25 16:50:17 <noway96> but also what does sstate cache contain? Won't it slow down the build?
Jan 25 16:50:35 <kanavin> noway96: exactly opposite
Jan 25 16:50:49 <kanavin> noway96: it contains build artefacts, and reuses them when possible
Jan 25 16:51:20 <noway96> but isn't the reusing part where the speedup will occur?
Jan 25 16:51:44 <kanavin> ah, "it" == "delete sstate cache"
Jan 25 16:51:56 <kanavin> not "it" == "sstate cache"
Jan 25 16:52:52 <kanavin> noway96: you can delete only really old stuff, like more than several weeks old - play with 'find <sstate-dir> -mtime (I think)"
Jan 25 16:53:39 <kanavin> but first check if it's really grown too big, or the issue is elsewhere
Jan 25 16:53:55 <noway96> kanavin thank you
Jan 25 16:54:03 <kanavin> noway96: like a build directory that you've forgotten about, if you use several
Jan 25 16:54:37 <kanavin> noway96: also 'du --max-depth=1" is really useful to determine where the available space disappeared
Jan 25 17:04:02 <noway96> du --max-depth=1 for the win!
Jan 25 18:50:48 <JPEWhacker> I was curious if anyone knows why mpg123 still has the "commercial" LICENSE_FLAGS? It seems like recent versions move to LGL might not have the old commercially restricted GPL license?
Jan 25 19:11:42 <khem> kergoth: I am not seen fails with argp-standalone
Jan 25 19:11:53 <khem> kergoth: but then I am severly backed up
Jan 25 19:50:45 <austenhc> Hey does anyone have experience stripping down systemd for an embedded system? I don't need or want udev, networking or any of the other junk it ships with. I also am not sure how to override a bbclass
Jan 25 19:56:05 <BubuIIC> austenhc: I don't think you can have systemd without udev
Jan 25 19:56:19 <BubuIIC> you can disbale the networking parts though
Jan 25 19:57:24 <austenhc> Hm. Maybe it doesn't fit my needs then. UDEV takes much longer than devtmpfs.
Jan 25 19:58:53 <BubuIIC> I think the two are complementory?
Jan 25 19:59:20 <austenhc> udev seems to handle things like hotplugging. On an embedded system with no external ports, that seem unecessary.
Jan 25 19:59:23 <BubuIIC> I.e. you still get most things through devtmpfs nowadays even if you have udev
Jan 25 19:59:45 <austenhc> Yeah, devtmpfs enumerates all my hardware and populates /dev really quickly.
Jan 25 20:00:20 <BubuIIC> udev is doing less and less I think because most things can be done directly in the kernel now
Jan 25 20:00:34 <austenhc> I'm working on a sub-5 second to app image. The pain points are the Init system, UBI (attach takes 2 full seconds!) and bootstrap.
Jan 25 20:00:50 <austenhc> I guess I'm stuck with SysVInit for now.
Jan 25 20:01:22 <BubuIIC> have you done some benchmarking systemd boot vs sysvinit?
Jan 25 20:02:02 <rburton> austenhc: if you want to boot reallly fast and you have a single app, then use something like busybox init which is basically /init -> sh -> your app
Jan 25 20:02:07 <BubuIIC> depending on what you are starting systemd might be faster. (or slower ...)
Jan 25 20:02:08 <austenhc> Yup. SysVInit takes about 6 seconds (mostly bluetooth daemon) Systemd takes about 20s
Jan 25 20:02:32 <austenhc> We're using busybox now! I'll check into it's init system
Jan 25 20:02:51 <rburton> systemd can be incredibly fast, presumably your setup is just doing lots
Jan 25 20:03:05 <rburton> you can probably just not install udev to be honest
Jan 25 20:03:28 <austenhc> Yeah, I couldn't find any good way of just building the core of systemd
Jan 25 20:07:12 <rburton> austenhc: who cares what you build, its what you install
Jan 25 20:08:02 <austenhc> I mean, with Yocto building an installing are basically the same process.
Jan 25 20:08:26 <austenhc> I couldn't find a way to tell systemd not to use udev or ntp services
Jan 25 20:30:42 <noway96> why? unparsed line: 'PV = gitr1'
Jan 25 20:31:13 <neverpanic> noway96: did you mean: PV = "gitr1"
Jan 25 20:32:32 <noway96> neverpanic: as a matter of fact, I did. Thanks!
Jan 25 21:59:39 <joshuagl> armpit: sorry, had to cancel your build. The old cluster has some WIP code on and the build was failing
Jan 25 23:04:24 <yocti> New news from stackoverflow: Yocto do_package_qa hangs for bin_package nodejs recipe <https://stackoverflow.com/questions/46412795/yocto-do-package-qa-hangs-for-bin-package-nodejs-recipe>
**** ENDING LOGGING AT Fri Jan 26 03:00:02 2018