**** BEGIN LOGGING AT Thu Oct 03 02:59:57 2019
Oct 03 07:42:33 <pepone> Hey, how can I get Warrior to build 5.3 kernel instead of default 5.0
Oct 03 07:43:59 <pepone> PREFERRED_PROVIDER_virtual/kernel = "linux-yocto-dev"
Oct 03 07:44:02 <pepone> do I need that?
Oct 03 08:34:17 <yocti> New news from stackoverflow: Add modules to php/apache2 in Yocto <https://stackoverflow.com/questions/58215105/add-modules-to-php-apache2-in-yocto>
Oct 03 09:04:21 <yocti> New news from stackoverflow: Awkward line wrap in Yocto <https://stackoverflow.com/questions/57952823/awkward-line-wrap-in-yocto>
Oct 03 10:48:14 <iceaway> I am trying to build an SDK based on a distro/image with multilib support. I get various error messages about conflicting files, where it looks like a conflict between the 32-bit and 64-bit version. Is it possible to generate an sdk that supports both 32-bit and 64-bit?
Oct 03 11:04:37 <yocti> New news from stackoverflow: Generate a file and deploy it to image, inside a yocto recipe <https://stackoverflow.com/questions/58217635/generate-a-file-and-deploy-it-to-image-inside-a-yocto-recipe>
Oct 03 11:09:53 <rburton> iceaway: libraries should be in separate paths and identical files are ignored so i believe it shoudl work. what recipes are conflicting?
Oct 03 11:12:06 <iceaway> https://pastebin.com/xNU7XGP9
Oct 03 11:12:13 <iceaway> This is the error output from bitbake.
Oct 03 14:24:02 <palate> do you guys know "legato"?
Oct 03 14:30:29 <khem> iceaway: which release are you on ? gmp is fixed by this https://git.openembedded.org/openembedded-core/commit/?id=a2b4552b96c49dbdc9aef4b64820c3ff16dc5bb6  icu-config is fixed by https://git.openembedded.org/openembedded-core/commit/?id=52f2dd97e9dd20dea0f3cdeb2df490d1a4c646aa
Oct 03 14:31:16 <khem> I am sure others are also fixed, please try with latest master and if it works then backport the needed bits to release branch you are on. Test them and send backport requests if that release is still maintained
Oct 03 14:32:46 <LetoThe2nd> palate: i have a classic education on two instrucments, so i certainly know "legato". but i guess thats not what you mean.
Oct 03 14:39:23 * palate looks if there is a yocto-based "staccato" system...
Oct 03 14:39:57 <palate> LetoThe2nd: Legato seems to be some sort of alternative to Yocto, but that claims it is compatible with it. Not sure what that means.
Oct 03 14:40:45 <palate> LetoThe2nd: my company is looking into it, and I don't like the idea because Yocto seems to me like there is a bigger community. Community is an important point to me
Oct 03 14:41:10 <LetoThe2nd> palate: seems like jsut another application platform built upon yocto/OE
Oct 03 14:41:26 <rburton> "yocto 1.7.2" !!
Oct 03 14:41:31 <rburton> (very old)
Oct 03 14:41:57 <palate> rburton: oh, I don't know if they support "from yocto 1.7.2 on" or only 1.7.2, that's a good point
Oct 03 14:42:03 <LetoThe2nd> palate: if they offer something that give you a benefit, lookinto it. but the yoctoians will probably care little for the problems and breakages of their metadata.
Oct 03 14:42:11 <rburton> but it looks like its just a yocto layer
Oct 03 14:42:30 <palate> but then I don't see the point... we should just make our own layer, and that's it.
Oct 03 14:43:31 <LetoThe2nd> palate: well they seem to offer some fancy api. if it suits your need, looking into it is good. know as much as you can.
Oct 03 14:43:42 <LetoThe2nd> rburton: fun fact: i am just cleaning my toaster. literally.
Oct 03 14:44:51 <palate> got it
Oct 03 14:46:19 <LetoThe2nd> palate: i mean, i kmow *A* *ALOT* of stuff that i hope i never ever have to touch. but therefore i know the *WHY*
Oct 03 14:47:02 <palate> LetoThe2nd: yes. I think somebody should know the "why", always...
Oct 03 14:50:22 <rburton> palate: it might just be a glorified BSP layer
Oct 03 14:50:41 <rburton> 'meta-swi - Hardware adaptations'
Oct 03 14:51:59 <palate> yes, I'm having a hard time understanding what legato is exactly -_-. If it's just a layer, then it's just yocto with a BSP layer. They may sell it as a a whole framework, but if it's just a layer, the system is still yocto...
Oct 03 14:52:07 <rburton> of course that yocto legato stuff is appears dead
Oct 03 14:52:26 <palate> rburton: like they branched away from yocto?
Oct 03 14:52:39 <rburton> 'Important: This page is now archived.  For Legato Application Framework Releases please visit legato.io.  For Linux Source Code and Toolchains visit your module's firmware page.  '
Oct 03 14:56:39 <palate> 'The Legato AF is designed to run on top of an embedded Linux Distribution. Each vendor maintains it's own customized Linux Distribution that contains a customized kernel and drivers to work within the constraints of the target device.'
Oct 03 14:56:49 <palate> AF being "application framework"
Oct 03 15:00:44 <rburton> | Copying files into the device: nic_AMDA0058.nffw: Directory block checksum does not match directory block while looking up "nic_AMDA0058.nffw"
Oct 03 15:00:44 <rburton> | mkfs.ext4: Directory block checksum does not match directory block while populating file system
Oct 03 15:00:46 <rburton> huh
Oct 03 15:00:59 <rburton> disk full?
Oct 03 15:26:55 <LocutusOfBorg> palate, I know it, unfortunately
Oct 03 15:27:12 <palate> LocutusOfBorg: sounds pretty bad xD
Oct 03 15:27:57 <LocutusOfBorg> palate, I avoided it in my dailyjob, and forced customer to not use it :)
Oct 03 15:28:16 <LocutusOfBorg> but another customer in the meanwhile adopted it so some of my colleagues are using it right now
Oct 03 15:29:05 <palate> LocutusOfBorg: and is it painful?
Oct 03 15:30:41 <palate> LocutusOfBorg: I'm still not sure what it does... does it bring some kind of system to install apps instead of writing recipes?
Oct 03 15:45:41 <khem> RP:whats your take on musl and llvm9 patches
Oct 03 15:46:10 <LocutusOfBorg> palate maybe we can followup in private :)
Oct 03 15:46:51 <khem> musl 1.24 release is around the corner I hope it happens this week but wont count on it we are on tip with latest patch I sent
Oct 03 15:50:36 <RP> khem: how many of them are there?
Oct 03 15:51:03 <RP> khem: you think the musl upgrade should be in 3.0?
Oct 03 15:51:09 <khem> RP:yes
Oct 03 15:51:16 <khem> it should be
Oct 03 15:51:30 <RP> khem: you didn't really give me much justification on why I should break the "no upgrades in M4" rule :/
Oct 03 15:51:38 <khem> ugh
Oct 03 15:52:16 <khem> well reason is that it has fixed for riscv which is related signal.h header and it will affect Apps, we dont want people to use it
Oct 03 15:52:43 <RP> khem: we're supposed to be frozen and bugfix only so it helps to spell things like that out. Was there other changes in there?
Oct 03 15:53:07 <khem> yes few bug fixes but no other major changes
Oct 03 15:54:32 <khem> it also has support for linux 5.2 headers fixes which we should need now that we are on 5.2
Oct 03 15:55:12 <RP> khem: ok, that helps to justify it
Oct 03 15:56:31 <khem> https://git.musl-libc.org/cgit/musl/log/?qt=range&q=1f0e9f9cc2e3fa354f94e18b3b362de5f1ec7272..2c2477da9a553c0b9b2fa18073a5dcdbe6d395af
Oct 03 15:56:43 <khem> this is the complete changeset
Oct 03 15:57:17 <PinkSnake> Hello all, I have some trouble to build a Ros package (ld: cannot find -lros  | ld: cannot find -lroscpp)  and I don't understand what I have to do because the recipe contains already DEPENDS & RDEPENDS information :S any idée ?
Oct 03 16:00:15 <khem> RP:also https://patchwork.openembedded.org/patch/165408/ is safe, it says on patchwork its on master-next but then it really is not
Oct 03 16:01:12 <khem> PinkSnake:you can check if these libs are in the sysroot of the package
Oct 03 16:01:18 <khem> and in standard path
Oct 03 16:05:26 <yocti> New news from stackoverflow: Enable php intl extension in Yocto <https://stackoverflow.com/questions/58222685/enable-php-intl-extension-in-yocto>
Oct 03 16:06:58 <PinkSnake> khem not sur to understand what I have to check because DEPENDS = "libscanstream roscpp catkin std-msgs sensor-msgs"
Oct 03 16:13:44 <qschulz> PinkSnake: WORKDIR/recipe-sysroot is the libros there? then is the sysroot passed to the configure/compile steps so that it can be found?
Oct 03 16:15:14 <PinkSnake> qschulz seems ok : /home/fds/fds/yocto-fsd/sdf/sdf/tmp/work/cortexa9t2hf-neon-aaa-linux-gnueabi/roscpp/1.11.21-r0/sysroot-destdir/opt/ros/indigo/lib/libroscpp.so
Oct 03 16:17:57 <PinkSnake> qschulz it's probably a trouble in the lib path because ros lib are installed in /opt/ros/version... and it's not a usual path
Oct 03 16:20:05 <qschulz> PinkSnake: this is in the recipe of roscpp, not the recipe requiring libros. We need to check if libros is in the sysroot of the recipe which requires it
Oct 03 16:21:04 <qschulz> PinkSnake: but your remark is a good note for later
Oct 03 16:27:34 <PinkSnake> qschulz You are right, i don't have any folder sysroot inside libros :S
Oct 03 16:27:42 <PinkSnake> RDEPENDS should be the trick ?
Oct 03 16:40:54 <qschulz> no, DEPENDS is for buildtime dependencies, RDEPENDS for runtime. So DEPENDS, if the DEPENDS is missing something then it won't find it obviously
Oct 03 16:41:26 <qschulz> you should have the recipe providing libros and libroscpp in the DEPENDS of your Ros package recipe
Oct 03 17:02:15 <khem> if libs are not in standard paths <sysroot>/usr/lib <sysroot>/lib then you have to add those paths to -L
Oct 03 17:05:35 <yocti> New news from stackoverflow: .h file not found when compiling a package using Bitbake <https://stackoverflow.com/questions/58223507/h-file-not-found-when-compiling-a-package-using-bitbake>
Oct 03 17:06:06 <RP> khem: I merged that but tweaked it to patch the inc file
Oct 03 17:06:17 <RP> khem: this way it fixes all images
Oct 03 17:06:25 <khem> RP:cool !
Oct 03 17:06:31 <RP> khem: forgot to fix the commit message :/
Oct 03 17:07:31 <khem> RP: on a different topic I wanted to start sending some QA reports for extended layers from meta-openembedded is it something we can do
Oct 03 17:08:06 <khem> right now, I have made meta openembeddded layers build breaks free, thats a huge milestone
Oct 03 17:08:19 <khem> want to take it to next level
Oct 03 17:13:31 <kergoth> meta-oe fully builds? really?
Oct 03 17:19:38 <Crofton> yes hell froze over
Oct 03 17:42:26 <rburton> khem: how many pnblacklist?
Oct 03 18:20:57 <mischief> is anyone using systemd-initramfs? trying to figure out how this is happening http://ix.io/1Xuw
Oct 03 19:14:37 <khem> rburton: a few
Oct 03 19:15:25 <khem> actually 7
Oct 03 19:18:19 <mranostay> so would it be non trivial to enable the pixbuf PACKAGECONFIG by default in freetype. so that color emoji fonts could render?
Oct 03 19:58:57 <khem> mranostay: sometimes these kind of changes can have unintended effects on indirect deps if you can contain those then it will be not that hard
Oct 03 20:07:34 <rburton> mranostay: you can enable it in your distro, obviously.
Oct 03 20:07:49 <rburton> mranostay: feel free to send a patch to enable it by default if you think thats the right thing to do
Oct 03 20:24:57 <kergoth> yeah, definitely take note of how many new recipes end up built in a typical build after enabling it
Oct 03 20:25:55 <rburton> looks like just libpng
Oct 03 20:47:57 <mischief> is there a standard method for removing a file from only a specific image? i want to delete something out of the initramfs, but not the normal image.
Oct 03 21:08:11 <kergoth> mischief: you probably want a ROOTFS_POSTPROCESS_COMMAND in the image reicpe
Oct 03 21:12:59 <mischief> is that different from IMAGE_PREPROCESS_COMMAND ? i just tried IMAGE_PREPROCESS_COMMAND in my initramfs recipe and it seems to work.
Oct 03 21:14:18 <mischief> though i'm not sure that's correct
Oct 03 21:14:57 <kergoth> if you're modifying the rootfs rather than messing with image construction i'd advise using the rootfs one, though in reality they run right after one another in most cases
Oct 03 21:15:01 <kergoth> just as a matter of principle..
Oct 03 21:18:26 <RP> khem: I don't quite understand what you want to do
Oct 03 21:26:21 <khem> RP: 1. Add remaining meta-openembedded layers to meta-oe task, create a mechanism to run ptests on meta oe packages, report that into test results
Oct 03 21:28:41 <RP> khem: its basically a question of changing the config.json in yocto-autobuilder-helper
Oct 03 21:28:57 <RP> khem: you could even test changes by specifying a different repo to pull the helper  config from
Oct 03 21:29:40 <RP> khem: the unmarked Branch/Repo/Revision boxes at the top of force build are for the helper
Oct 03 21:45:06 <RP> khem: for the ptests you'll probably need to create an image of the things you want to run ptests for
Oct 03 21:56:31 <mischief> kergoth: thanks!
Oct 03 23:15:45 <rburton> kanavin_: can you remember what 0001-mesonbuild-environment.py-check-environment-for-vari.patch was for?
Oct 03 23:54:32 <mischief> does anyone know if ROOTFS_BOOTSTRAP_INSTALL is documented somewhere? or if there is a flag to define packages to only install in the initramfs on a per-machine basis?
Oct 04 00:28:38 <aehs29> Im trying to get travis to build automatically through github integration, but I (obviously) hit the 50 minute time limit (free travis), does anyone know if theres a better solution?
Oct 04 00:29:04 <aehs29> trying to avoid using some s3 workaround to get the sstate
Oct 04 00:29:23 <rburton> run your own worker and persist the sstate
Oct 04 00:29:47 <rburton> s3 can be too slow (and expensive) to persist sstate on
Oct 04 00:30:08 <rburton> you can try using the sstate on yoctoproject.org but obviously you need to pin to the release and use poky etc
Oct 04 00:30:35 <rburton> but the real solution is your own hardware, the runner on it, and a persistant sstate
Oct 04 00:31:44 <rburton> (your own hw could obviously be on aws or whatever, you just need persistant local storage)
Oct 04 00:32:19 <rburton> oh and github actions are much nicer than travis :)
Oct 04 00:33:11 <rburton> ah but actions don't do self-hosted yet so ignore me
Oct 04 00:33:32 <rburton> half one, definitely time for bed
Oct 04 00:33:35 <rburton> 'gnight
Oct 04 00:34:06 <Crofton> good night john boy
Oct 04 00:48:01 <neverpanic> aehs29: other services give you longer build times in the free tier, github actions or azure pipelines, for example.
**** ENDING LOGGING AT Fri Oct 04 03:02:52 2019