**** BEGIN LOGGING AT Mon Nov 27 03:00:01 2017
Nov 27 09:39:33 <yocti> New news from stackoverflow: How can I use new toaster version in old yocto <https://stackoverflow.com/questions/47507753/how-can-i-use-new-toaster-version-in-old-yocto>
Nov 27 10:06:04 <jdmssmkr> How do I solve the error "multiple providers are available for runtime python3-setuptools (python3-setuptools, python3-distribute)"? I've added "PREFERRED_PROVIDER_python3-setuptools ?= "python3-setuptools"" to my image.bb but I keep getting that error...
Nov 27 10:14:14 <open-nandra> hi guys, anybody here with experience with networkmanager? I've used recipe from meta-openembedded it builds fine but not working with wifi
Nov 27 10:14:26 <open-nandra> cannotbring interface up
Nov 27 10:14:35 <open-nandra> so I was curious if somebody hit same issue
Nov 27 10:20:56 <LetoThe2nd> open-nandra: the most obvious things first: have you enabled its wifi supprt? http://cgit.openembedded.org/meta-openembedded/tree/meta-networking/recipes-connectivity/networkmanager/networkmanager_1.4.4.bb?h=master#n83
Nov 27 10:41:37 <open-nandra> LeoThe2nd: yes I've enabled it
Nov 27 10:41:49 <open-nandra> I have wpa_supplicant installed
Nov 27 10:42:59 <open-nandra> LeoThe2nd: I tracked it down that nm is trying to bring up interface and also wap_supplicant which then fails
Nov 27 10:43:59 <open-nandra> LeoThe2nd: and then basically it's dead (I just want to be sure that I'm not doing anything wrong)
Nov 27 10:45:20 <LetoThe2nd> open-nandra: and the interface works if you manually bring it up?
Nov 27 10:45:40 <LetoThe2nd> open-nandra: hint: if you use tabcompletion, then you'll also get my nick right and highlighting works :-)
Nov 27 10:45:46 <open-nandra> LetoThe2nd: if I stop nm and do ifconfing eth0 up then it works fine
Nov 27 10:46:29 <LetoThe2nd> open-nandra: um, eth0 is not wifi, i assume
Nov 27 10:46:42 <open-nandra> sorry wlan0
Nov 27 10:47:24 <open-nandra> I have this log in syslog:
Nov 27 10:47:26 <open-nandra> Nov 25 22:40:58 rpi user.warn kernel: [    9.803109]
Nov 27 10:47:26 <open-nandra> esp_op_add_interface interface 2 requested by - wpa_supplicant - 269
Nov 27 10:47:26 <open-nandra> Nov 25 22:40:58 rpi daemon.info wpa_supplicant[269]: Could not set
Nov 27 10:47:26 <open-nandra> interface wlan0 flags (UP): Operation not supported
Nov 27 10:47:26 <open-nandra> Nov 25 22:40:58 rpi daemon.info wpa_supplicant[269]: nl80211: Could
Nov 27 10:47:26 <open-nandra> not set interface 'wlan0' UP
Nov 27 10:47:43 <open-nandra> esp_op_add_interface is kernel driver where I've added some printouts
Nov 27 10:47:54 <open-nandra> but same request came before from nm
Nov 27 10:48:04 <open-nandra> esp_op_add_interface interface 2 requested by - NetworkManager - 231
Nov 27 10:48:05 <open-nandra> Nov 25 22:40:58 rpi user.warn kernel: [    9.512234]
Nov 27 10:48:05 <open-nandra> esp_op_add_interface only support MAX 2 interface
Nov 27 10:48:24 <LetoThe2nd> no idea then
Nov 27 10:49:12 <open-nandra> LetoThe2nd: ok thanks anyways
Nov 27 11:59:06 <rburton> armpit: can i kill the processes that are hanging on f26?
Nov 27 13:41:35 <Tamis> If you want to create an image that will contain only a singe recipe inside, no kernel, no pkg manager, no anything else. What is the best approach?
Nov 27 13:43:31 <LetoThe2nd> Tamis: there nothing wrong with having an image recipe that sets IMAGE_INSTALL = "yourpackage"
Nov 27 13:44:28 <LetoThe2nd> that pulls in that one package and its dependencies. the kernel does not get installed anyways unless you explicitly add it. and you can also keep it from getting built by using linux-dummy
Nov 27 13:45:09 <Tamis> LetoThe2nd: You gave me the answer I wanted.
Nov 27 13:45:13 <LetoThe2nd> Tamis: you can find some inspiration here: https://github.com/LetoThe2nd/meta-containerization
Nov 27 13:45:27 <Tamis> LetoThe2nd: My problem was the kernel. That was getting build.
Nov 27 13:45:27 <LetoThe2nd> Tamis: thats basically the infrastructure to do exactly that :)
Nov 27 13:45:51 <LetoThe2nd> Tamis: well then wwhy don't you ask "how do i keep the kernel from getting built?" ;-)
Nov 27 13:46:34 <Tamis> LetoThe2nd: :) sometimes you need to ask some more to get what you want :)
Nov 27 13:47:34 <LetoThe2nd> Tamis: maybe in real life. on IRC, its like this: ask the question that you want an answer for. works out waaaaay better, usually :-P
Nov 27 13:59:02 <nayfe> Hey, after hostnamectl set-hostname XXX, avahi-daemon still respond to previous hostname, do we need to restart avahi-daemon on hostname change by hand ?
Nov 27 13:59:43 <kanavin> nayfe: this should be asked from avahi developers and users
Nov 27 13:59:53 <kanavin> yocto is not an all-purpose channel ;)
Nov 27 14:01:13 <nayfe> kanavin>  indeed but just in case :)
Nov 27 14:40:26 <yocti> New news from stackoverflow: How to determine dependencies of recipes in yocto? <https://stackoverflow.com/questions/34303946/how-to-determine-dependencies-of-recipes-in-yocto>
Nov 27 15:06:15 <zarzar> console output of eclipse mars and oxygen is broken when using the yocto plugin and creating an yocto project
Nov 27 15:42:59 <luneff> can I remove an inherit in a recipe in bbappend?
Nov 27 15:44:15 <luneff> basically, the parent recipe wants a update-rc.d, but I do not
Nov 27 15:47:27 <luneff> oh, found another way around. let it start on unneeded runlevel
Nov 27 15:47:31 <neverpanic> luneff: update-rc.d.bbclass does nothing if a) DISTRO_FEATURES does not contain sysvinit, b) INHIBIT_UPDATERCD_BBCLASS is true, c) INITSCRIPT_PACKAGES is set and empty
Nov 27 15:47:38 <neverpanic> So I suggest you choose one of those.
Nov 27 15:48:02 <luneff> oh, thanks a lot!
Nov 27 15:59:33 <Tamis> LetoThe2nd: I saw your example and I made it work. It works on the images that I don't want the kernel to be built.
Nov 27 15:59:56 <LetoThe2nd> Tamis: :)
Nov 27 16:00:06 <Tamis> LetoThe2nd: But the same happens also on one image that I want the kernel to be built
Nov 27 16:00:39 <Tamis> LetoThe2nd: I tried to override the option on the image. but it doesn't seem to work
Nov 27 16:01:06 <LetoThe2nd> Tamis: i don't get what you are asking.
Nov 27 16:01:38 <LetoThe2nd> Tamis: its a question of the MACHINE. the example i gave you implements a MACHINE with linux-dummy as the selected kernel. hence, it doesn't get built.
Nov 27 16:02:14 <Tamis> LetoThe2nd: ok I see.
Nov 27 16:03:13 <Tamis> LetoThe2nd: What I wanted was to have a layer with some image recipes. Which all of them expect one do not build the kernel.
Nov 27 16:03:58 <rburton> live with it?
Nov 27 16:04:00 <LetoThe2nd> Tamis: if the kernel gets built or not primarily depends on the MACHINE, not on the image
Nov 27 16:05:15 <Tamis> LetoThe2nd:  hmm the machine is the same on both situations.
Nov 27 16:05:20 <rburton> exactly
Nov 27 16:05:39 <rburton> personally i'd live with the kernel being built.  it will be built once and hardly rebuilt again.
Nov 27 16:06:04 <LetoThe2nd> Tamis: if it doesn't get installed, and nothing else happens besides one time build cycles wasted, i agree with rburton.
Nov 27 16:06:20 <Tamis> I think that's what I will go for also
Nov 27 16:06:38 <Tamis> thanks a lot for your input guys
Nov 27 16:16:01 <zarzar> i'm trying to use libxml with a yocto generated toolchain, the includes are not set up properly, libxml include files are not found, how do I fix this? i have tried adding include paths but they do not get used during the build, the includes are not passed to the toolchain command line
Nov 27 16:16:34 <zarzar> continued: i am using yocto plugin for eclipse
Nov 27 16:17:35 <LetoThe2nd> zarzar: here, have a look: OE carries quite some patches for libxml, probably also to take care of thing like that: http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-core/libxml/libxml2_2.9.7.bb
Nov 27 16:18:39 <nayfe> <zarzar> : what do you mean by yocto generated toolchain ? yocto sdk ?
Nov 27 16:18:51 <zarzar> yocto sdk
Nov 27 16:19:08 <nayfe> zarzar: you need DEPENDS = "libxml2" on your recipe
Nov 27 16:19:40 <zarzar> nayfe: libxml is in the image
Nov 27 16:19:51 <nayfe> then if it's not included in sdk, you maybe need to add it with TOOLCHAIN_TARGET_TASK += " libxml2 "
Nov 27 16:20:11 <zarzar> i am trying to use the toolchain installer with yocto plugin, trying with a cmake project now
Nov 27 16:20:48 <zarzar> i am writing applications not trying to build the image just to be clear
Nov 27 16:21:49 <nayfe> zarzar: yocto sdk does not include all stuffs, you may need to add some stuff with TOOLCHAIN_TARGET_TASK variable
Nov 27 16:22:17 <LetoThe2nd> ah yes, ignore what i jsut said. didn't read properly
Nov 27 16:22:24 <zarzar> i do not build the image or toolchain
Nov 27 16:23:29 <nayfe> yocto sdk is not a toolchain, it's an environment
Nov 27 16:23:50 <rburton> zarzar: i'd first check that it does contain the libxml2 headers :)
Nov 27 16:24:07 <zarzar> nayfe: yocto generates a toolchain installer does it not?
Nov 27 16:24:36 <zarzar> rburton: already have
Nov 27 16:25:13 <nayfe> yocto sdk includes everything you need to compile an application, gcc + headers + some packages etc
Nov 27 16:25:18 <zarzar> all the headers are there, i just am not sure how to set the paths for headers and libraries in the yocto cmake project
Nov 27 16:25:27 <rburton> nayfe: sdk contains everything that was in the image you generated it from
Nov 27 16:25:41 <rburton> zarzar: the cmakelists should search for libxml2 as usual, and then it *should* just work
Nov 27 16:26:18 <zarzar> rburton: how do i tell it to search for them?
Nov 27 16:26:37 <rburton> zarzar: dunno, never used cmake
Nov 27 16:26:47 <rburton> google libxml2 cmake i guess :)
Nov 27 16:27:09 <zarzar> has nothing to do with libxml2
Nov 27 16:27:30 <nayfe> do you inherit from pkg-config ?
Nov 27 16:27:48 <zarzar> nayfe: i have no idea
Nov 27 16:28:07 <zarzar> i'm trying to use the yocto plugin with and installed sdk
Nov 27 16:28:31 <rburton> i've never used the eclipse plugin directly but its a lot easier if you use something like cmake/autotools that can find the libraries. which is why i said google libxm2 cmake.
Nov 27 16:28:49 <rburton> otherwise you're writing bare makefiles and that means pain
Nov 27 16:28:57 <zarzar> i created a yocto cmake project
Nov 27 16:29:14 <rburton> then use the right function in cmake to find libxml2
Nov 27 16:29:29 <rburton> https://cmake.org/cmake/help/v3.0/module/FindLibXml2.html <-- first google hit
Nov 27 16:29:32 <zarzar> good thing yocto is free
Nov 27 16:29:46 <rburton> or pay for it and you get paid support
Nov 27 16:32:00 <Crofton|work> "Free"
Nov 27 16:36:29 <mrc3> hey, all. i have a problem with do_package_qa and a virtual/pkg: my recipe rdepends on that virtual/pkg but it ends in error (rocko): ERROR: kmscube-git-r0 do_package_qa: QA Issue: /usr/bin/kmscube contained in package kmscube requires libgbm.so()(64bit), but no providers were found in RDEPENDS_kmscube? [file-rdeps]
Nov 27 16:37:10 <mrc3> i looked at insane.bbclass and certainly, by the time the virtual/pkg is being examined (oe.packagedata.read_subpkgdata()), it comes up empty
Nov 27 16:38:06 <mrc3> would it make sense to retrieve instead the runtime-rprovides for that virtual/pkg?
Nov 27 16:39:08 <rburton> it should have a rdepends on a real package name not virtual/pkg
Nov 27 16:39:17 <rburton> because virtual/pkg is a *build time* name
Nov 27 16:41:29 <zarzar> cmake and autotools were bad choices for yocto plugin, includes show as unresolved, are never resolved in the IDE because they are dependent on sysroot parameter thqat is not known to eclipse IDE
Nov 27 16:41:54 <rburton> they're also used by 99% of all free software out there
Nov 27 16:42:02 <rburton> so not supporting them would be stupid
Nov 27 16:42:45 <rburton> the eclipse plugin does need a bit of love unfortuntately
Nov 27 16:43:23 <zarzar> almost everything in a source file has either yellow warning or red error, but it does build
Nov 27 16:43:42 <zarzar> CDT managed make project is superior
Nov 27 16:44:10 <RP> zarzar: we would love some help improving the eclipse experience
Nov 27 16:45:00 <zarzar> wouldn't be that hard ro generate text file listing of configs needed to allow users to set up a CDT managed project, so they can use eclipse CDT features instead of negating almost all the good features
Nov 27 16:45:20 <mrc3> rburton, all right, i may have gotten wrong my naming here. case in point is kmscube, which requires libgbm.so; the mali driver rprovides libgbm, but rdepending on it is still not enough
Nov 27 16:46:12 <zarzar> RP: i can work on this, how do i get started?
Nov 27 16:46:58 <RP> zarzar: its open source, take what we have improve it, send the changes back (or write a new plugin or whatever makes sense)
Nov 27 16:47:29 <zarzar> ok i don't know where to get the project files
Nov 27 16:48:09 <zarzar> RP: where do i pull the source?
Nov 27 16:48:28 <RP> zarzar: for the existing eclipse plugin?
Nov 27 16:49:46 <zarzar> RP: yes if you don't mind, i'm not sure where to find
Nov 27 16:50:01 <rburton> mrc3: if i build kmscube then i get a RDEPENDS=libgbm and no qa warning, which suggests that its the mali driver's fault
Nov 27 16:50:07 <rburton> (i'm using mesa)
Nov 27 16:50:14 <RP> zarzar: http://git.yoctoproject.org/cgit/cgit.cgi/eclipse-poky/
Nov 27 16:51:08 <zarzar> RP: would it be better to generate the needed info during the build of the sdk installer?
Nov 27 16:51:45 <RP> zarzar: I don't know enough about what is needed to be able to answer that
Nov 27 16:51:53 <mrc3> rburton, right, mali4xx-userland rreplaces, rprovides, and rconflicts libgbm
Nov 27 16:51:55 <zarzar> RP: ok
Nov 27 16:52:07 <zarzar> RP: thanks for the link I will look into it
Nov 27 17:18:20 <mrc3> rburton, after tweaking a few things and making sure that kmscube rdepends on libgbm (but depends on virtual/gbm, now provided by mali), during inspection of libgbm (file-rdeps in insane.bbclass) there are no FILERPROVIDES with libgbm.so
Nov 27 17:18:45 <rburton> mrc3: sounds like a mali bug :)
Nov 27 17:19:02 <mrc3> rburton, i take that for a given
Nov 27 17:19:21 <rburton> mesa builds libgbm with FILERPROVIDES_/usr/lib/libgbm.so.1.0.0_libgbm: libgbm.so.1()(64bit)
Nov 27 17:19:29 <mrc3> i'm just trying to understand what needs to be fixed in mali4xx-userland so that it builds
Nov 27 17:21:13 <mrc3> rburton, yes, because mesa depends on libgbm_10.0.0.bb. mali provides libgbm.so instead
Nov 27 17:24:34 <mrc3> i could only get the right FILERPROVIDES by fetching the runtime-rprovides for said rdependency, in this case rprovided by mali
Nov 27 18:20:02 <zarzar> cmake uses hard coded paths, breaking CVI builds: CMake Error: Cannot open file for write: /home/nth/Desktop/eclipse-ws-3/yocto_cmake_project/Debug/CMakeCache.txt.tmp
Nov 27 18:20:16 <zarzar> yocto plugin should not use cmake projects
Nov 27 18:20:26 <zarzar> Ci builds***
Nov 27 18:20:34 <zarzar> very bad
Nov 27 18:20:42 <khem> are you using SDK
Nov 27 18:23:26 <zarzar> yes with the eclipse yocto plugin, its not looking good, so many bad things abot cmake projects
Nov 27 18:23:40 <zarzar> hard coded paths, so very very very very very very very bad
Nov 27 18:29:50 <khem> are you sourcing the environment script
Nov 27 18:30:19 <khem> usually the environment script should do the needful
Nov 27 18:30:32 <khem> for containing the SDK cross compile env
Nov 27 18:35:39 <zarzar> kehm: "are you sourcing the environment script" ???
Nov 27 18:35:54 <zarzar> no idea what that means, i ran the script if that is what you are asking
Nov 27 18:54:11 <nathani_> so I'm trying to build an eSDK for my image, but it complains that I don't have the qt layer, by I'm BBMASKED the recipes that require the qt layer. eSDK seems to ignore
Nov 27 18:54:37 <kergoth> bbmask prevents recipes from even being parsed. it's impossible for anything to ignore that
Nov 27 18:54:46 <kergoth> those recipes don't exist from bitbake's perspective
Nov 27 18:57:00 <nathani_> pastebin.com/T2vGjaX4
Nov 27 18:58:33 <khem> may be you need to include qtbrowser-qt5 in your BBMASK regexp
Nov 27 18:58:46 <khem> zarzar: can you build this app on cmdline ?
Nov 27 18:58:50 <nathani_> I'm just confused where it's picking up the qtbrowser recipe from, because I have /meta-boundary/recipes-qtbrowser in my BBMASK
Nov 27 18:59:37 <khem> nathani_: perhaps your BBMASK is not what you are expecting it to be
Nov 27 19:00:04 <khem> may be btbake -e can show you what it actually is seeing
Nov 27 19:03:37 <nathani_> -e confirms that it is as expected. it only complains on the populate_sdk_ext. Not on a normal build or on populate_sdk
Nov 27 19:06:22 <zarzar> khem: yes, but CI clones to temp workspace, so the hardcoded paths are all wrong
Nov 27 19:41:29 <yocti> New news from stackoverflow: How can I use new toaster version in old yocto? <https://stackoverflow.com/questions/47507753/how-can-i-use-new-toaster-version-in-old-yocto>
Nov 27 20:11:36 <yocti> New news from stackoverflow: build a library with yocto <https://stackoverflow.com/questions/47518760/build-a-library-with-yocto>
Nov 27 20:12:07 <flashburn> i'm somewhat new to yocto. my understanding is that oe_make runs a Makefile. Where can I find parameters and options that I can pass to it?
Nov 27 20:13:00 <kergoth> oe_runmake just runs 'make'
Nov 27 20:13:07 <kergoth> see the gnu make manpage for details
Nov 27 20:13:16 <kergoth> EXTRA_OEMAKE variable is used to pass arguments along tom ake
Nov 27 20:13:57 <bluelightning> note that what vars you can set on the make command line are of course dependent on the makefile itself
Nov 27 20:17:30 <rburton> flashburn: oe_runmake is just a convenience wrapper around calling make directly, you can see what it does in base.bbclass
Nov 27 20:19:39 <flashburn> rburton: thanks
Nov 27 20:40:49 <flashburn> i'm doing testing. the source code for one library keeps changing. how do i force a reload of the source every time i run the recipe? looked online but couldn't find the answer
Nov 27 20:43:43 <rburton> AUTOREV if its a git/svn/whatever SRC_URI
Nov 27 20:48:18 <zarzar> is the environment file generated as part of sdk installer supposed to run as a bash script?\
Nov 27 20:49:03 <rburton> yes, don't run it under a silly shell like csh
Nov 27 20:49:07 <rburton> its posix sh
Nov 27 20:53:41 <zarzar> dioesn't run in dash i thinnk
Nov 27 20:58:10 <bluelightning> zarzar: are you sourcing it? i.e:   . /path/to/environment....
Nov 27 21:04:21 <zarzar> bluelightning: kehm asked the same, i have no idea what you mean
Nov 27 21:04:40 <zarzar>  ./scripts/environment-setup-cortexa7hf-neon-dey-linux-gnueabi.sh: 5: export: -march: bad variable name
Nov 27 21:05:08 <bluelightning> zarzar: the shell script is designed to be sourced, not run - so that variables it sets are actually set in the environment afterwards
Nov 27 21:05:44 <bluelightning> zarzar: you do that by preceding the shell script with ". " (or "source " with some shells)
Nov 27 21:10:01 <zarzar> source variable not found
Nov 27 21:10:16 <zarzar> with "." same error march bad variable name
Nov 27 21:19:59 <flashburn> rburton: for some reason it doesn't work. i'm still getting the older coder version
Nov 27 21:37:11 <rburton> i do love how we excel at finding race conditions in build systems
Nov 27 21:37:19 <rburton> s/love/hate/
Nov 27 21:37:21 <rburton> so close
Nov 27 21:37:50 <rburton> todays failure is bash erroring out because its trying to compile a .h at the same time that its being written
Nov 27 21:48:06 <halstead> Race conditions: from intriguing to frustrating in 60 seconds.
Nov 27 22:46:31 <rewitt> does devtool work with menuconfig? and if so is there documentation?
Nov 27 22:47:09 <rewitt> The dev-manual still says to run "bitbake -c menuconfig", but I don't want to recommend that if devtool can do it and has docs
Nov 27 22:48:52 <rewitt> bluelightning: ^
Nov 27 22:49:42 <bluelightning> rewitt: work with? yes, it should, but it has no built-in front-end to menuconfig
Nov 27 22:49:53 <bluelightning> (which means you can't use it in the eSDK, only next to the build system)
Nov 27 22:51:27 <rewitt> bluelightning: I think that's fine, I just need the *best current way* to tell them how to change the kernel config. But honestly even menuconfig won't be enough on it's own if they want persistence. There is no, easy button.
Nov 27 22:52:20 <bluelightning> rewitt: the persistence comes when you do the next step i.e. devtool finish - that should take what changes you have made, generate a config fragment and put it next to the recipe
Nov 27 22:52:49 <rewitt> bluelightning: Oh, I was not aware it did that. It diffs versus the default?
Nov 27 22:52:59 <bluelightning> rewitt: yes
Nov 27 22:53:06 <rewitt> bluelightning: Oh that's sweet
Nov 27 22:54:00 <rewitt> bluelightning: Still trying to figure out if they even need it though. They say they just want to turn of some modules. Do you know if you can remove some from "kernel-modules-all" or whatever the meta package is?
Nov 27 22:54:31 <rewitt> bluelightning: I think that would be the idea solution, don't change the config for the kernel, just don't install some of the modules in the image
Nov 27 22:54:52 <bluelightning> rewitt: I think you'd have to generate your own packagegroup / more specific selection of modules
Nov 27 22:55:02 <rewitt> blargh
Nov 27 23:16:56 <rewitt> bluelightning: So then is there documentation for using devtool with menuconfig?
Nov 27 23:55:45 <bluelightning> rewitt: I don't think there is I'm afraid, but apart from the menuconfig step it's pretty much the same as any other recipe
Nov 28 00:16:00 <majuk> Hi all. I was trying to switch to a new kernel [3.14.28->4.4] in Yocto. Got my config setup, but bitbake appears to hang on do_fetch
Nov 28 00:16:08 <majuk> Any questions or comments appreciated.
Nov 28 00:17:20 <majuk> It's possible it is actually downloading the source for the kernel... but I'm not sure how to go about confirm/denying that.
Nov 28 00:17:41 <majuk> Been working on do_fetch for about 90 minutes thus far.
Nov 28 01:03:31 * armpit scratches head... morty.. such a troubled child
Nov 28 02:12:37 <yocti> New news from stackoverflow: How to explain this glibc modification on libpthread? <https://stackoverflow.com/questions/47522791/how-to-explain-this-glibc-modification-on-libpthread>
**** ENDING LOGGING AT Tue Nov 28 03:00:01 2017