**** BEGIN LOGGING AT Wed Aug 28 02:59:57 2019 Aug 28 06:56:46 LetoThe2nd / qschulz: I've just tried to run with "stock" poky and get exactly the same error about the *.bb files not being found. I've uploaded the entire repo to https://github.com/proffalken/yocto-testing and would really appreciate it you or someone else could take a look and see where I'm going wrong given that this isn't specific to the original project I was working on? Aug 28 07:16:24 Proffalken: the repo doesn't seem to contain the layer in question? **** BEGIN LOGGING AT Wed Aug 28 07:17:39 2019 Aug 28 07:20:42 LetoThe2nd: one of us misinterpreted :) Let's see what Proffalken was trying to say Aug 28 07:22:18 :) Aug 28 07:55:38 Hmmm, hang on, let me find out what's going on here... Aug 28 07:55:47 The layer should be in the "top level" directory Aug 28 07:57:46 hahahah, I pushed the wrong branch... try https://github.com/proffalken/yocto-testing/tree/my-yocto-2.7.1 Aug 28 07:58:12 (as you can see, I've just cut&pasted the commands from the quickstart to get the branch name etc) Aug 28 08:05:29 ok, now it makes sense :) Aug 28 08:55:58 Proffalken__: which machine are you building for? Aug 28 09:01:41 Proffalken__: cd /tmp; git clone https://github.com/proffalken/yocto-testing -b my-yocto-2.7.1; cd yocto-testing; source oe-init-build-env; [add /tmp/yocto-testing/meta-telegraf to BBLAYERS in conf/bblayers]; echo 'DISTRO_FEATURES+="systemd"' >> conf/local.conf; MACHINE=beaglebone-yocto bitbake telegraf; Aug 28 09:01:46 no warning Aug 28 09:01:47 building Aug 28 09:03:13 I got the warning only when building qemux86 (default machine) because your recipe is buildable for arm only (and you explicited it). Then building beaglebone-yocto which is arm-based would have the same warning because telegraf can be built only when DISTRO_FEATURES has systemd in it Aug 28 09:03:41 hence, you've got a layer without recipes in it because all the recipes (telegraf only) can't be built for the specific distro or machine Aug 28 09:03:53 qschulz: interesting, i only set qemuarm, and still got the warning. didn't make the mental transition to enabling systemd too :-( Aug 28 09:03:56 since this is resolved durnig parsing, you're including a layer which is empty Aug 28 09:04:45 LetoThe2nd: I saw it only because I tried to build telegraf only and it failed for the two aforementioned reasons :) Aug 28 09:04:57 qschulz: kudos, good find! Aug 28 09:05:56 LetoThe2nd: \o/ Aug 28 09:07:18 https://www.irccloud.com/pastebin/D3A4jVKf Aug 28 09:07:35 qschultz: Ah, cool, ok, so I need to add a systemd layer as well? Aug 28 09:07:57 I'm building for armhf (pi3/pi0) if that makes a difference? Aug 28 09:09:07 Proffalken__: well that makes the recipe OK from the compatible_host/machine point of view. But you still need systemd in your DISTRO_FEATURES Aug 28 09:09:38 because of https://github.com/proffalken/yocto-testing/blob/my-yocto-2.7.1/meta-telegraf/recipes-telegraf/telegraf/telegraf_0.1.bb#L22 , I assume Aug 28 09:10:03 so 1. remove this if it's not a hard requirements Aug 28 09:11:18 or 2. add DISTRO_FEATURES += "systemd" in the correct place (and I never know where it should be, maybe conf/local.conf but I don't like putting things there, or conf/machine/machineyoureusing.conf, or recipes-*/*/imageyourebuilding.bb or conf/distro/distroyourebuilding.conf) Aug 28 09:11:40 Cool, thanks, I'll give that a go... Aug 28 09:12:00 Proffalken__: OR, have more than one recipe that does not depend on the compatible or distro features :) Aug 28 09:12:20 but anyway, to be able to build telegraf you'll need a correct machine and a correct DISTRO_FEATURES Aug 28 09:13:02 so having a second recipe "compatible-free and systemd-free" is just going to remove the warning but not going to make telegraf buildable Aug 28 09:13:09 New news from stackoverflow: YOCTO : Where can I find my linux image kernel headers Needed to build drivers and how can I install them? Aug 28 09:29:54 I'm trying to create an extended sdk for dev team but i'm stuck here : checking target system type... Invalid configuration `x86_64-custdistro-sdk-linux': system `sdk-linux' not recognized. I can't find where the trouble is inside the distro conf ... any idea ? Full log: https://pastebin.com/Yuvm6KWc Aug 28 09:43:15 New news from stackoverflow: How to structure Yocto conf files for two different machines and same bblayers Aug 28 10:30:09 Hi, i have an issue whit my buid system/sdk ( i cant get geoclue to build ). arm-oe-linux-gnueabi-gcc gives the error unrecognized command line option '-fstack-clash-protection' Aug 28 10:31:01 and i can finde somthing in geoclue that is truning on the option Aug 28 10:43:26 New news from stackoverflow: Rotate framebuffer works only with 180° and 0° Aug 28 11:13:29 New news from stackoverflow: Why does the yocto bblayers.conf file use absolute paths? Aug 28 11:31:36 Hi, just noticed that my patch (posted on 16th of august) havn't received any comments or got merged. How far does the patch backlog stretch and when can I expect that my patch get some attention? :) Aug 28 11:43:35 orzen, wiki page says its perfectly fine to send a reminder ping to the list, but the project is in feature freeze now I believe, so perhaps patch review/merge takes longer Aug 28 11:58:31 kroon: I see, thank you! Aug 28 12:27:44 RP: sorry for not sending that series. *both* my builders had full disks, so I had to deal with that before starting my round of builds. Aug 28 12:40:45 zeddii: np, I'm having fun with plenty of other patches Aug 28 12:44:02 I'm 3/4 of the way through the builds, so it will be out today. Aug 28 12:45:25 Hi all Aug 28 12:46:57 I'm trying to create a recipe from a git repository with devtool add ssh://git@stash.xxx.xx.com:7999/yyyy/zzzz.git Aug 28 12:48:40 but I get an error Failed to fetch URL xxxx with ERROR: Command 'script -e -q -c "recipetool --color=always create --devtool -o /tmp/devtoolertuu76w 'git://git@stash.xxx.xx.com:7999/yyyy/zzzz.git' -x /home/guillaume/data/Fusion/stuff/oe-core-SCR2/build/workspace/sources/devtoolsrc5vqso2w4" /dev/null' failed Aug 28 12:51:12 Having a look to the recipetool commands, it seems the final SRC_URI generated is something like SRC_URI = "git://git@stash.xxx.xx.com/7999/yyyy/zzzz.git;protocol=ssh". It seems there is an issue with the custom ssh port 7999. Aug 28 12:52:15 Does someone know the right way to pass URL to devtool / recipetool for git repository accessed through SSH on port 7999? I have tried to add ";port=7999" but it don't fix the issue Aug 28 13:13:24 RP: Do you know where the checkvnc comes from? I am trying to duplicate the failure that was detected to see whats up. It will be interesting to understand the cause/effect, because I would not have expected any kind of drastic change with the fast fail tty for systemd. Aug 28 13:13:49 New news from stackoverflow: How to build a working TPM2 image for Raspberry Pi with Yocto? || Change YOCTO Linux kernel version Aug 28 13:13:57 The log you pointed me at showed: Running '. ./oe-init-build-env; /home/pokybuild/yocto-worker/qa-extras2/yocto-autobuilder-helper/scripts/checkvnc; DISPLAY=:1 bitbake core-image-sato:do_testimage -k' with output to /home/pokybuild/yocto-worker/qa-extras2/build/build/command.log.6c Aug 28 13:14:04 That is the checkvnc I was asking about. Aug 28 13:20:01 jwessel: you mean the contents of that script? Aug 28 13:20:26 jwessel: http://git.yoctoproject.org/cgit.cgi/yocto-autobuilder-helper/tree/scripts/checkvnc Aug 28 13:20:29 yes. It seems to be related to the autobuilder. I am not sure if that is part of the failure, part of the test or what. Aug 28 13:20:45 jwessel: its ensuring there is a DISPLAY available Aug 28 13:23:47 So in theory I just have to run: bitbake core-image-sato:do_testimage Aug 28 13:30:43 jwessel: yes Aug 28 13:30:54 jwessel: that is just autobuilder boilerplate around the command Aug 28 14:22:16 I find the issue, it has already been fixed : https://github.com/openembedded/openembedded-core/commit/4290e04b69360b5e1da9f37166015e30f66cb335 Aug 28 14:44:03 New news from stackoverflow: Yocto recipe python whl package Aug 28 15:32:44 <__angelo> is there a suggested linxu distro to work with yocto ? Aug 28 15:33:01 <__angelo> *linux :) Aug 28 15:33:16 no, but there are basic functional requirements.. Aug 28 15:33:16 __angelo: anything listed in the manual as being supported? Aug 28 15:33:35 fray: "it boots", "it can run vi" Aug 28 15:33:47 Ubuntu seems to be the easiest to get a modern system to build easily, but RHEL/CentOS 7 works fine.. I've not used RHEL 8 yet.. Aug 28 15:34:03 but RHEL/CentOS 7 may require some additional (newer) software to be installed.. Aug 28 15:34:13 <__angelo> LetoThe2nd, the fact is that i fight to work with an updated linux. Ok will check. Aug 28 15:34:49 my vote would be in ubuntu too. if in doubt, yank up a docker container running it on whatever your preferred host system is (that my technique) Aug 28 15:34:59 if you are on a specific release, say 'Thus' (YP 2.8).. then you should create a VM or Container with a Linux OS from that era.. say Ubuntu 18.04 LTS.. and use that for development.. because newer OSes will change interfaces that the system may depend on. Aug 28 15:35:15 but for 'master', it should work on the latest versions of the common OSes Aug 28 15:36:16 <__angelo> fray, LetoThe2nd ok thanks Aug 28 15:43:24 nrossi: let me know if/as/when you're around Aug 28 16:52:01 * zeddii waits for the last builds to complete .... Aug 28 17:03:45 * kanavin_ is running a world build in a python2-less environment Aug 28 17:04:05 so far, ca-certificates has failed, and, surprisingly, pseudo Aug 28 17:06:17 kanavin_: I did a bit of tinkering with that yesterday as well, I'm wondering did you tweak things so python -> python3, or did you remove it altogether? Aug 28 17:07:08 smurray, I removed it on the host distro, then dropped it from HOSTTOOLS, and also dropped the python-native and python recipes for good measure Aug 28 17:07:26 python -> python3 is a really wrong thing to do Aug 28 17:08:20 kanavin_: slight issue is some distros have done it, so there are some checks for it Aug 28 17:08:53 smurray, I have removed that bit as well Aug 28 17:09:37 kanavin_: you went further than I did, I left the python recipes in place so python-native could work, as I thought that was RP was implying on the call yesterday Aug 28 17:10:08 smurray, there are basically 3 recipes that need them, ltp, perf and kernel-devsrc Aug 28 17:10:17 I fixed ltp, and removed perf and kernel-devsrc :) Aug 28 17:10:28 (and tweaked the packagegroups to not include them) Aug 28 17:10:41 I trust Bruce will get to fixing the last two Aug 28 17:10:57 okay llvm-native is the latest to go bust Aug 28 17:11:02 kanavin_: I think my having python -> python3 avoided some issues, ca-certificates and pseudo built here Aug 28 17:11:38 smurray, good to know they're compatible with py3 then, it's probably a matter of patching them to use python3 explicitly Aug 28 17:11:41 kanavin_: systemtap blew up for me, but I suspect its python3 detection logic is wonky and may work with just python3 available Aug 28 17:14:38 kanavin_: I did look at kernel-devsrc a bit, afaict all the scripts in the kernel "scripts" directory are python3 compatible, though some of them use /usr/bin/python :/ Aug 28 17:16:23 smurray, right, I trust Bruce will find the right way to handle this, as he knows better what scripts are important Aug 28 17:16:31 kanavin_: yeah Aug 28 17:20:56 kanavin_, smurray: this all sounds like good data. I suspect pseudo's python usage is really simple and easily tweaked to python3 explicitly Aug 28 17:21:07 zeddii: cool Aug 28 17:22:07 kanavin_: I think the only other thing I noticed that would need cleanup is the gpgme recipe, the python2 PACKAGECONFIG would need to be removed Aug 28 17:23:05 smurray, yes, but it is not enabled by default, so not a priority really :) Aug 28 17:23:13 kanavin_: and I just noticed with grep that there's a PACKAGECONFIG in alsa-tools that references "python-core python-pygobject pyalsa" Aug 28 17:23:20 smurray, same there Aug 28 17:23:28 easy enough to fix Aug 28 17:23:31 RP: I fixed ca-certificates and pseudo, and resumed the build Aug 28 17:23:56 RP, it's at gcc-cross now, so the bulk of recipes are yet to be tests Aug 28 17:24:02 *tested Aug 28 17:24:10 kanavin_, yah. I have the kernel ones done. they'll be in my series in a few hours Aug 28 17:30:17 heh, I need a second build machine, I switched off of python tinkering to churn some AGL builds Aug 28 19:59:27 smurray: As long as they're not on my default they're ok. We have precedent for things which add dependencies external to oe-core as long as they're not default Aug 28 20:03:44 RP: is the thinking that the python 2 recipes will shift to meta-python? Aug 28 20:12:46 smurray: probably a dedicated meta-python2 Aug 28 20:13:02 * zeddii nods. Aug 28 20:13:15 RP: you’ve seen my series, I think the kernel can be out from under the python2 thumb now :D Aug 28 20:14:34 * zeddii is glad to get that fixed up. Aug 28 20:54:41 zeddii, thanks Aug 28 20:58:20 zeddii: yes, sounds good thanks :) Aug 28 21:15:47 RP: it occurs to me that u-boot is unresolved wrt python. I asked Tartarus, aiui the python3 support for libfdt is tied to other changes that currently block a simple upgrade of libfdt in the u-boot tree Aug 28 21:16:55 And we're in turn grumbling at libfdt folks Aug 28 22:06:41 Tartarus: :/ Aug 28 22:06:57 smurray: on the call the other day it seemed uboot would be the final hurdle Aug 28 22:07:37 but that doesn't inject py2 into the image? Aug 28 22:07:46 RP: yep Aug 28 22:07:56 Crofton|work: correct Aug 28 22:08:13 zeddii: can I just remove that foo patch? Aug 28 22:08:14 sometimes random bikeshedding on patch series tires me out. Aug 28 22:08:21 RP. yup. Aug 28 22:08:28 I just git rm'd it and rebased the patch. Aug 28 22:09:10 * zeddii goes into only reply to useful comments mode. Aug 28 22:09:32 zeddii: thanks, I'll just delete it than rejuggle patches :) Aug 28 22:09:47 yah. my v2 was more overhead than it was worth :D Aug 28 22:10:03 that'll teach me to "git add ." in myeta/recipes-kernel/ :D Aug 28 22:10:18 zeddii: just trying to think of the easiest way to fix it in my tree :) Aug 28 22:10:53 I hear you. I'm not a fan of a single v2 in the middle of a series, but re-issuing it, seemed worse in this case. Aug 28 22:11:10 zeddii: right, that is fine. I just had to check that was the only difference Aug 28 23:05:40 zeddii: first error is in: https://autobuilder.yoctoproject.org/typhoon/#/builders/76/builds/975 :/ Aug 28 23:07:32 zeddii: looks like some scripts still reference "python" :/ Aug 28 23:15:27 New news from stackoverflow: How to install files in the native sysroot with yocto when doing populate_sdk? Aug 28 23:45:32 New news from stackoverflow: Cross compiling a kernel module issue in Makefile Aug 29 00:11:35 RP: bugger! will send an updated patch. **** ENDING LOGGING AT Thu Aug 29 02:59:57 2019