**** BEGIN LOGGING AT Tue Jan 22 02:59:56 2019 **** BEGIN LOGGING AT Tue Jan 22 07:44:59 2019 Jan 22 08:09:30 New news from stackoverflow: How to use openjdk-8-native as default java in yocto build? Jan 22 08:19:06 Can someone give me a hint on how to install custom dtbo device tree overlays on the raspi boot partition when using mender? when using "inherit devicetree" in the recipe they appear in /boot/devicetree/, but not in /uboot/overlays/ where the raspi firmware blob needs them at boot time :/ Jan 22 08:35:51 Hi all, someone here has hint to build add a function/methode after the image build process ? Jan 22 08:38:36 What do you want to do in that method? Jan 22 08:39:52 hmm, trying to switch to openssl 1.1.0 in sumo but bitbake claims I didn't set PREFERRED_VERSION_nativesdk-openssl = "1.1.%" correctly but I did according to bitbake -e.. any ideas? Jan 22 08:40:02 PinkSnake: ROOTFS_POSTPROCESS_COMMAND Jan 22 08:40:14 malanecora: build a custom image ( in fact I need just to encrypt the Yocto image :) like a post process but i need to use the generated image as output Jan 22 08:40:41 LetoThe2nd: yes but it's not used just to made modification inside the rootfs ? Jan 22 08:40:50 PinkSnake: hm Jan 22 08:41:11 PinkSnake: And what about using a script? Jan 22 08:41:13 PinkSnake: possibly. well i still think that you should add a custom image tyoe Jan 22 08:41:22 type. Jan 22 08:41:44 (as you did ask about this already some days ago) Jan 22 08:42:02 malanecora: LetoThe2nd Thx for info, i will double check image type :) Jan 22 08:42:46 LetoThe2nd: Yes I have made the same thing some days ago in order to bluid a FIT image and all works like charme :) Jan 22 08:43:00 regarding my question above, looks like IMAGE_BOOT_FILES_append = "devicetree/foo.dtbo;overlays/" does the trick. Jan 22 08:45:29 "Initialising tasks...ERROR: Multiple versions of openssl are due to be built (/home/builder/src/base/poky/meta/recipes-connectivity/openssl/openssl_1.0.2p.bb /home/builder/src/base/poky/meta/recipes-connectivity/openssl/openssl_1.1.0i.bb). Only one version of a given PN should be built in any given build. You likely need to set PREFERRED_VERSION_openssl to select the correct version or don't depend on Jan 22 08:45:35 multiple versions." Jan 22 08:46:41 Does anyone know how to solve the "No manifest generated from: pkg_name" ERROR? It's driving me crazy since the manifest generation is done by the build system... Jan 22 08:46:51 mcfrisk: Which Yocto version are you using? Jan 22 08:49:32 malanecora: sumo Jan 22 08:49:38 mcfrisk: Set the preferred version in your distro configuration Jan 22 08:50:14 I did, but it seems the packaging is changed from 1.0.2 to 1.1.0 since "NOTE: preferred version 1.1.% of openssl not available (for item libssl)" etc Jan 22 08:50:39 mcfrisk: However, if you're trying to build 1.1.x is highly likely that you'll expericence issues related to other packages' dependencies Jan 22 08:52:15 yes, exactly. bitbake showed them as NOTE: when parcing recipes and the actual error message was bogus. But I know what to look for now Jan 22 08:54:12 Hi all, Jan 22 08:54:13 mcfrisk: Btw, I came here asking the same question as you a week ago. The recommendation I get here was to upgrade Yocto to Thud (basis openssl for other recipes is 1.1.x) Jan 22 08:54:24 got* Jan 22 08:57:09 malanecora: ok, that's good to know. might have to do the same but it's a big transition on my project and will takes months.. Jan 22 08:58:25 The main question is always, how to depends of the image name do_image_complete ( depends upon non-existent task do_image_complete...) Jan 22 08:58:34 mcfrisk: I know it can be tedious...but the most straightforward solution from my point of view Jan 22 08:59:40 mcfrisk: You can always try to patch the other recipes as well :P Jan 22 09:11:14 yes, but things like separate binary packages for libcrypto may increase target disk usage with openssl 1.1.0. I thought both version would be kind of supported in sumo.. Jan 22 09:50:48 Hi, I'm trying to use devtool (thud) on a git repo that does not contain master branch and it seems srcbranch option is not working: $> devtool add -B develop -S 902ceaf8a1d53ea1b70cf9f3f0302da87f5a7f8c sol2 https://github.com/ThePhD/sol2 fails with "Unable to find revision 902ceaf8a1d53ea1b70cf9f3f0302da87f5a7f8c in branch master even from upstream" Jan 22 09:59:30 nayfe: maybe set SRCREV on tag value ? Jan 22 10:02:25 PinkSnake: I also tried -V option, tags, without specifiing revision and only srcbranch, all failed :/ Jan 22 10:33:10 Hi all, how do I set up a default MACHINE variable when none is specified in a bitbake command ? Jan 22 10:43:21 prabhakarlad: look into the script setup-environment. There's a line for the usage. Jan 22 10:44:00 The script is called "setup-environment" Jan 22 11:02:57 prabhakarlad, you can do it in your local.conf, or directly in command line: MACHINE= bitbake Jan 22 11:03:13 yoctonator: not everyone has setup-environment Jan 22 11:03:50 prabhakarlad: as kanavin said, the fallback default is in local.conf. Jan 22 11:07:55 rburton: any idea why the axis fonts are corrupt in the perf list emails? Jan 22 11:08:09 RP: no, happens randomly Jan 22 11:08:23 missing fonts on whatever builder send the mails, maybe? Jan 22 11:08:25 rburton: I guess at least we're getting them now. Are they ever legible? Jan 22 11:08:29 yes Jan 22 11:08:44 rburton: just wondering that, missing font would be deterministic though? Jan 22 11:09:05 yeah but i've never paid enough attention to spot if there's a pattern in the builder that sent the mail Jan 22 11:09:14 rburton: fair enough :) Jan 22 11:09:31 might be better to see if we can remove the crazy generate-png-from-chart code with just embedding a svg chart Jan 22 11:10:16 rburton: One step at a time... Jan 22 11:10:29 sure Jan 22 11:10:33 * RP wants build perf data included with release builds for example Jan 22 11:10:45 maybe even the html report since the release ends up on a webserver Jan 22 11:14:20 Does anyone know why I don't have eth0 after booting my result of "bitbake core-image-minimal" on my eval board? Somehow the boot sequence doesn't show something like "attached PHY driver Atheros 8031 ethernet" Jan 22 11:15:08 ifconfig only shows "lo". Jan 22 11:15:22 but no "eth0" Jan 22 11:17:56 missing kernel module? Jan 22 11:18:01 minimal means minimal, there's not a lot in it Jan 22 11:21:40 Does it mean I should try to add it with "make menuconfig"? Jan 22 11:23:35 yoctonator: kanavin: rburton: thank you for the pointers. Jan 22 11:27:36 yoctonator: they get built, but not installed. add to local.conf: IMAGE_INSTALL_append = " kernel-module-asix" (in my case) Jan 22 11:28:06 don't know what module is needed by your hw Jan 22 11:40:11 New news from stackoverflow: Yocto: Create a New Directory in etcdir Jan 22 11:45:07 RP: re fonts in charts, the latest build perf i saw 12 minutes ago from ypperf-ubuntu16 has legible charts Jan 22 11:45:52 rburton: its not an autobuilder one Jan 22 11:46:08 rburton: I need halstead to help fix the AB 1604 machine Jan 22 11:46:10 no, but i mean the charts *can* work in the mails Jan 22 11:46:17 i suspect its missing fonts Jan 22 11:46:18 rburton: oh, yes, agreed. Jan 22 12:01:19 My system lacks of /lib/init/vars.sh (e.g. needed by gpsd) Jan 22 12:02:19 Is this issue related with sysvinit in Thud? Jan 22 12:07:56 Dear Yocto Team, I have practically no experience in yocto.I am trying to modify the snmp sources to add a mib module and I have read that I need to install "devtool". I execute the following commad. $ bitbake fsl-image-gui -c populate_sdk_ext Then I get the following error message: ERROR: Uninative selected but not configured, please set UNINATIVE_URL I'm checking the mailing lists to find a discussion where the same probl Jan 22 12:08:19 reviewed meta-freescale and openembedded core mailing list and now I'm going to review yocto-builds mailing lists. Could you help me solve the error or give me some guidelines to do it? Thanks. Jan 22 12:08:29 you don't need to build a sdk to use devtool Jan 22 12:14:04 I wonder how I get to know what the name of the kernel-module for the Atheros ethernet driver is Jan 22 12:14:34 kernel-module-atheros doesn't work LOL Jan 22 12:15:45 (Error: unable to find a match) Jan 22 12:18:25 Thanks rburton, according to the Yocto Project Application Development and the Extensible Software Development Kit (eSDK) manual I can download a tarball installer, but this is based off the core-image-sato and core-image-minimal images.It would be a problem? Jan 22 12:20:01 Do I need to install eSDK to use devtool? Jan 22 12:30:02 How do I find what the name of my linux kernel recipe is for the call "bitbake -c menuconfig"? Jan 22 12:31:15 Is that a name of a subdir or is it mentioned in a config file somewhere? Jan 22 12:32:08 Today I'm totally stranded with Yocto again :-( Jan 22 12:33:54 I searched the whole internet for how to get that eth0 to work... In a GUI config tool, I just would select it from a combobox... Jan 22 12:34:39 for including it into the distro. So I hope "make menuconfig" gives me a hint. Jan 22 12:40:26 yoctonator: just write "bitbake virtual/kernel -c menuconfig" Jan 22 12:40:58 yoctonator: anyway, you should have the preferred virtual/kernel provider set in yout machine.conf Jan 22 12:45:39 yoctonator: have a look at the files in tmp/deploy/*/kernel-module-* and see if you can spot the name there Jan 22 12:55:12 hey, thanks malanecora! "bitbake virtual/kernel -c menuconfig" opens the GUI and it has the label "linux-imx" which I think is the name of the linux kernel recipe :-) Jan 22 12:56:10 anyway "bitbake linux-imx -c menuconfig" works in the same way then :-) Jan 22 12:59:30 yoctonator: You're right, it is the same recipe indeed! virtual/kernel is just a placeholder to make recipes provider-agnostic Jan 22 13:16:50 I'm trying to move the IMAGE_BOOT_FILES_append into the meta-package/recipe where the dtbo files to be included are generated - but seems like I didn't get the right location/file for this. Jan 22 13:17:06 inside local.conf it works fine, see https://paste.noname-ev.de/9110 Jan 22 13:18:03 any hints on where I should look? Jan 22 13:40:32 zeddii: I merged those kernel patches but I think its broken qemuarm poky-lsb runtime Jan 22 13:48:07 zeddii: reverting 1873c9605dcfb0af6eb57936c8adf3a9e23dcfb0 does fix it Jan 22 14:06:31 The subdir /etc/network is missing in my root filesystem. Do you have any idea what I'm missing in my configuration? Jan 22 14:06:54 Gpsd initscript in Yocto Thud is rotten Jan 22 14:07:57 RP: I had no luck finding something useful in the long list in /tmp/deploy Jan 22 14:13:41 RP: I had mentioned that it might break things before the full switch over was working. It will be chosing the new machine versus the old by default right now. Jan 22 14:14:01 so that must have been it. Jan 22 14:14:20 I could drop the match of the new machine definition on "qemuarm" and that would fix it as well. Jan 22 14:14:55 does anyone know why openssl-bin from 1.1.1a RDEPENDS on perl? Jan 22 14:18:16 yoctonator: init-ifupdown, if you want debian-style network ifup/ifdown Jan 22 14:18:57 yoctonator: if you're starting with core-image-minimal, maybe don't. core-image-full-cmdline is a bit more featured. Jan 22 14:23:10 anyone know of a good emacs major mode for .dtsi files? Jan 22 14:25:11 zeddii: hmm, I'd assumed this wouldn't break since this means we cant pull kernel changes into stable branches safely? :/ Jan 22 14:26:08 zeddii: Choice is to either fix somehow or revert the change for now Jan 22 14:26:27 the change switched the BSP completely. For -stable, I would have tweaked it. Jan 22 14:26:43 I can send a follow up patch that drops the 'qemuarm' from the new definition. Jan 22 14:27:45 zeddii: I don't know the underlying code so I'll need to go off your judgement there, I just see the srcrevs Jan 22 14:28:10 I have noticed that the transition from Rocko to Thud has made my Wi-Fi iface to be recognised really late Jan 22 14:28:39 zeddii: we're only doing this in master, not the stable branches so I worry we may otherwise pull thiis into the other stable branches accidentally though Jan 22 14:28:39 Before that, the iface was risen up at boot (along with eth0) Jan 22 14:29:19 ack'd. I'll put something together now and then make the changes safe for backport (if it happens on purpose, or just through process). Jan 22 14:29:29 single patch headed your way now Jan 22 14:30:00 rburton: yeah, cool! I added "init-ifupdown" to the string which is assigned to IMAGE_INSTALL_append in local.conf. After "bitbake core-image-minimal" and booting that new image I initially don't have eth0 but after "/etc/init.d/network stop" and "/etc/init.d/network start" I have eth0 now! :-) Jan 22 14:30:23 yoctonator: i'd recommend looking at your needs and maybe just using connman, so much nicer Jan 22 14:32:02 rburton: never heard about connman... Is it a tool for pimping up a core-image-minimal with network support? :) Jan 22 14:32:16 its a tool for management of networking Jan 22 14:32:18 like network-manager Jan 22 14:32:50 I wonder which packages I also need to add to core-image-minimal for a working network Jan 22 14:33:11 init-ifupdown is one solution, it assumes that you write the config files for your networking configuration Jan 22 14:33:25 whereas connman will plug-and-play wired networking for you Jan 22 14:33:28 (and handle wifi) Jan 22 14:34:18 zeddii: thanks. I should have tested this in isolation too, sorry Jan 22 14:36:39 rburton: Is connman a tool for the setup of building a yocto image or a tool for configuring the network on my running yocto linux on my eval board? Jan 22 14:36:51 yoctonator: recipe you install into the image Jan 22 14:36:56 (latter) Jan 22 14:38:52 rburton: ah OK, do I need to add "connman" to IMAGE_INSTALL_append in local.conf as well? Jan 22 14:39:01 yes Jan 22 14:39:23 rburton: thanks. I'll give it a try Jan 22 14:40:08 RP: bah. no worries. I didn't wait for a follow up to my question before sending the patches either. I can't fully test the new machine yet, so I can't pickup runtime issues as easily (although I could have found this by trying some things with the old qemuarm tests). Anyway, that patch should fix it, if it doesn't, yell at me right away Jan 22 14:41:20 zeddii: will do, will put it in for testing and check things are good Jan 22 15:00:55 RP: can we do another pass at virgl-gtk? This time everything should be fixed, except centos is known to fail one of the virgl selftests. Jan 22 15:01:13 rburton: Yippie! :-D After I've added the "connman" package to IMAGE_INSTALL_append in local.conf the eval board reports it has eth0 now!!! And even it gets an IP address from DHCP, already. Awesome! :-) Jan 22 15:01:43 * yoctonator is ordering a new round of free beer for the whole chat here :-) Jan 22 15:02:51 RP: rburton : do you know where/who uses core-image-minimal-initramfs.bb, and why it 'depends' on "live" stuff? the "live image" thingy creates a strong dependency on grub (and x86). the name core-image-minimal-initramfs would tend to mean that it should be a 'generic' initramfs image for any arch. Jan 22 15:03:11 removing the the live modules from that image makes it work on any arch. Jan 22 15:03:13 Any information about calling a python function from a bash function ? :D Jan 22 15:08:25 kanavin: we can Jan 22 15:08:49 ndec: I do not remember Jan 22 15:09:08 PinkSnake: needs a new python interpreter so you have no context Jan 22 15:10:05 kanavin: https://autobuilder.yoctoproject.org/typhoon/#/builders/85/builds/145 Jan 22 15:10:35 RP: thx but could you please retry with simple words ? ;) Jan 22 15:11:04 RP: i'm looking for the opposite of Jan 22 15:11:12 PinkSnake: you can only do it if its standalone code Jan 22 15:11:20 PinkSnake: can't share bitbake context Jan 22 15:11:56 RP: it's not possible to use the opposite of bb.build.exec_func ? And thx again for repplying ;) Jan 22 15:12:22 RP: is this image used in the testing? e.g. are we breaking anything if we remove the live things there? Jan 22 15:12:23 ndec: I think the usecase for that image was as an installer and it should really be called "initramfs-installer" Jan 22 15:12:55 ndec: some renaming and a generic initramfs image would be good Jan 22 15:13:14 ndec: I think its used by the live image type? Jan 22 15:13:48 PinkSnake: the shell equaivalent is "python /path/to/python/script" Jan 22 15:14:01 RP: thanks! Jan 22 15:14:37 RP: thx! Jan 22 15:16:49 Ok, I feel like I'm going a little insane.... where is the FAKEROOT variable set? I can see it being used, but not assigned Jan 22 15:20:22 JPEW: bitbake image-recipe-name -e > en.dot Jan 22 15:22:00 JPEW: It may be obsolete in the way we run pseudo Jan 22 15:24:17 RP: I did see the comment about that. Does that mean that https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/bitbake/lib/bb/build.py#n370 and https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/lib/oe/rootfs.py#n172 are dead code? Jan 22 15:26:35 JPEW: I suspect so Jan 22 15:27:37 RP: Fair enough. It seemed like really important code, I couldn't figure out how it would have worked correctly :) Jan 22 15:28:03 JPEW: replaced by FAKEROOTCMD iirc Jan 22 15:28:09 RP: Yep Jan 22 15:28:26 JPEW: that was a while ago :) Jan 22 15:28:58 -EDATACOVEREDINDUST Jan 22 15:29:53 RP: Ok, I'll submit a patch to remove the dead code so no one else falls prey to it Jan 22 15:30:26 JPEW: well, keep in mind that someone using bitbake generically could conceivable find the old API useful Jan 22 15:30:50 JPEW: the rootfs.py mention is just odd though as that code is much newer Jan 22 15:31:08 Ya Jan 22 15:48:48 Uhm, my subdir ./tmp/deploy/ipk does not contain a file Packages. How is it generated? Because calling "opkg update" gives an error about Packages.gz is missing Jan 22 15:51:03 on your target, if you don't have package-management (image feature) enabled, you won't get a database in the rootfs Jan 22 15:52:24 I have added "package-management", "opkg" and "package_ipk" Jan 22 15:52:47 yoctonator: bitbake package-index Jan 22 15:53:40 i've got half a class to use deploydir directly as a feed that handles both that and running a http daemon but its not quite ready ye Jan 22 15:53:41 yet Jan 22 15:53:48 ndec: the live images use it, iirc Jan 22 15:55:58 Woah!! This is awesome here B-) I've did "opkg update" and "opkg install mytestpackage" and it succeeded!!! I love you all guys. Thanks so much! Jan 22 15:56:12 bitbake package-index did the job Jan 22 15:56:33 * yoctonator is ordering another round of free beer for the whole chat B-) Jan 22 15:58:11 I followed the steps described here and I just can recommend those 3 tutorial videos: https://www.youtube.com/channel/UC4nV9mp2HJbt0FzrkqacTBQ Very good for beginners Jan 22 15:58:59 This combined with your hints here made my day, guys. Thanks a lot and see you next time! Bye Jan 22 16:05:12 rburton: i find it confusing that core-image-minimal-initramfs turns out to be specifically designed to work with live images. no? core-image-live-initramfs would seem more appropriate.. and it would also mean we would have core-image-minimal-initramfs as a good generic initramfs .. Jan 22 16:27:55 * armpit https://marketplace.eclipse.org/content/nodeclipse-phantomjs Jan 22 16:28:06 they do have one ; ) Jan 22 16:29:40 armpit: just don't Jan 22 16:36:13 RP: qemuarm-lsb fails in qemu, known issue? Jan 22 16:42:14 did anyone else got unsubscribe for the ML due to bounces? Jan 22 16:57:57 mckoan: Yes, it seems to happen somewhat regularly for me on my gmail account Jan 22 17:05:19 kanavin: yes, -next should be fixed Jan 22 17:08:46 RP: right, I thought so. otherwise, looks good so far - one unrelated failure in oe-selftest, seems to be one of those random qemu failure flukes Jan 22 17:09:06 kanavin: hmm, which random fail was it? :/ Jan 22 17:09:24 RP: https://autobuilder.yoctoproject.org/typhoon/#/builders/56/builds/149 Jan 22 17:09:26 rburton: "One of the SDK test cases “/poky/meta/lib/oeqa/sdk/cases/python.py”, while executed, is giving a result “unknown”. We are not sure about the reason of this behavior and investigating more into it. This test case is a part of seven ADT test plans." Jan 22 17:09:34 rburton: can you remember whether that was something you fixed? Jan 22 17:09:57 mckoan, Yes. It is happening enough we are working toward a new set up. Jan 22 17:11:21 kanavin: boot login timeout. I wonder if the test used qemuarm Jan 22 17:12:44 kanavin: x86_64 so no. Odd and we're not tracking a failure like that :( Jan 22 17:14:04 RP: indeed - qemu is used in plenty of other tests in oe-selftest, so it seems to generally work Jan 22 17:14:23 kanavin: I'll rerun selftest with the branch Jan 22 17:17:01 RP: thanks. I'd like to confirm it's not my fault :) Jan 22 17:17:27 armpit: talking with halstead we've decided to move the script from poky scripts/contrib into yocto-ab-helper Jan 22 17:17:56 armpit: that should remove some of the backporting needed there and let us beat it into shape for our infrastructure Jan 22 17:19:17 kanavin: https://autobuilder.yoctoproject.org/typhoon/#/builders/56/builds/151 Jan 22 17:20:05 RP: cheers, let's see how it goes Jan 22 17:22:23 RP. ok Jan 22 17:23:12 armpit: we're likely going to modify things quite a bit and drop phantomjs for example Jan 22 17:23:26 just place the html reports onto a webserver Jan 22 17:24:12 that would make it simpler going forward Jan 22 17:24:39 armpit: yes, I like simpler Jan 22 17:24:55 emails can have an http link Jan 22 17:25:16 yep Jan 22 17:25:27 * armpit heading out of the hill Jan 22 17:30:49 RP: the previously failed oe-selftest case has now passed, oe-selftest is ongoing otherwise Jan 22 17:31:16 kanavin: you're off the hook then :) Jan 22 17:32:59 RP: I had also written down the parallelization of upstream checks from your suggestions, it's not the prettiest code ever, but it does cut down the time from 12 minutes to 90 seconds: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=akanavin/package-version-updates&id=c69c22b9d2ed961045e9f9c0c6497ac062026ac5 Jan 22 17:34:03 with this, writing a plugin to bitbake-layers to present this info is feasible again (replacement for the old csv code) Jan 22 17:34:14 kanavin: probably worth doing then. You could neaten it slightly with a list of variables and function Jan 22 17:34:46 RP: yes, wip :) I was eager to see if the parallel part works, and added those variables one by one Jan 22 17:35:15 also, ProcessPoolExecutor, not ThreadPoolExecutor, because GIL :( Jan 22 17:35:20 kanavin: I wasn't sure quite how bad it would be and its probably not as bad as I feared Jan 22 17:35:34 kanavin: not surprised :/ Jan 22 17:36:39 RP: I didn't try ThreadPool tbh, just went straight to processes. It would probably still work, as I/O waits do unblock the threads. Jan 22 17:36:50 but if you're doing computation in threads, python is a bad choice :( Jan 22 17:37:11 kanavin: it is what it is... Jan 22 17:38:17 kanavin: I've grown to quite like python Jan 22 17:39:57 RP: yes, I like it too, there was really nothing quite like Python before it came along. But I have to wonder if it's going to be misplaced by Go in the medium to long term. Jan 22 17:40:44 nothing quite like Python - readable syntax ('executable pseudocode' as some said), 'batteries included' - rich standard library, interpreted for rapid prototyping Jan 22 17:40:53 I think this was the first language to have all three Jan 22 17:41:10 i doubt go will be replacing python anytime soon. only for certain cases Jan 22 17:41:19 kanavin: Ironic I set out with bitbake thinking python was its speed problem Jan 22 17:41:33 a lot of devs love go, but a lot aren't fans of its opinionated nature and error handling overhead Jan 22 17:41:40 * kergoth shrugs Jan 22 17:42:01 textual overhead, that is, not performance overhead Jan 22 17:42:31 Go, Rust, and Elixir are all on my list to play with, but need to find the motivation to code outside of work to do that... Jan 22 17:43:51 Anyone else played with selinux + docker in a yocto/oe-based distro yet? Not seeing a container-selinux recipe yet Jan 22 17:44:44 * RP should play with other things more Jan 22 17:45:18 hard to find the time, energy, and motivation :\ Jan 22 17:45:27 when i'm not at work i just cannot care less about coding nowadays Jan 22 17:45:32 which makes me sad Jan 22 17:46:07 i think i'd really enjoy playing with elixir. the benefits of erlang but nicer syntax? can't argue with that Jan 22 17:47:11 ugh, refpolicy stages sepolgen.conf to point to the include/Makefile for building modules on target, but none of hte paths are approrpiate for use by another recipe to build at build time. why bother staging it then? Jan 22 17:50:57 kergoth: At the weekend I did sort out a load of electronics stuff and clear the workbench, think I'm more likely to play with electronics hardware than software... Jan 22 17:52:14 halstead: I re-subscribed thanks Jan 22 18:57:37 RP: the old test was nonsense, what release was that against? Jan 22 19:55:55 rburton: 2.6.1 QA report Jan 22 20:07:36 zeddii: that fixed it thanks Jan 22 21:30:40 Is there a IMAGE_FSTYPES I can just dd to a partition, re-run grub-mkconfig, and reboot into the Yocto image ? Jan 22 21:31:32 broadly speaking.. Jan 22 21:39:18 kroon: sounds a bit like the live one Jan 22 21:42:17 New news from stackoverflow: How to setup an own device tree for a RaspberryPI in yocto? Jan 22 21:44:01 RP: ah yes.. maybe thats what I really want Jan 22 21:50:41 kroon: we kind of hate that image type but it does have uses Jan 22 21:52:08 RP: my plan was to just run a Yocto dist natively on my laptop Jan 22 21:56:15 selfhosting Jan 22 21:57:12 gotta read up on booting Jan 22 22:00:06 kroon: see the build-appliance image too - that is a selfhosting image Jan 22 22:02:50 RP: thanks Jan 22 22:31:37 Hi guys, I'm writing a .bb file for my package and I'm having trouble in the do_install process. Basically I have lots of files nested in multiple directories structure. I'm using install to copy files but install don't support in creating directories recursively. Any tips on how to get this right? Jan 22 22:33:17 use cp and then correct the ownership with chown if needed Jan 22 23:29:05 | install: cannot stat 'arch/arm/boot/fitImage.initramfs': No such file or directory Jan 22 23:29:17 any ideas why I see this using kernel-fitemage class Jan 22 23:29:26 trying to build a fitimage with a radisk Jan 22 23:29:43 I've done this on like rocko, but thud seems different Jan 22 23:32:33 Crofton|work: fitimages don't have good regression tests? Jan 22 23:32:50 * RP suspects that wasn't the right answer Jan 22 23:40:20 * zeddii_home suspects fitimages should hit the gym! Jan 23 00:06:29 * armpit Crofton|work is the regression test Jan 23 00:07:43 damn it Jan 23 00:08:15 open a bug Jan 23 00:08:48 * armpit wonders if one can drink beer in a driver-less car? Jan 23 00:09:26 manju may have already fixed in master Jan 23 00:09:35 0f87f8522a2c1b5a7042738887952b98060fbd9d Jan 23 00:09:39 look ssuspicious Jan 23 00:09:40 * armpit looks Jan 23 00:11:40 Crofton|work, you can only use Poky commits in this channel Jan 23 00:11:49 oh shit Jan 23 00:11:53 wrong channel Jan 23 00:12:06 this is embaressing Jan 23 00:12:26 I had always wondered what the difference was between this and #oe, now it is clear Jan 23 00:12:48 : ) Jan 23 00:13:13 Crofton|work, I'll grab that commit Jan 23 00:13:13 ok build finished, now must test, but mayb enot this evening Jan 23 00:13:33 armpit, thanks Jan 23 01:43:06 New news from stackoverflow: How to build redhat root file system? Jan 23 02:13:11 New news from stackoverflow: How to build redhat root file system? [on hold] Jan 23 02:30:50 should do_populate_sdk honor ASSUME_PROVIDED? I have nativesdk-perl in ASSUME_PROVIDED, but it fails to install nativesdk-autoconf due to unresolved dependency **** ENDING LOGGING AT Wed Jan 23 02:59:57 2019