**** BEGIN LOGGING AT Wed Nov 22 03:00:02 2017 Nov 22 06:52:03 I want to the customer kernel header files in a recipes, I found do_configure[depends] += "virtual/kernel:do_shared_workdir" Nov 22 06:52:31 but it doesn't work for me, so I want to check whether the kernel header files generate from the kernel is installed Nov 22 06:52:53 where is the staging directory of the kernel? Nov 22 07:14:45 does do_configure[depends] also apply for those recipes inherit from cmake class? Nov 22 07:17:23 it does Nov 22 07:29:07 joshuagl, I think I know what I miss, do_shared_workdir looks like not install the files like the headers_install does Nov 22 07:29:35 so my kernel headers is not installed at all Nov 22 07:31:51 do_compile[depends] += "virtual/kernel:do_shared_workdir" ? Nov 22 07:33:05 nayfe, it doesn't work either. the problem is the what do_shared_workdir does seems not what I want Nov 22 07:33:58 the kernel files I need is a file placed at include/uapi/video/ but I read the kernel class Nov 22 07:34:24 the do_shared_workdir seems not do kernel header_install first then copy files Nov 22 07:35:31 you also set DEPENDS = "virtual/kernel" is suppose ? Nov 22 07:37:44 nayfe, it work neither Nov 22 07:44:42 stdint: that EXTRA_OECONF = "--kernel-headers=${STAGING_KERNEL_DIR}/include" ? Nov 22 07:47:04 nayfe, it doesn't work, You see those files are installed into there http://paste.debian.net/996983/ Nov 22 07:47:39 there are not installed into STAGING_KERNEL_BUILDDIR at all Nov 22 07:49:31 STAGING_KERNEL_DIR and STAGING_KERNEL_BUILDDIR are identical ? Nov 22 07:58:15 nayfe, Yes, I tried, it failed. I think the cmake doesn't support kernel-headers either Nov 22 08:03:03 even the EO_SOURCE_DIR doesn't not work for me Nov 22 08:04:31 stdint: sorry, i joined irc late, can you tell what you want to do ? and if you can pastebin your recipe maybe ? Nov 22 08:05:44 nayfe, I have a create a recipe for a library, the library will communicate with the kernel Nov 22 08:06:18 using the ioctl, the define of the ioctl is in the kernel, I have verified the headers_install will install it Nov 22 08:06:37 then I can't find a way to make the recipe to find that kernel headers Nov 22 08:07:04 the previous doesn't use the external kernel header, and I have verified it work Nov 22 08:07:38 with the toolchain created by yocto, once I place that header files into sysroot, it can compile that package Nov 22 08:08:08 stdint: is it a private lib ? Nov 22 08:08:23 nayfe, no, you can find it at github Nov 22 08:08:42 https://github.com/rockchip-linux/mpp Nov 22 08:21:19 New news from stackoverflow: Manually building a Kernel source from Yocto build Nov 22 08:45:53 stdint: maybe you need to add something like https://github.com/linux-rdma/rdma-core/blob/master/buildlib/RDMA_LinuxHeaders.cmake to mpp Nov 22 08:47:13 nayfe, that is nosense, I just need the kernel files are installed in the /usr/include Nov 22 08:47:21 that is the standard linux way Nov 22 08:50:25 stdint: you have to tell cmake where to search for kernel headers, and while cross-compiling, they can be anywhere. If mpp does not have any KERNEL_DIR variable or equivalent, it wont work Nov 22 08:50:47 nayfe, no, the cmake will search the sysroot/usr/include Nov 22 08:51:10 they don't need KERNEL_DIR or something else Nov 22 08:51:36 the cmake doesn't regard it as a kernel header files, just a regular file Nov 22 08:52:10 you will find the yocto has installed them into the staging of for the recipe but just the one used to build libc Nov 22 08:52:14 not the customer one Nov 22 09:16:41 stdint: true, that's why you need to point to STAGING_KERNEL_DIR and not linux-libc-headers Nov 22 09:17:14 nayfe, I think the normal is install the files into recipe-sysroot/usr/include Nov 22 09:18:03 so what I need is a way to install the kernel header files into there, but I can't find a proper way Nov 22 09:23:45 stdint for me something like EXTRA_OECONF = "CPPFLAGS="-I${STAGING_KERNEL_DIR}/include/uapi -I${STAGING_KERNEL_DIR}/include"" should be enough, but, if you prefer to copy files you can do something like do_configure_prepend() { cp -rf ${STAGING_KERNEL_DIR}/include/uapi ${SYSROOT_DIRS}/.... } Nov 22 09:27:08 stdint: similar to what nayfe told you, CXXFLAGS += " -I${STAGING_KERNEL_DIR}/include should work, otherwise your build configuration has something wrong Nov 22 09:30:31 zero_note, yes it will work, but not the correct Nov 22 09:31:09 you can't use the kernel header from kernel directly, you should always use the header_install Nov 22 09:31:17 or some wrong file will be installed Nov 22 09:32:20 and it looks ugly, is there a way to make the customer kernel header become a package Nov 22 09:33:17 stdint: you mean something like kernel-devsrc? Nov 22 09:35:23 DEPENDS += "kernel-devsrc" still won't install the kernel header into recipe-sysroot Nov 22 09:39:06 wait, is your goal to compile your library with the yocto-created toolchain? with the kernel headers, etc... Nov 22 09:41:22 stdint: customer kernel is not the one shipped ? Nov 22 09:42:09 there is solution but i'm not sure it is the one to use: https://lists.yoctoproject.org/pipermail/linux-yocto/2014-April/002178.html Nov 22 09:42:11 stdint: if so, that's nothing related to your library recipe Nov 22 09:42:13 nayfe, the customer kernel is compiled by the yocto, and as the prefer kernel Nov 22 10:24:17 nayfe, I think that is what I want https://paste.fedoraproject.org/paste/MY0voVXFvywylhYOQynvtg/ Nov 22 10:24:34 but the problem is how to make the recipe depends on the linux-rockchip-headers? Nov 22 10:24:57 I can only set DEPENDS += "linux-rockchip" Nov 22 10:27:56 linux-rockchip-headers is installed with linux-rockchip, so DEPENDS += "linux-rockchip" is ok Nov 22 10:31:05 nayfe, yes, and I saw the recipe name appears at recipe-sysroot/sysroot-providers/ Nov 22 10:31:17 but the files are not installed into the recipe-sysroot Nov 22 10:35:46 Hi all Nov 22 10:37:16 or I should have a try on sysroot_stage_all_append Nov 22 10:45:18 I think right solution is to use STAGING_KERNEL_DIR, maybe add an issue in mpp github to get cmake to support it. Mixing kernel includes looks bad to me. Nov 22 10:45:46 nayfe, I don't think so Nov 22 10:46:04 STAGING_KERNEL_DIR is a wrong settings Nov 22 10:46:08 you can see in http://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-kernel/linux-libc-headers/linux-libc-headers.inc begins it looks to be the right thing to do Nov 22 10:47:04 first do_configure[depends] += "virtual/kernel:do_shared_workdir" won't install the necessary files for you Nov 22 10:47:49 second settings an external directory in default cmake is a wrong way, that is what the cmake toolchains files' job Nov 22 11:21:53 New news from stackoverflow: Python Subprocess.Popen not work for Cross platform Yocto Nov 22 11:51:58 New news from stackoverflow: How to include GL/gl.h in SDK generate with Yocto Nov 22 13:34:15 hello Nov 22 13:34:39 are there official dropbox images for building images with yocto? Nov 22 13:34:55 i'm interested in centos 7 images Nov 22 13:35:08 dropbox images? Nov 22 13:35:11 ... Nov 22 13:35:28 oops Nov 22 13:35:32 docker, i mean :) Nov 22 13:36:13 cornel: kind of. look at the CROPS stuff Nov 22 13:36:14 cornel: yocto creates a custom distro, so you will not have centos Nov 22 13:37:07 cornel: I bookmarked that https://github.com/gmacario/easy-build never used it for now Nov 22 13:37:16 LetoThe2nd, thank you Nov 22 13:37:49 nayfe, i don't know who is gnmacario but i trust the yocto project guys Nov 22 13:39:20 I think AGL provide yocto docker envs, you could probably use one of those Nov 22 13:40:29 cornel: depending on your needs, jsut roll a custom one based on the official fedora/whatever install instructions. especially when you want to use volume mounts, this is often an easier solution. Nov 22 13:41:26 https://github.com/crops/yocto-dockerfiles/blob/master/dockerfiles/centos/centos-7/centos-7-base/Dockerfile ? :p Nov 22 13:41:32 LetoThe2nd, i hoped i would never have to manually pack a linux file system again :) Nov 22 13:42:09 CTtpollard, i'm looking after them now, thank you Nov 22 13:58:27 hi! anyone here dealing with ACLs? :D Nov 22 14:05:11 cornel: automating ubuntu + a handful of packages installed in a dockerfile is not exactly magic :) Nov 22 14:07:46 btooth: via xattrs? Nov 22 14:11:31 rburton: using setfacl, we use fido (still, and yes fido :-) - but for this issue it should be same for HEAD of yocto) - we introduced a "fakeroot sandboxing()" task before do_package_write_ipk after do_packagedata -- inside this task we call setfacl on files in ${WORKDIR}/packages-split/${PN} Nov 22 14:11:44 this works and we find the acls inside the ipks Nov 22 14:12:05 you did? didn't think opkg supported those Nov 22 14:12:32 but when doing setfacl on a directory, the directory looses the acl and owner/group, falling back to system user. Nov 22 14:13:02 rburton: yes we did. we use patched libarchive and options to opkg for this Nov 22 14:13:09 and it is working Nov 22 14:13:16 can you send patches? :) Nov 22 14:13:30 but seems there is a bug when appling on directories Nov 22 14:14:46 rburton: i will try to extract all the things we did/patched ;-) Nov 22 14:29:08 rburton: is it too late to file a bug for major version updates that I think should be done in this release cycle? Nov 22 14:29:15 python 3.6 in particular stands out Nov 22 14:34:41 rburton: is it too late to file a bug for major version updates that I think should be done in this release cycle? python 3.6 in particular stands out Nov 22 14:34:52 kanavin: no, please do Nov 22 14:37:40 rburton: we already missed that update twice Nov 22 14:37:47 3.6's been out for almost a year Nov 22 14:38:02 and once 3.7 is out, we have double the amount of work to update Nov 22 15:00:08 rburton: looks like python is really the only problematic spot Nov 22 15:00:24 rburton: perl is also a bit behind (5.24->5.26), but who cares :) Nov 22 15:07:10 hi all, I'm trying to apply a patch in my example recipe. I'm getting error that patch file is not good... so can anyone tell me what is the default way to make a patch file, method I'm using now is "diff -u file1 file2 > file.patch" Nov 22 15:08:36 xtron: i personally have everything i want to patch in a git repo, and there it is just git format-patch -o $WHEREYOUWANTIT $THEREVISIONYOUWANT Nov 22 15:09:08 xtron: IIRC otherwise, the canonical form is diff -Naur in the root directory of the sources. Nov 22 15:10:00 the yocto plugin is not very good, screwing up console output constantly Nov 22 15:10:36 LetoThe2nd: does directory matter when we create patch file ? Nov 22 15:11:02 xtron: of course it matters. Nov 22 15:12:40 LetoThe2nd: what is the problem if we create patch file while standing in the directory where both files are, after all it is about the different between files not their paths Nov 22 15:14:22 xtron: and how would the patch utility know where to apply it? there is no guarantee that in a source tree there are no files of identical name, but separate directories Nov 22 15:14:47 xtron: and I am very sure that the error message was not "error that patch file is not good" Nov 22 15:17:11 LetoThe2nd: obivously :P ... here is the error https://pastebin.com/AqFkq2uv Nov 22 15:20:49 xtron: its exactly what i said: the directories are off. please read up on patch (-> man patch), and learn about the -p switch. it really makes sense being able to do the steps manually. and also take a good look at the patch file you generated: the files and pathes are explicitly stated in there. Nov 22 15:24:13 LetoThe2nd: ok Nov 22 15:28:57 With systemd and read-only-rootfs feature, do you know how to make a service to run before remounting RO ? Nov 22 15:30:13 that's not a good path to take. what exactly are you trying to accomplish? Nov 22 15:30:16 nayfe: before? why would there be a remount at all? Nov 22 15:30:34 if the kernel cmdline has ro, it'll start read only, not start rw and remount ro Nov 22 15:30:43 it'd only remount from rw to ro if it was ro in fstab, i believe Nov 22 15:30:49 regardless, it's a bad approach Nov 22 15:31:02 if you just want to populate a path or something, use systemd's tmpfiles mechanism Nov 22 15:34:08 indeed, i just see it now, i didn't see that /etc/machine-id was tmpfs Nov 22 15:34:22 i want to make something close :) Nov 22 15:34:33 my mistake ;) Nov 22 15:34:55 :) Nov 22 16:04:02 So Im looking for kernel params passed to uboot. I can see this recipe in poky Nov 22 16:04:03 meta-fsl-arm/recipes-kernel/linux/linux-imx.inc Nov 22 16:04:10 KERNEL_EXTRA_ARGS Nov 22 16:04:36 of course, this refers to my board: Nov 22 16:04:37 meta-fsl-arm/conf/machine/imx28evk.conf Nov 22 16:06:21 Would be nice to specify extra kernel params in my local.conf if thats possible? Nov 22 16:07:28 how they're passed completely depends on the board Nov 22 16:07:55 some u-boots are configured with support to load an external environment file, that's the easiest method afaik Nov 22 16:08:39 right. Its hard to find where this comes in. I did hear about the env files yes Nov 22 16:09:54 the best resource would be the info about your hardware and how it boots, or directly examining its images / partitions. i.e. if it boots from sd, is there an environment file on a functioning sd card image for that board already? if not, you'd probably want to examine the u-boot recipe to see how u-boot is being configured, to see if such support is enabled Nov 22 16:10:35 Yeah, the latter im looking for. The image is on an SDcard. Not sure if I can mount it locally mind you Nov 22 16:12:09 There appears to be a compressed zimage and not a lot else :/ Nov 22 16:12:53 can't say i'm super familiar with the boot process of freescale arm boards, could check the readmes in the layer too Nov 22 16:13:01 or any docs that come with the sdk that freescale releases Nov 22 16:13:11 otherwise yeah, check the config file for the u-boot build Nov 22 16:13:22 Yeah, the board in question comes from a third party. Its very light on the docu and support Nov 22 16:13:34 or ask otavio, if he happens to be around? Nov 22 16:13:35 * kergoth shrugs Nov 22 16:13:51 Ultimately, I just want the screen to appear. My working theory is that the framebuffer isnt attached Nov 22 16:14:00 so I need to pass the correct video params to the kernel Nov 22 16:14:03 its a thought Nov 22 16:16:08 thanks :) I shall persevere Nov 22 16:16:40 ah, yeah, video params are almost certainly the issue, sadly i can't be of much use there, configuration for lcd panels is voodoo to me :) Nov 22 16:16:45 good luck with it Nov 22 16:17:03 thanks :) Its the only thing I've seen in the logs that uboot complains about Nov 22 16:17:09 'no video parameter' Nov 22 16:17:31 tbh, I'd love to know more about framebuffers anyway so I can have my own fancy bootup screens o Nov 22 16:21:04 otavio, JaMa: I'm trying to sync up on rocko with meta-qt5 and meta-boot2qt, but qtwebengine is failing on a glibc2.26 issue that was fixed, but later removed (cfree vs free). I can see the fix being added, but it's been removed in all branches I've seen so far. Nov 22 16:21:14 rburton: WARNING: neon-0.30.2-r0 do_fetch: Failed to fetch URL http://www.webdav.org/neon/neon-0.30.2.tar.gz, attempting MIRRORS if available Nov 22 16:21:33 kanavin: yes, there's a bug. my proposal was to remove neon :) Nov 22 16:21:33 rburton: how come this isn't caught by autobuilder? I thought checkuri is exactly for that purpose? Nov 22 16:21:38 looks like the patch was removed during a larger merge from upstream or something, but why? Nov 22 16:21:48 rburton: yes, I will send a patch that removes neon in a minute ;) Nov 22 16:22:07 kanavin: maxin had the bug so check with him first in case he's sitting on patches Nov 22 16:22:18 (needs to be removed from a few other mentions in oe-core too) Nov 22 16:22:38 rburton: yeah, already done - I just finish up the meta-oe move, as it needs to be taken from debian mirror Nov 22 16:23:02 rburton: anyway my concern is that this isn't caught by checkuri, or other checks, or is it? Nov 22 16:23:45 checkuri spotted it, thats how theres a bug Nov 22 16:23:48 wto: where it was removed? Nov 22 16:24:02 kanavin: https://bugzilla.yoctoproject.org/show_bug.cgi?id=12361 Nov 22 16:24:03 Bug 12361: normal, Medium+, 2.5 M1, maxin.john, NEW , Fetcher failure for URL: "http://www.webdav.org/neon/neon-0.30.2.tar.gz" Nov 22 16:24:05 wto: it might be removed when the same fix was applied in upstream repo Nov 22 16:24:23 maxin: I have the patch already Nov 22 16:24:35 JaMa: http://code.qt.io/cgit/yocto/meta-qt5.git/commit/?h=upstream/rocko&id=bbb304456978fa9713932e9ff529d18a5ef2a947 for upstream/rocko Nov 22 16:24:55 kanavin: ok.. please go ahead then Nov 22 16:24:58 wto: but currently it builds OK with master, so if you're using the revision defined in meta-qt5 then it should work Nov 22 16:25:01 rburton: I ran checkuri, and it was happy, but maybe I ran it wrongly Nov 22 16:25:10 hm checkuri has been failing for a long time. fired a fresh one against master to see. Nov 22 16:25:23 wto: that's not my and otavio's meta-qt5, as The Qt Company... Nov 22 16:25:39 Oh, I see. Nov 22 16:25:50 wto: https://github.com/meta-qt5/meta-qt5 is the one which works for me Nov 22 16:26:03 maxin: yeah, it's just that I was doing the usual upstream version check cleanups, and this stood out as 'the dead upstream' Nov 22 16:26:28 JaMa: alright, I'll try that. Thanks. Nov 22 16:26:43 JaMa: which one should be considered "upstream" in general? Nov 22 16:27:04 wto: I'd take ours ;-) Nov 22 16:27:10 ofc :) Nov 22 16:27:19 wto: https://github.com/meta-qt5/meta-qt5 is the upstream, http://code.qt.io/cgit/yocto/meta-qt5.git is the fork of it by Qt Company Nov 22 16:28:34 Alright, nice. Nov 22 16:31:28 rburton: checkpkg is nowadays also capable of reporting dead upstreams ;) Nov 22 16:40:44 rburton: ah, checkpkg should not be run directly, but rather via selftest? Nov 22 16:40:52 *checkuri, sorry Nov 22 16:42:57 This may be a very silly question but where do the targets (like core-image-minimal) get their package lists from? Nov 22 16:43:12 theres a lot of stuff here I don't need. Would be nice to remove it Nov 22 16:43:29 oni: images are defined in recipes, and those include the list of what to pull in Nov 22 16:43:47 oni: the best way is to simply write your own image recipe Nov 22 16:43:53 Ok, I'll take a closer look at that. Cheers Nov 22 16:55:15 rburton: https://autobuilder.yocto.io/builders/nightly-checkuri/builds/435 Nov 22 16:55:21 rburton: checkuri was a success :-/ Nov 22 16:55:27 interesting Nov 22 16:56:20 oh i think checkuri follows some mirrors Nov 22 16:56:25 rburton: https://autobuilder.yocto.io/builders/nightly-checkuri however here you can see there is a different test, where checkuri does fail Nov 22 16:56:32 yes Nov 22 16:56:34 very odd Nov 22 16:56:51 rburton: I guess it needs to be configured to not fall back to the yocto mirror Nov 22 16:57:43 rburton: and there's another oe-selftest failure here due to missing distrodata https://autobuilder.yocto.io/builders/nightly-checkuri/builds/434 Nov 22 16:59:41 kanavin, I have seen that on Morty.. "--run-tests distrodata.Distrodata.test_checkpkg" looks wrong for that branch Nov 22 17:00:05 is that an AB config thing? Nov 22 17:00:12 armpit: it's a test that is only meant for tip of master branch really Nov 22 17:00:22 oh Nov 22 17:00:27 armpit: it bitrots over time, and we don't keep it up to date anywhere except master Nov 22 17:00:37 includes added to yocto project (eclipse yocto plugin) are completely ignored Nov 22 17:01:24 k. AB needs a tweek for other branches.. guess I should log a bug Nov 22 17:01:42 kanavin, thanks for the info Nov 22 17:02:49 armpit: the test makes sure that latest upstream versions are detected correctly, which is something master branch needs, but stable branches, not really Nov 22 17:08:10 the video on this page is from 2010 LOL https://www.yoctoproject.org/tools-resources/projects/eclipse-ide-plug Nov 22 17:21:02 * armpit wonders what a meta-ab would look like. Yocto ab for embedded devices Nov 22 17:40:02 armpit: I have bad news about morty, its hanging everywhere :( Nov 22 17:40:23 armpit: looks like do_rootfs in smart, some kind of db pthread mutex issue :( Nov 22 17:40:39 armpit: I tried to run a release build but its dead in the water Nov 22 17:41:00 armpit: also f26 issues with fontconfig-native and tumbleweed issues with ccache-native and signing Nov 22 17:41:14 I think rburton is looking at the fonctconfig issue Nov 22 17:41:20 (and knows how to fix it) Nov 22 17:50:17 wow Nov 22 17:51:55 does the yocto plugin work with oxygen cdt eclipse? Nov 22 17:51:56 yeah just sent the patch Nov 22 17:52:00 armpit: ^ Nov 22 17:52:43 Is there a way to do a make clean with bitbake? Nov 22 17:52:58 re-read and reparse recipes and remove intermediates? Nov 22 17:53:09 I see this Nov 22 17:53:11 bitbake -c clean core-image-minimal Nov 22 17:57:16 but it doesnt seem to reset the state as it were Nov 22 17:58:42 the include paths i add to a yocto plugin project in eclipse oxygen are completely ignored, not sent to compiler command line Nov 22 18:02:04 RP, how did it get worse ? the build before that was a lot cleaner. what changed? hosts order? Nov 22 18:03:32 * armpit looks at the failures.. Nov 22 18:04:33 armpit: I don't know :( Nov 22 18:04:41 armpit: I'm overloaded with issues :( Nov 22 18:04:57 RP no worries Nov 22 18:05:32 * armpit will see what can be done Nov 22 18:07:14 armpit: thanks. I'm just braindumping state here so at least others know... Nov 22 18:08:27 I have the font patch in my morty-next. should the current morty build continue? Nov 22 18:26:38 armpit: is it getting anywhere is it hung? Nov 22 18:27:40 armpit: we probably need to analyse the hangs, see if we can figure out why Nov 22 18:27:45 armpit: do you have ssh access in? Nov 22 18:27:53 armpit: if not, talk to halstead Nov 22 18:31:36 the include paths i add to a yocto plugin project in eclipse oxygen are completely ignored, not sent to compiler command line Nov 22 18:32:30 does the yocto plugin work with oxygen cdt eclipse? Nov 22 18:37:48 does the yocto plugin work with oxygen cdt eclipse? Nov 22 18:50:33 does the yocto plugin work with oxygen cdt eclipse? Nov 22 18:57:11 RP I don't have ssh access Nov 22 18:57:41 I will talk to Mr. Halstead Nov 22 19:19:07 what the heck is an autotools project and why does yocto plugin use it? Nov 22 19:20:36 autotools is the framework that generates most ./configure && make && make install-based build systems Nov 22 19:29:06 i guess CDT managed was too good to use? Nov 22 19:29:17 I can't even figure out how to udate the includes Nov 22 19:29:28 i hate open source Nov 22 19:30:47 autotools can be pretty painful... we have little choice about using it, we do because many of the third-party projects we build use it Nov 22 19:31:20 in some quarters it is being replaced by things like meson, so there is some hope Nov 22 19:35:40 no reason to use it for application development Nov 22 19:36:07 yocto is inspiring me to get out of the field asap Nov 22 19:36:17 what does yocto have to do with anything? Nov 22 19:36:38 yocto plugin Nov 22 19:36:54 if you don't want to use autotools, dont' use autotools Nov 22 19:37:05 + auto tools nightmare projects for application development when a yocto toolchain installation isrequired Nov 22 19:37:25 impossible to config to use a yocto generated toolchain installer Nov 22 19:37:31 with a CDT project Nov 22 19:40:32 * armpit so nice living it AG country.. smells like sewer outside :/ Nov 22 19:44:55 zarzar2: it's fair that our current eclipse plugin is lacking Nov 22 19:45:45 CDT managed C/C++ projects are the best choice for application development Nov 22 19:46:03 * rob_w leans back, grepping for popcorn Nov 22 19:52:49 yocto plugin with mars and oxygen breaks the console output so it freezes, have to restart eclipse constantly to get it to work Nov 22 20:00:15 concerning the convoluted yocto toolchain installer: all you had to do was list the location of g++, gcc, as, ld, list the defines, list the include paths, and list the library paths and we could just dump that in a CDT managed project Nov 22 21:13:57 halstead: so the kernel upgrade didn't fix kvm on f26, and it took a good week of builds to happen again Nov 22 21:17:26 hi all, can anyone point me how to generate "deprecated" global sysroot for Pyro or Rocko? Nov 22 21:17:42 now, we have separate sysroots due to reproducible builds.. Nov 22 21:18:04 but somewhere I read about some fallback (can't recall the name of recipe which puts it together) Nov 22 21:19:04 phatina: bitbake build-sysroots may be what you want, but it's definitely a workaround Nov 22 21:19:23 phatina: can you explain what you're trying to support? Nov 22 21:20:53 zarzar2: I'm afraid it's not that simple, our compilers are per-architecture not per-machine so they do need defined command lines in order to properly match the target, and these are set up by the environment setup script - that side of things is pretty straightforward though Nov 22 21:26:24 bluelightning: just to follow my workflow - I am developing some SW for raspi0w and need to crosscompile and link against some lib. I have already hacked. Nov 22 21:26:53 bluelightning: meta-ide-support gave me best tools to crosscompile and easily test SW when working on it Nov 22 21:27:11 bluelightning: I can understand, that once I have all the SW ready, it makes sense to separate sysroots... Nov 22 21:27:22 bluelightning: but now, it's a bit annoying Nov 22 21:27:36 phatina: we have our standard SDK or devtool to make that sort of thing a little easier, FWIW Nov 22 21:28:18 bluelightning: oh, do you have any link, please? Nov 22 21:30:17 phatina: for devtool (for now, ignore that it says "Extensible SDK" - devtool can be run next to bitbake as well): http://www.yoctoproject.org/docs/2.4/sdk-manual/sdk-manual.html#using-devtool-in-your-sdk-workflow Nov 22 21:35:41 rburton: would you like to upgrade and reboot today and start building uptime again? Nov 22 21:37:34 bluelightning: having seen devtool --help <- I am not sure, how I can use this thing to cross-compile my new autotools-based project which depends on already built packages... Nov 22 21:39:36 phatina: use devtool add to set the thing up for building (supplying the path to the source tree), then you can use devtool build (or bitbake) to build it and devtool deploy-target to install the latest built version to your target machine Nov 22 21:40:46 beer time Nov 22 21:41:58 * bluelightning could use a beer but the boss probably wouldn't look too kindly upon that Nov 22 21:44:00 bluelightning: thanks! will play with it Nov 22 21:49:08 night time, bye and thanks once again! Nov 22 22:13:55 bluelightning, the work boss or real boss ; ) Nov 22 22:14:20 armpit: both :) Nov 22 22:14:39 well I work remotely and my work boss wont know Nov 22 22:15:31 not only that, its the day before a holiday.. no one is really working anymoer Nov 22 22:17:20 well.. that may all change next year : / Nov 22 22:31:45 halstead: can we please not reboot f26 until we've poked more at the f26 kvm issue on it Nov 22 22:31:59 RP, absolutely. Nov 22 22:32:42 halstead: thanks Nov 22 23:04:17 * armpit the internet needs a reboot Nov 22 23:04:50 * armpit wonders if it would actual come back up Nov 22 23:08:10 it would depend on what version of u-boot it was running. Nov 22 23:13:05 or if it it running in qemu mips : ) Nov 22 23:23:54 bluelightning, I will drink on your behalf Nov 22 23:36:51 armpit: great :D **** ENDING LOGGING AT Thu Nov 23 03:00:01 2017