**** BEGIN LOGGING AT Wed Apr 05 03:00:04 2017 Apr 05 03:05:33 bluelightning: thanks for checking :-) although everything seems to be working for me now Apr 05 06:49:23 Pharaoh_Atem: not working on it right now, and it's assigned to marquiz, not me :) we cannot update the dnf stack now as we're in a pre-release freeze, so that will have to wait for a couple weeks or so Apr 05 06:50:32 morning Apr 05 07:15:41 we're having a new board that bears a '.' in the device tree file name. while the whole build process works fine, it breaks in the end when trying to create the shortened symlinks. is there some official stanza on what file names are allowed? (requirements/restrictions) Apr 05 08:27:09 RP: I'd be happy to get some advice in https://bugzilla.yoctoproject.org/show_bug.cgi?id=11239 Apr 05 08:27:10 Bug 11239: normal, Medium+, 2.3 M4, jussi.kukkonen, NEW , native recipes should not care about DISTRO_FEATURES / MACHINE_FEATURES Apr 05 08:34:11 Hi everybody =) Does anyone have already used yocto with I.MX6 ? Apr 05 08:34:23 Chrys: many, many people. Apr 05 08:34:54 Chrys: but you probably mean poky as the openemebdded based reference distribution :) Apr 05 08:34:56 I would like to add the plugins gstreamer-imx in my yocto project. Apr 05 08:36:27 So It's better to say " i use poky on I.MX6" instead of " I worked with Yocto on I.MX6" Apr 05 08:36:57 Chrys: technically, yes :-) so what is your question? Apr 05 08:36:58 I thought Poky was more an base example of what should be a distro with the use of Yocto Project tools. Apr 05 08:37:21 So my question is, ,i would like to add meta-fsl-arm-krogoth to my yocto project. Apr 05 08:37:43 ..."and"? Apr 05 08:37:50 The goal is to decode mp4 files using gstreamer-1.0 on I.MX6 Apr 05 08:38:07 So I thought that could be nice to use the gstreamer-imx plugins Apr 05 08:38:18 I've add on the bblayers.conf the meta-fsl Apr 05 08:38:27 I've bitbake my image Apr 05 08:38:45 But still when I do gst-inspect I don't have those plugins. Apr 05 08:38:57 Chrys: did you add the package(s) to your image? Apr 05 08:39:06 or the packagegroup, respectively? Apr 05 08:39:23 You mean with IMAGE_INSTALL ? Apr 05 08:40:31 thats the classic way, yes. Apr 05 08:40:33 I have IMAGE_INSTALL_append = " \ ( with a list of other package ) and add : gstreamer1.0-plugins-imx \ Apr 05 08:41:19 and when you inspect your image build with bitbake -e, that actually gets added? Apr 05 08:42:02 I'm finishing my bitbake and I will let you know thx for the command. Apr 05 08:44:04 Have you any experience with qt ? Because I would like to remove qtwebkit as it takes 1 hour to " do_compile" each time I bitbake. Apr 05 08:44:31 0: qtwebkit-5.6.2+gitAUTOINC+bb43bca166-r0 do_compile (pid 10490) Apr 05 08:44:45 well then why do you build an image that pulls it in? Apr 05 08:45:09 I don't know where it pulls it in. Apr 05 08:45:53 I've tried using IMAGE_INSTALL_remove. But I didn't tried with "PACKAGES_EXCLUDE" Apr 05 08:47:35 i mean, why do you start out with an image like it. you could also begin with core-image-minimal or such. Apr 05 08:48:23 jku: ouch :/ Apr 05 08:50:20 jku: what happens if you set this in native_virtclass_handler ? Apr 05 08:51:21 jku: I'd be tempted to swap those two if statements at the top and set this if PN has -native in it Apr 05 08:51:22 Why I shouldn't ? It's an image with qt + gstreamer + all the stuff from the board I use + special settings and so on for screens. Apr 05 08:52:37 Chrys: i'm just asking. well then find out what pulls it in. as always, bitbake -e is your friend Apr 05 08:53:06 The think is, I have meta-qt5. And Meta-qt5 include all the recipes for all the packages of qt5. Apr 05 08:53:33 So in the conf of the meta-qt5, every recipes in the meta is parsed. Apr 05 08:53:47 There is a way to exclude only one recipe ? Apr 05 08:54:47 Chrys: the recipes always get paresed. but if something gets built and pulled that you don't want, then you have to find out what causes it. usually some kind of dependency. Apr 05 08:55:58 I will try bitbake -e, thanks. Apr 05 08:58:28 RP: that seems to work based on "bitbake -e"... I'll run some more tests Apr 05 09:30:18 Hello! I have a custom recipe in which I install a systemd unit in the do_install routine, then I add this file to a package via FILES_{PN}, however after it builds, I get a QA check complain as the file has been installed but not shipped (in any package).. find shows the file under package/ subdir, do you have any hints or ideas on what could be doing wrong there? Apr 05 09:41:54 LetoThe2nd : How do you read / use bitbake -e ? Apr 05 09:43:55 I guess the best is to push the result in a file and search where qtwebkit is compile. Apr 05 09:48:45 Chrys: bitbake -e $MYIMAGEORRECIPEORWHATEVER | less Apr 05 09:49:02 Chrys: there also is some dependency graph generation, but i fogot the syntax. Apr 05 09:49:09 RP: ^^^ maybe? Apr 05 09:51:25 LetoThe2nd: bitvake -g ? Apr 05 09:53:43 oO( "bitwake" ) Apr 05 09:53:52 LetoThe2nd : Just by the fact of changing the priority of the layer meta-fsl-arm-krogoth, i get extra plugins... Apr 05 09:55:04 Chrys: funny. sounds like something in the layer (or your layer setup) is not completely right, then. Apr 05 09:55:23 otavio: ^^^ does this ring a bell? Apr 05 10:01:00 LetoThe2nd : So complicated to add gstreamer-imx plugins :s Apr 05 11:59:44 ?!? Apr 05 12:00:02 LetoThe2nd: can you give me a summary? Apr 05 12:02:05 otavio: sure. OP wanted to add gstreamer-plugins-imx, and concluded they only install if he "changed the priority" of meta-fsl-arm-krogoth Apr 05 12:05:03 Hello all, little question here. Have you any clue of why some meta when bitbake are like in the build configuration : meta-fsl-arm-krogoth = ":" Apr 05 12:06:53 Usually I get meta-name =": Apr 05 12:07:53 wll did you get the layer from git? Apr 05 12:08:19 I did. Apr 05 12:08:24 That's why is strange Apr 05 12:08:52 and what does git status say? Apr 05 12:09:27 That's the problem : Not a git repository (or any of the parent directories): .git Apr 05 12:09:35 It's not recognized as a git repository Apr 05 12:10:07 did you get it as some subdirectory of a git repo, then? Apr 05 12:10:32 or something else that results in then .git directory being somewhere else than the layer root directory? Apr 05 12:12:15 LetoThe2nd: OP ? Apr 05 12:12:30 LetoThe2nd: are you running latest meta-fsl-arm? Apr 05 12:12:30 otavio: Original Poster (Chrys) Apr 05 12:12:42 oh Apr 05 12:12:44 Chrys: Apr 05 12:12:48 Chrys: otavio is fsl-arm guy ;-) Apr 05 12:13:17 I'm redownloading it again. I'm using krogoth branch. Apr 05 12:13:22 Chrys: meta-freescale, meta-browser, meta-chicken, meta-miracast and meta-marvell ;-) Apr 05 12:13:36 Chrys: is there a specific need of using krogoth and not morty? Apr 05 12:13:58 otavio : the BSP support of my board is based on krogoth Apr 05 12:14:20 Chrys: which board is that? Apr 05 12:14:34 otavio : Phyboard I.MX6 solo Apr 05 12:14:34 Chrys: morty is way nicer ;-) Apr 05 12:14:51 otavio : the company don't supply for morty. Apr 05 12:15:25 Chrys: too bad! Apr 05 12:15:38 otavio : the only think i would like is to run a mp4 video on a Phytec PHyboard using the gstreamer-imx plugins Apr 05 12:15:38 Chrys: ok, lets see what we can do. Apr 05 12:15:47 Chrys: and? Apr 05 12:15:53 otavio : Because I would like to use the hardware decoder. Apr 05 12:16:02 otavio : thanks for your help ! Apr 05 12:16:36 Chrys: you're welcome. Push them to put the boards on community Apr 05 12:16:46 Chrys: or to hire us to do so ;-) Apr 05 12:17:05 Chrys: the addition of plugin should be straightforward Apr 05 12:17:23 Is there an open Git where I caan see their laer? Apr 05 12:18:11 Can multiple vars be excluded in the same VAR[vardepsexclude]= statement, or do I need one for each of them? Apr 05 12:18:37 otavio : git.phytec.de https://git.phytec.de/ ( use of meta-phytec and meta-yogurt ) Apr 05 12:20:02 otavio : the board phyboard-mira have an h264 hardware decoder, and I would like to use it on for mp4 video Apr 05 12:20:26 otavio : that's why i would like to add meta-fsl-arm as it gives gstreamer-imx plugins Apr 05 12:21:45 otavio : Also , i don't know if it is good but I have this error : ERROR: /opt/PHYTEC_BSPs/phyBOARD-MIRA/sources/poky/../meta-fsl-arm/recipes-fsl/images/fsl-image-mfgtool-initramfs.bb: No IMAGE_CMD defined for IMAGE_FSTYPES entry 'sdcard' - possibly invalid type name or missing support class Apr 05 12:22:05 otavio : What I do, is that i remove recipes-fsl, but what should be the best practices ? Thanks :p Apr 05 12:27:34 Chrys: you have a problem; they forked it all Apr 05 12:27:52 Chrys: you cannot mix meta-fsl-arm with this mess Apr 05 12:28:28 otavio : Mhhh Apr 05 12:28:30 Chrys: meta-phytec duplicates most of our recipes and God knows which changes they did Apr 05 12:29:05 otavio : I've see that they made an update on march and add gstreamer-imx plugins Apr 05 12:29:25 otavio : when I remove fsl-arm, i have the error that imx-codec isn't provides Apr 05 12:29:48 Chrys: I am very sorry but you need to push them. Apr 05 12:30:16 otavio : they didn't know how to lunch a mp4 video ahah Apr 05 12:30:25 Chrys: https://github.com/Freescale/meta-freescale-3rdparty/tree/master/conf/machine Apr 05 12:30:32 Chrys: those are all well supported Apr 05 12:31:15 otavio : do you know a way of doing h264 decoding by software then ? Apr 05 12:31:41 Chrys: using gst-libav Apr 05 12:31:46 Chrys: but will be slow Apr 05 12:32:01 otavio : i have gst-libav but no h264 decoder in it Apr 05 12:32:06 otavio : i can only run ogg video Apr 05 12:32:22 Chrys: you must enable it likely Apr 05 12:32:41 Chrys: do you have other imx board? Apr 05 12:32:52 if so, our demo images should be of help Apr 05 12:32:55 otavio : No. Do you know a good board for GUI + video ? Apr 05 12:33:18 Chrys: http://blink.ossystems.com.br/ Apr 05 12:33:31 Chrys: most imx6 are good Apr 05 12:33:44 we work a lot with most of vendors Apr 05 12:33:55 phytec is not one of them Apr 05 12:34:38 Chrys: tell us more about the project (pvt msg) and possibly we might help Apr 05 12:35:08 otavio : How do I do pvt msg ? Apr 05 12:37:24 hi everyone Apr 05 12:37:45 I have a strange behaviour with /etc/group file Apr 05 12:38:31 the avahi group is present in the image build directory, but not in the final image version... Apr 05 12:38:53 hi! i have a distro built on yocto. i need python 3.5 so i added the default python3 recipe. but it's very, very slow!! running "pip3 --help" takes 27 seconds on a raspberry pi! Apr 05 12:39:05 if anyone can help thanks Apr 05 12:39:09 python2 is about four times faster Apr 05 12:39:47 so i tried with the default raspberry distro ("raspbian"). same test, "pip3 --help", takes only 7 seconds, on the same hardware (same sd-card). Apr 05 12:40:01 is there any optimizations that i'm missing? Apr 05 12:54:25 Hi there! Apr 05 12:54:48 Anyone willing to help with this probably RSS-related issue? https://pastebin.com/PGumJfQG Some googling showed several pastebins with similar logs. Me too I was hit by something similar last january but RP wizardry helped there... Apr 05 12:55:54 I'm a newbie in Yocto who's trying to compile a custom image. I have added several recipies, but I have problems with two of them Apr 05 12:56:17 I have added my recipes under bblayer.conf Apr 05 12:56:36 This one is from meta-rauc layer (https://github.com/rauc/meta-rauc). Seems like this issues tend to pop up when recipes depend set dependencies on build artifacts (images). Apr 05 12:57:20 here the offending line seems: Apr 05 12:57:28 and I have added the library installations I want to make under local.conf, adding the as "IMAGE_INSTALL_append = " Apr 05 12:57:33 do_fetch[depends] = "${@' '.join([d.getVar(image, True) + ":do_build" for image in ['RAUC_SLOT_' + slot for slot in d.getVar('RAUC_BUNDLE_SLOTS', True).split()]])}" Apr 05 12:58:11 When I try to build using *bitbake core-minimal-image" breaks Apr 05 12:58:16 aurele: hi Apr 05 12:58:42 but if I compile each recipe separately does not break, typing *bitbake * Apr 05 12:58:57 maxin, hi Apr 05 12:59:00 So I'm confused about what I'm doing wrong Apr 05 12:59:55 aurele: trying to recreate the problem here.. are you installing avahi-daemon in the image ? Apr 05 13:00:01 the error printed is, among a huge list, "core-image-minimal-1.0-r0 do_rootfs: linux-xlnx not found in the base feeds" Apr 05 13:00:06 maxin, yes Apr 05 13:00:30 any clue about what could be the issue? Thanks in advance Apr 05 13:01:24 If there's any way I can provide more specifical info, let me know, I'll paste it in PasteBin or whatever :D Apr 05 13:01:24 aurele: ok, firing a build to see if that fails here too.. Apr 05 13:01:55 mugurumov: building the image without IMAGE_INSTALL_append in local.conf works? If yes, Can you paste the the IMAGE_INSTALL_append line Apr 05 13:02:31 jku: yes Apr 05 13:02:53 this is the local.conf Apr 05 13:02:54 IMAGE_INSTALL_append = " busybox device-tree hostapd rtl8812au init-ifupdown python u-boot-xlnx linux-xlnx" Apr 05 13:03:13 the one failing is the last one linux-xlnx Apr 05 13:03:24 I can compile it alone, bitbake linux-xlnx Apr 05 13:03:31 maxin, I'm following a lead : I have a recipe wich provides /etc/shadow and /etc/passwd files, I just fired avahi from passwd file (the group value was wrong in it). (for the record, those files are legacy files from older linux based targets) Apr 05 13:03:44 mugurumov: do you really want your kernel to be added into your rootfs? Apr 05 13:03:58 aurele: group file has avahi:x:999: in the build here.. Apr 05 13:04:04 mugurumov: Using CORE_IMAGE_EXTRA_INSTALL is a better choice to IMAGE_INSTALL Apr 05 13:04:26 behanw: huh? Apr 05 13:04:37 gizero: prefered, rt-patch Apr 05 13:04:48 behanw: I can try that Apr 05 13:04:59 aurele: ok, so, there was a custom recipe which provides passwd file ? Apr 05 13:05:17 same happens with another recipe I have that contains libraries like i2cdevtools,... Apr 05 13:05:17 maxin, yes Apr 05 13:05:32 LetoThe2nd: A way to add to IMAGE_INSTALL generically. Used to be called POKY_EXTRA_INSTALL before it was merged into meta Apr 05 13:05:35 so right now, I prefer having that working Apr 05 13:05:41 mugurumov: follow suggestion by jky. If commenting out IMAGE_INSTALL_append altogether works, try removing just two last recipes in the list Apr 05 13:05:46 so I can have libraries I need Apr 05 13:06:14 maxin, I will have to test from complete fresh build, but removing my bloody avahi line from passwd seems to fix the problem... sory for bothering Apr 05 13:06:52 aurele: good to know ... Apr 05 13:07:31 behanw: got a pointer to some rationale, or should i just try to memoize it that way? Apr 05 13:08:27 mugurumov: first, a recipe is not the same as package (you build the former, but IMAGE_INSTALL requires the latter) Apr 05 13:09:36 jku: ok, I still misuse some terms as I'm new to this Apr 05 13:10:38 mugurumov: second, not sure what you want to do with kernel, but adding a package to the image is not going to mean that kernel is then used. I think you need to use PREFERRED_PROVIDER_virtual/kernel instead but this is not something I'm familiar with Apr 05 13:12:28 jku: the kernek stuff is secondary right now, I have another package that contains libraries, and same happens Apr 05 13:12:58 so I will go for fixing this first. It contains stuff like i2cdevtools, wpasupplicant,... Apr 05 13:13:34 And, the same, if I compile it alone, bitbake compiles, but not if I add it to local.conf Apr 05 13:13:53 mugurumov: so with regards to recipes vs packages: i2cdevtools may provide a library package that has a different name. Try "bitbake -e i2cdevtools | grep ^PACKAGES=" Apr 05 13:13:59 mugurumov: if you are trying to build a base image with maybe some additional packages (i.e. libraries) but for your custom kernel (linux-xlnx ?) I'd suggest reading the "Yocto Project Board Support Package (BSP) Developer's Guide" Apr 05 13:14:41 LetoThe2nd: I believe that strictly speaking IMAGE_INSTALL is for image recipes and CORE_IMAGE_EXTRA_INSTALL is for conf files. Apr 05 13:15:16 in case it helps, the error printed in both cases is: "ERROR: core-image-minimal-1.0-r0 do_rootfs: linux-xlnx not found in the base feeds " Apr 05 13:15:21 The idea being that you can add arbitrary packages to your build without changing IMAGE_INSTALL directly. Apr 05 13:15:28 again... depending on the recipe, the package name you have to append IMAGE_INSTALL with may be different Apr 05 13:15:55 LetoThe2nd: I don't have a link handy to point to it however. Apr 05 13:16:25 gizero: +1 Good advice (about the dev guide) Apr 05 13:16:38 mugurumov: yes, it means no recipes ever provided a package with that name... even if "bitbake linux-xlnx" eventually succeded Apr 05 13:16:50 gizero: yeah, if I switch linux-xlnx to my libray package, the name in error changes Apr 05 13:16:58 ok gizero Apr 05 13:18:58 behanw: ah ok. thanks! Apr 05 13:24:17 gizero: are you using rm_work? Apr 05 13:24:43 RP: nope Apr 05 13:24:59 gizero: is there much in /home/gizero/work/smartliving/distro/repo-master/build-poky/tmp/sysroots-components/cortexa7hf-neon-vfpv4/glibc/ ? Apr 05 13:25:58 RP: 637 files! Apr 05 13:26:21 gizero: so the question is why some bits of glibc from there appear to have gone missing Apr 05 13:28:10 gizero: glibc should contain "rwhod.h" and it should be in that directory. Even rm_work wouldn't remove that so why its missing I really don't know/understand Apr 05 13:28:32 gizero: a manifest is referencing it, but that would seem to be correct Apr 05 13:29:05 missing? there's also that "FileExistsError" even before in the log... isn't that happening first? Apr 05 13:29:13 gizero: not sure what else I can say :/ Apr 05 13:29:42 gizero: maybe grep the build directory and see if you can find any references to anything removing it? Apr 05 13:29:59 gizero: I assume this was an incremental build and not a build from scratch? Apr 05 13:30:14 and that the file is still missing now with the build stopped? Apr 05 13:30:26 RP: maybe I can try to wipe tmp/ and see if it helps Apr 05 13:31:06 gizero: I have no doubt it probably would. Doesn't explain why it broke though :/ Apr 05 13:31:42 I'll try to break it again ;-) Apr 05 13:45:53 In a function, how do I access machine specific vars? e.g. MYVAR_mymachine="" is accessed in sh functions using ${MYVAR_${MACHINE}} ? Apr 05 13:51:06 hi. I want to keep layer-related additions in a (collection) subfolder (i.e. qt5-layer). But, afaics, that does not work with classes, right? Apr 05 13:52:20 is there a way to append bbclasses' path? Apr 05 13:53:42 or solely path convention? `$LAYER/classes`? Apr 05 13:53:57 RP: cleaning tmp/ doesn't help much if I then bake the final recipe again... same result (FileExistsError)s. I'm leaving now, but if I got it right, I'd better figure out which build step left that much stuff in /home/gizero/work/smartliving/distro/repo-master/build-poky/tmp/sysroots-components/cortexa7hf-neon-vfpv4/glibc/, correct? Apr 05 13:55:34 gizero: Hmm, I was looking at the FileNotFoundError part :/ Apr 05 13:56:21 gizero: two tasks failed, one with fileexists and the other filenotfound? Apr 05 13:56:51 gizero: have you "do_build" dependencies somewhere in your dependency chain? Apr 05 13:56:59 https://pastebin.com/PGumJfQG Apr 05 13:57:27 gizero: is that the whole log though? Apr 05 13:57:29 yes! I have this: do_fetch[depends] = "${@' '.join([d.getVar(image, True) + ":do_build" for image in ['RAUC_SLOT_' + slot for slot in d.getVar('RAUC_BUNDLE_SLOTS', True).split()]])}" Apr 05 13:58:12 gizero: I suspect if you swap do_build for an earlier dependency e.g. do_image_complete, it might work Apr 05 13:58:22 * RP really needs to make this work better Apr 05 13:58:40 I don't understand the filenotfound bit though Apr 05 13:58:53 RP: that was my suspect, but didn't really know which task to move it to... ;-) Apr 05 13:59:06 RP: I'll give it a spin asap Apr 05 13:59:16 ...I'll report back Apr 05 14:04:25 just for the record: The class only needs to be found *first* within the BBPATH. See poky/bitbake/lib/bblayers/query.py for details. Apr 05 14:11:43 I need to add ERROR_CFLAGS="" with make to make a package(gstreamer) pass compiling Apr 05 14:12:08 I export that variables in do_compile_prepend() {} it seems not work Apr 05 14:12:28 where should I add that variable Apr 05 14:14:36 EXTRA_OEMAKE seems match Apr 05 14:38:20 gizero: If you were able to check I'd love to know if this solves the issue btw: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip-rss2&id=2a7eed9118df1013502744398539c69a2c1c86ff Apr 05 15:03:36 kanavin: :( Apr 05 15:04:03 marquiz: if you need any assist on dnf stack stuff, let me know Apr 05 15:04:21 I did DNF stack integration for Mageia, so I have a decent idea of what you need to do to make things sane Apr 05 15:11:13 Pharaoh_Atem: thanks, I plan to get to it in the next few days Apr 05 15:11:48 does the build system (bitbake) ignore .bbappend that do not contain a matching COMPATIBLE_MACHINE ? Apr 05 15:12:45 T_UNIX: COMPATIBLE_MACHINE is used to skip a recipe Apr 05 15:12:52 bbappend is used to modify the metadata of a recipe Apr 05 15:12:57 or will those be appended anyway, eventually rendering the entire recipe unused Apr 05 15:13:19 they'll be appended regardless, then the recipe will be skipped when anonymous python functions are run at the end of parsing Apr 05 15:13:41 COMPATIBLE_MACHINE just results in raising an exception in an anonymous python function, you can see it implemented in base.bbclass Apr 05 15:13:43 ah, okay. Thanks kergoth :-) Apr 05 15:13:48 np Apr 05 15:18:34 the MACHINE value is derived from the file name in conf/machine/MACHINE.conf right? Apr 05 15:31:37 T_UNIX: backwards. the value of MACHINE is used to control what file is parsed in conf/machine. Apr 05 15:31:41 not the other way around Apr 05 15:52:07 If I do SOMEVAR="${@myfn(d)}" and I must ensure that SOMEVAR never is clobbering the taskhash. How do I do that? SOMEVAR[nodepsexclude] can't be set on a py function, can it? Apr 05 15:54:15 ...or vardepsexclude to be precise Apr 05 15:56:01 sveinse: it does work on py functions Apr 05 15:59:49 hello everyone Apr 05 16:00:18 i am using yocto project for beaglebone black Apr 05 16:00:47 i use eclipse IDE to develop application for beaglebone black Apr 05 16:01:40 i follow http://www.yoctoproject.org/docs/2.2.1/mega-manual/mega-manual.html#resources-mailinglist Apr 05 16:01:50 to setup eclipse Apr 05 16:02:35 now i want to develop GUI application using gtk+ Apr 05 16:03:05 i include this statement in source code file: #include Apr 05 16:03:11 but when i build project Apr 05 16:03:46 hello.c:33:21: fatal error: gtk/gtk.h: No such file or directory Apr 05 16:05:27 if you have a way to solve it. Please show me Apr 05 16:07:01 hello Apr 05 16:17:04 kergoth: I have an interesting dilemma. do_build of glibc-initial and the other -initial friends are causing quite some problems :/ Apr 05 16:18:00 kergoth: the issue is that in a "collapsed" inter-sstate task graph do_build is removed and you end up with xxx:do_image_complete -> glibc-initial:do_populate_sysroot :/ Apr 05 16:18:21 We can't even use the depvalid trick to unmask some of them since the starting task isn't an sstate one Apr 05 16:19:49 kergoth: I'm wondering about putting a deltask do_build into glibc-initial but that would need a bitbake change to allow recrdeptask to work Apr 05 16:23:14 could deltask it and then addtask it again, so it exists, just doesn't depend on anything? Apr 05 16:23:22 dunno, i'm not very awake today, i'll give it somet hought Apr 05 16:26:04 kergoth: I have patches and the dependency graph changes look sane... Apr 05 16:26:21 kergoth: keep thinking its a pretty major thing to do though... Apr 05 16:26:44 even the current index error if a recrdeptask doesn't exist could be classed as a real bug I guess Apr 05 16:28:14 kergoth: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip-rss2&id=24608feee0a4195db13c29ac95aadec4a2573d9a and http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip-rss2&id=23edd35eb6abc37f623c62d20b35e4ff53841497 seems too simple Apr 05 16:29:33 I guess I should test an eSDK with that Apr 05 16:30:06 lsandov: do you know whom to ask about oe-selftest coverage support? Looks like it's not fully working. Apr 05 16:31:25 ed2: can you define "not fully working"? Apr 05 16:33:03 lsandov: I'm adding test cases for not covered code, but it doesn't reflect coverage report for some reason Apr 05 16:33:30 please help me Apr 05 16:33:56 i am having a trouble with build gtk+ application in eclipse Apr 05 16:34:20 ed2: ah, that coverage, I understand now Apr 05 16:34:23 i used yocto plugin Apr 05 16:35:06 rocket42: perhaps if you pastebin the specific error you're getting, or describe the problem someone might know Apr 05 16:36:03 yes, i am doing Apr 05 16:36:48 I have created "Yocto Project SDK Autotools Project" project which is GUI application using gtk+ Apr 05 16:37:03 in my source code Apr 05 16:37:13 #include atal error: gtk/gtk.h: No such file or directory Apr 05 16:39:05 although when i add path of gtk.h to include directories ( project > properties > c/c++ general > paths and symbols). Apr 05 16:39:14 rocket42: did you point eclipse at a toolchain which has the gtk headers in it? Apr 05 16:39:27 yes, Apr 05 16:40:09 i pointed to toolchain by set up Yocto Project SDK (Window>preferences) Apr 05 16:40:42 And I follow part 10.3 in chapter 10 in http://www.yoctoproject.org/docs/2.2.1/mega-manual/mega-manual.html#sdk-developing-applications-using-eclipse Apr 05 16:41:25 rocket42: where did you get that toolchain from? I'd check that {OECORE_TARGET_SYSROOT}/usr/include/gtk+ exists Apr 05 16:41:47 rocket42: sadly I don't use the eclipse plugin so I'm not much help there Apr 05 16:42:05 i get toolchain from http://downloads.yoctoproject.org/releases/yocto/yocto-2.2.1/toolchain/x86_64/ Apr 05 16:42:45 rocket42: which one specifically did you choose? Apr 05 16:43:35 poky-glibc-x86_64-core-image-sato-armv5e-toolchain-2.2.1.sh Apr 05 16:44:21 i installed in folder: ~/poky/2.2.1 Apr 05 16:45:14 rocket42: that does sound like it should work... Apr 05 16:45:44 yes Apr 05 16:45:58 i follow official document from yocto project Apr 05 16:46:59 when i add manually path of gtk.h to including directories Apr 05 16:47:06 rocket42: ah, I think this might be because you need to add in the gtk+ compiler flags. Did you create any autoconf pieces to add that in? Apr 05 16:47:33 Eclipse can navigate to this file automatically Apr 05 16:47:52 but when rebuild, this error still Apr 05 16:48:51 rocket42: you usually have to add in something like `pkg-config gtk+-2.0 --cflags` Apr 05 16:48:56 (and ldflags) Apr 05 16:49:10 but this bit is gtk specific, not really the yocto project Apr 05 16:49:19 oh, Apr 05 16:50:14 i just checked Includes folder in Project Explorer Apr 05 16:50:50 that is /usr/include Apr 05 16:51:25 not ~/poky/2.2.1/usr/include Apr 05 16:51:39 do you know why? Apr 05 16:51:46 rocket42: --sysroot=XXX is probably on the compiler arguments? Apr 05 16:53:15 --sysroot=~/poky/2.2.1/ Apr 05 16:54:00 so i think that including directory must be "~/poky/2.2.1/usr/include" Apr 05 16:54:09 right, RP? Apr 05 16:54:23 rocket42: right Apr 05 16:54:52 rocket42: however gtk lives in /usr/include/gtk-2.0/gtk/gtk.h Apr 05 16:55:17 rocket42: the pkg-config command I gave above will add in -I /usr/include/gtk-2.0/ amongst other things Apr 05 16:56:15 but /usr/include/gtk-2.0/gtk/gtk.h for native compilation, not for cross-compilation Apr 05 16:57:20 i think gtk is used must be ~/poky/2.2.1/usr/include/gtk-2.0/gtk/gtk.h Apr 05 16:58:43 i am cross-develop gtk+ application from ARM architecture Apr 05 16:59:07 i am cross-developing gtk+ application for ARM architecture Apr 05 18:39:00 ah ha! Apr 05 18:40:09 when you run bitbake -e somerecipe without a populated cache, variable tracking is left on during the up front parsing, not just the final full parse we use to dump the metadata for -e. results in a parsing memory usage of 5gb instead of 2.5gb here (with/without var tracking). if i bitbake -p first, the bitbake -e uses 400mb regardless Apr 05 18:40:26 obviously the memory usage drops down after that initial parse, but in some cases that can be too late, if we got OOM killed :) Apr 05 18:41:15 guessing we want it enabled during the config metadata parsing, then disabled during the full recipe parsing to populate the recipe cache, then re-enable for the -e parse and dump Apr 05 18:59:29 kergoth: that sounds like a enhancement to be tracked in bugzilla Apr 05 18:59:57 kergoth: are you using mem_profiler as we discussed some weeks ago? Apr 05 19:07:49 lsandov: right now i'm just running bitbake with memory_profiler's mprof command line app, havent' ahd a chance to debug a problem with the patches to let bitbake do it itself yet Apr 05 21:08:26 if I move the SDK I need to adapt the path of the found binaries in the relocate script on my own, or have I missed some parameter? Apr 05 21:11:17 mdnneo: right, the SDK isn't designed to be moved after installation... the relocation script that is run on install gets deleted after it finishes successfully Apr 05 21:12:09 bluelightning: there is a parameter to keep it and it also creates an according .sh script ... but yes just the one which had been valid for the particular installation Apr 05 21:14:53 so seems like it is not ment for "another" move Apr 05 21:17:47 ok but then I think I either modify the kept one via search replace or use the lines of the installer script to call it accordingly ... just thought I ask first if there is a existing solution Apr 05 21:18:15 in theory it should be able to be used again, but I'm not sure it's recommended procedure Apr 05 21:19:45 bluelightning, I sent a patch for meta-qt4 a while ago Apr 05 21:25:09 Crofton|work: hmm yes I am behind on those, let me check if I already picked it up and just need to push Apr 05 21:25:13 (sorry) Apr 05 21:25:36 no worries Apr 05 21:25:42 nothing changes in the layer Apr 05 21:25:44 so not stale Apr 05 21:26:08 python-pyqt: Update to version 4.12. Apr 05 22:06:58 kergoth: sounds like a good lead Apr 05 22:11:43 damn it, now I can't find where I even started merging meta-qt4 patches, I guess I'll just have to do it again Apr 05 22:30:16 hi, i'm trying to upload a custom image to AWS and it is complaining about my kernel version. Apr 05 22:30:52 someone have ever upload a yocto image to the aws cloud system,? Apr 05 23:42:26 Crofton|work: so I just did a build test and discovered something I'd not realised - sip is duplicated between meta-oe and meta-qt4 Apr 05 23:42:47 Crofton|work: the one in meta-qt4 hasn't been updated and thus building python-pyqt fails without meta-oe - should we be doing something about that? Apr 05 23:44:33 I don't see anything in the entire meta-openembedded repo that refers to sip, so does it need to be there? Apr 06 01:46:40 bluelightning, did sip also end up in core Apr 06 01:47:38 Crofton|work: nope: http://layers.openembedded.org/layerindex/branch/master/recipes/?q=sip Apr 06 01:47:48 (now that our search actually works well for this kind of thing...) Apr 06 01:48:25 I think it was update sin meta-oe for something else Apr 06 01:49:13 oh, right... now I see the commit message Apr 06 01:49:16 sigh Apr 06 01:50:06 ah yes, need for qt5 Apr 06 01:50:38 reminds me, nede to work on pywt5 Apr 06 01:51:15 pyqt5 :) Apr 06 01:51:28 so at the moment I can't merge this patch because python-pyqt will stop building without meta-oe and meta-oe isn't a dependency Apr 06 01:51:53 pyqt won't build with newer sip Apr 06 01:52:11 no, and with the patch it won't build without it ;) Apr 06 01:52:44 short term the fix would be to copy the new sip version over to meta-qt4 Apr 06 01:52:48 well, I think I am the only user of pyqt Apr 06 01:53:13 we really do not want to dupe recipes Apr 06 01:53:23 not really, no, but we already have it Apr 06 01:53:32 either meta-qt4 and meta-qt5 depnd on meta-oe Apr 06 01:53:37 alternatively we state that python-pyqt won't build without meta-oe Apr 06 01:53:39 or sip goes to core Apr 06 01:53:56 I don't think we can add a layer dependency on meta-oe just for this Apr 06 01:54:43 so you break pyqt for anyone using meta-oe? Apr 06 01:54:48 nope Apr 06 01:54:59 that's not what I am suggesting Apr 06 01:55:25 you want to carry a dupe recipe then? Apr 06 01:55:37 no, not that either Apr 06 01:55:48 I'm saying, we don't add the layer dependency explicitly, but we force you to add meta-oe if you want to build pyqt Apr 06 01:55:56 we can document that in the readme Apr 06 01:56:06 ok fine with me :) Apr 06 01:56:18 I just need it to work Apr 06 01:56:23 and it is late here :) Apr 06 01:56:33 and I really need to shift to qt5 Apr 06 01:56:37 ok, I will merge your patch and send a patch to remove sip Apr 06 01:57:15 that seems like the best approach Apr 06 01:57:23 I have had people ask about pyqt5 Apr 06 01:57:32 because you find an email from me asking about it Apr 06 01:57:49 Thanks for getting caught up Apr 06 01:57:53 I'm off to bed Apr 06 01:58:19 heh I also have to add an EXCLUDE_FROM_WORLD to poky's inc file as well Apr 06 01:58:22 fun times Apr 06 01:59:00 goodnight :) Apr 06 02:48:57 hello, everyone Apr 06 02:49:26 i am having problem with cross-develop GUI application for beaglebone black Apr 06 02:49:51 tootchain i used is to get from yocto project Apr 06 02:50:24 image for beagelebone black also get from yocto project (santo image) Apr 06 02:51:04 i also use eclipse yocto plug-in to cross-develop Apr 06 02:51:25 i use gtk+ library for my application Apr 06 02:51:34 but when i build my application Apr 06 02:52:26 fatal error is always appear "fatal error: : no such file or directory" Apr 06 02:52:48 i don't know how to solve it Apr 06 02:55:30 it is very strange that **** ENDING LOGGING AT Thu Apr 06 03:00:02 2017