**** BEGIN LOGGING AT Fri Jan 13 03:00:02 2017 Jan 13 07:01:57 nrossi: Hello, yesterday I was able to compile "bitbake qtwebengine" but somehow I didn't find any of the examples got build with it. Jan 13 07:29:20 Ramose: qtwebengine-examples package? Jan 13 07:29:55 Ramose: I remember last time I used qtwebengine i don't think the examples were enabled by default. But this was qt4 or something Jan 13 07:30:27 nrossi: I tried building it bitbake qtwebengine-examples Jan 13 07:30:37 but didn't get anything Jan 13 07:30:42 let me try again Jan 13 07:31:11 Ramose: you cant build packages like that, only recipes qtwebengine should generate the qtwebengine-examples package if they are enabled Jan 13 07:31:37 nrossi: But then how should I enbale it Jan 13 07:32:24 I place it in meta-qt5-mel/recipes-core/packagegroups/packagegroup-qt5.bb Jan 13 07:32:44 nrossi; under RDEPENDS_${PN} += Jan 13 07:35:55 Ramose: Not sure tbh, the qtwebkit recipe uses the PACKAGECONFIG 'examples' Jan 13 07:38:38 nrossi: Am I missing something like meta-qt5-mel/recipes/qt5/qtwebkit_%.bbappend ? Jan 13 07:38:50 I mean I should same for qtwebengine ? Jan 13 07:39:46 Ramose: looks like all you need to do is add 'examples' to PACKAGECONFIG. https://github.com/meta-qt5/meta-qt5/blob/c0ba8ab590ba4e17a2a04925cf48915233d4f493/recipes-qt/qt5/qtbase_git.bb#L71 Jan 13 07:40:55 nrossi; Let me try Jan 13 07:41:00 Ramose: sorry wrong link -> https://github.com/meta-qt5/meta-qt5/blob/9aa870eecf6dc7a87678393bd55b97e21033ab48/recipes-qt/qt5/qt5.inc Jan 13 07:42:10 nrossi: I am confused now :( Jan 13 07:43:01 Ramose: when you add 'examples' to PACKAGECONFIG of qtwebengine, it will add "QT_BUILD_PARTS+=examples" to the configure, which should trigger the build of examples Jan 13 07:43:28 Ramose: adding PACKAGECONFIG_pn-qtwebengine_append = " examples" in your local.conf should do the trick Jan 13 07:44:36 Ramose: sorry, wrong way around, PACKAGECONFIG_append_pn-qtwebengine* Jan 13 07:48:54 nrossi: ok I placed PACKAGECONFIG_append_pn-qtwebengine = " examples" in build/conf/local.conf Jan 13 07:49:07 lets see now Jan 13 07:56:02 nrossi: it didn't work , This is where I have my packages included https://paste.ubuntu.com/23791113/ Jan 13 07:56:55 Ramose: did it error? or just no files? Jan 13 07:57:54 nrossi: bitbake qtwebengine run well but not files in tmp/work/coretex..... Jan 13 07:58:08 let me try it again Jan 13 07:59:00 Ramose: check the log.do_compile for qtwebengine, grep for "examples" see if it is even trying to build them Jan 13 07:59:10 ok Jan 13 08:04:11 nrossi: No , there is no string "examples" in log.do_compile of qtwebengine Jan 13 08:08:39 nrossi: this line PACKAGES =. "${PN}-qmlplugins-dbg ${PN}-tools-dbg ${PN}-plugins-dbg ${PN}-qmldesigner ${PN}-qmlplugins ${PN}-tools ${PN}-plugins ${PN}-mkspecs ${PN}-examples-dev ${PN}-examples-staticdev ${PN}-examples-dbg ${PN}-examples " already there in qt5.inc Jan 13 08:12:28 nrossi: but I do see this ./log.do_configure.32590:NOTE: QA Issue: qtwebengine: invalid PACKAGECONFIG: examples [invalid-packageconfig] when I grep -r "examples" . in qtwebengine/5.5.0+gitAUTOINC+qtwebengine-r0/temp folder Jan 13 08:15:56 Ramose: Hmmm thats odd, run bitbake qtwebengine -e and pipe to a file, then search for PACKAGECONFIG and look at the flags for [examples] Jan 13 08:16:50 ok Jan 13 08:47:11 How come my central sstate cache and my locally built package disagree on the package's files, yet with the same hash ID? Jan 13 08:47:59 I have to investigate this because ultimately it challenges our trust in the sstate cache for production environments :( Jan 13 08:49:23 sveinse: non-deterministic builds Jan 13 08:49:34 easily solved Jan 13 08:50:00 look at what files are different, then look at the build which controls those files, identify what codepaths are automatically detected Jan 13 08:50:18 like the configure.ac might see if libfoo is present and enable building something Jan 13 08:50:46 then the presence of libfoo in the sysroot will change the output, but if it isn't listed as a build dep then bitbake has no way of knowing that changes the build Jan 13 08:51:03 RPs recipe specific sysroots work makes this disappear, btw Jan 13 08:51:28 rburton: The package is uim, and it does use configure. So it can be a change in the sysroot which would pass undetected into the stateinfo? Jan 13 08:51:54 yes Jan 13 08:52:21 its fairly common for configure.ac to do something like AC_CHECK_LIBS to see if a library is present and then enable/disable building more code automatically Jan 13 08:52:44 there's a script in oe-core to exercise the builds to detect this Jan 13 08:53:16 it basically does a world build, rebuilds everthing with a full sysroot, then rebuilds everything with an empty sysroot. any changes are bugs. Jan 13 08:53:24 rburton: I'm curious, why isn't the hashsum of the sstate artefact stored? With it you could at least validize its contents as a part of siginfo Jan 13 08:54:15 We've had problems with uim before. We had to disable parallel installs, because it fails occationally on massive parallel builds Jan 13 08:57:52 Is there a possibility to enter the SeaBIOS setup when running it under qemu? I can attach to the console with xvncviewer, but by the time I do that, it already started booting. Jan 13 08:58:40 So as a QA initiative, we could do a non-sstate cache build and compare the local sstate cache with the central sstate cache? Jan 13 09:00:20 sveinse: bitbake myimage-with-has-uim-in ; bitbake uim ; wipe-sysroot ; bitbake uim -C unpack; buildhistory-diff Jan 13 09:00:54 rburton: the problem is that I don't have any information about the package stored in the central sstate cache. I don't have its build logs Jan 13 09:01:31 ignore that, you can replicate the problem locally Jan 13 09:05:23 rburton: yup, I'll try that. I think I see where you're going with this: That uim gets some leakage from the other recipes from the same build Jan 13 09:06:09 yes Jan 13 09:06:19 the sysroot is full of stuff that uim can decide to use Jan 13 09:07:09 If this is the case, then RPs fix is the only way round. Or is there a way to instruct yocto to wipe the sysroot cache from the recipe? Jan 13 09:09:23 RPs improvement is the one true fix Jan 13 09:09:33 currently, its a bug if the recipe doesn't control the build Jan 13 09:09:51 you'll notice that oe-core has several recipes where we add enable/disable toggled to stop this behaviour Jan 13 09:32:50 rburton, can you name one recipe where this is used? Jan 13 09:33:51 sveinse: we do store information about the hash in the siginfo/sigdata files, those tell us how its constructed. We checksum the inputs (the configuration and variables that were used to configure the build), not the output. Jan 13 09:37:07 RP, I understand that. And of course, it would not matter if you're doing a build and you fetch artefacts from the sstate cache. But with it I could compare two sstate caches. ..hmm, which I can anyways just by comparing files, because both would be correct. Yeah, forget it Jan 13 09:38:47 Is RPs fix a simple fix that can be cherry picked into an older poky (jethro)? Jan 13 09:41:47 sveinse: its not even in master yet, still a work in progress Jan 13 09:41:58 sveinse: nobody would describe it as simple Jan 13 09:42:09 RP: ok, thanks Jan 13 10:07:37 nrossi: I set it up again and run bitbake qtwebengine and see this ERROR: QA Issue: Package version for package qtwebengine-examples went backwards which would break package feeds from (0:5.5.0+git0+qtwebengine-r0.2 to 0:5.5.0+git0+qtwebengine-r0.1) [version-going-backwards] Jan 13 10:12:18 Ramose: not really sure what you have done, but looks like something is not cleaned properly? Jan 13 10:14:56 nrossi: yeah, after cleansstate I don't see this Jan 13 10:15:48 nrossi: there is no qtwebengine-examples folder yet Jan 13 10:17:51 nrossi: Actually I needed to run a QT based webrowser on target , for same I tried fancybrowser that comes with qtwebkit-examples but some how not able to play any video into it Jan 13 10:18:03 like youtube or .ogg files Jan 13 10:18:29 so I thought of trying fancybrowser that comes with qtwebengines Jan 13 10:21:27 Ramose: does the gstreamer you build (which is used by qtwebkit) support the target codecs you need? Jan 13 10:22:20 nrossi: sorry but how do I check it ? Jan 13 10:24:05 Ramose: gstreamer-inspect to check available codecs, and mediainfo to check your .ogg format. Youtube has a "detailed info" on the right click that shows the codec being used Jan 13 10:25:14 nrossi: something this https://www.youtube.com/html5 ? Jan 13 10:25:39 Ramose: that works too Jan 13 10:26:16 nrossi: let me check on target fancybrowser Jan 13 10:30:31 nrossi: I am seeing this on target when runs youtube ** (fancybrowser:663): WARNING **: add element stream Jan 13 10:32:05 Ramose: ok now you are probably entering webkit bug territory. Jan 13 10:33:02 yeah , kind of Jan 13 10:33:32 Ramose: if i remember correctly, qtwebkit is deprecated since 5.5. So yer... :), maybe back to qtwebengine? Jan 13 10:34:15 but there kind of not able to get any examples to run Jan 13 10:34:51 codec on target looks to be avc1.420001E Jan 13 10:37:42 Ramose: so you were able to build the examples? Jan 13 10:38:41 nross: No, only qtwebkit-examples Jan 13 10:38:46 I am able to buikd Jan 13 10:38:52 *build Jan 13 10:39:11 but qtwebengine-examples not building Jan 13 10:43:02 nrossi: I can see qtwebkit-examples has its own recipe file Jan 13 10:43:13 like qtwebkit-examples.bb Jan 13 11:14:47 Hello everyone Jan 13 11:16:57 I'm trying to make a basic yocto image for qemu using toaster, but when I add a new recipe I got "DEBUG runbuilds: No build env" on the toaster log. I've set the enviroment with the "source oe-init-build-env" command Jan 13 11:17:13 any ideas why I get this? Jan 13 12:20:12 are you running toaster from the same enviroment you sourced the buildenv? Jan 13 12:28:11 yes Jan 13 12:28:50 "source oe-init-build-env" and then "source toaster start" Jan 13 12:50:40 nrossi: its just strange I just couldn't find any *.o file from inside build/tmp/work/cortexa9hf-vfp-neon-nt-linux-gnueabi/qtwebengine/5.5.0+gitAUTOINC+qtwebengine-r0 Jan 13 13:11:29 python 2.8, pleas no Jan 13 13:12:22 "The appearance of a "Python 2.8" got the attention of the Python core developers in early December. It is based on Python 2.7, with features backported from Python 3.x. In general, there was little support for the effort—core developers tend to clearly see Python 3 as the way forward—but no opposition to it either. The Python license makes it clear that these kinds of efforts are legal and even encouraged—any real opposition to the project lies in Jan 13 13:12:55 can't get to the lwn article, anyone has a direct link? https://lwn.net/Articles/711061/ Jan 13 13:14:06 kanavin: https://lwn.net/SubscriberLink/711061/ed5f5a5caf7563c4/ Jan 13 13:14:18 neverpanic: cheers Jan 13 13:14:37 tl;dr: 1man project, will get different name, there's not going to be Python 2.8 Jan 13 13:16:27 neverpanic: except there's a ton of people and companies who are complacent about moving to python 3, and may coalesce around this Jan 13 13:16:51 (Yocto is not one of them) Jan 13 13:26:51 we've taken the python3 pain on the most part now apart from smart.. Jan 13 13:30:32 RP: dnf stack has a configure switch --with-python=3, which I'm not yet using :) Jan 13 13:33:19 kanavin: right, I know you're working on that :) Jan 13 13:48:08 hmm, how to detect in a recipe when -b is being used :/ Jan 13 13:57:18 Hello, I've built a image recipe and on genericx86 some packages are not being included in the rootfs. What are some of the obvious things I should look for? Jan 13 13:57:54 Strike5150: something in the image you're building would need to have a dependency on them Jan 13 13:59:06 RP: So the package is included using IMAGE_INSTALL_append = " xxxx" Jan 13 13:59:50 RP: Normally that should suffice? I'm noticeing that the package is being built under core2 when machine type is set to genericx86 Jan 13 13:59:53 Strike5150: perhaps check IMAGE_INSTALL has the value you think it has with bitbake -e ? Jan 13 14:01:48 RP: IMAGE_INSTALL lists the package as being included, its also part of the rootfs.manifest Jan 13 14:03:06 Strike5150: in which case it should be part of the image. Does the package contain files or is it empty? Jan 13 14:05:36 RP: The package has a size so assume its got the content intended Jan 13 14:06:08 RP: The strange thing though is the package is under core2-32-poky-linux/ not built under genericx86-poky-linux/ Jan 13 14:06:20 RP: I don't know if this is normal Jan 13 14:06:39 RP: by size I mean roughly what I would expect normally Jan 13 14:06:42 Strike5150: that is normal if the package isn't machine specfiic which most aren;t Jan 13 14:07:47 RP: I could post the gist of image environment if that helps? Jan 13 14:08:16 ll Jan 13 14:09:42 Strike5150: Does the package manager within image think the package is installed? Do the rootfs task logs show the package being installed? Jan 13 14:10:31 RP: I don't know where to find the package manager logs Jan 13 14:10:43 RP: I'll check the task logs Jan 13 14:11:00 Strike5150: which package backend are you using? Jan 13 14:11:23 rp: debian Jan 13 14:11:30 rp: deb :D Jan 13 14:11:58 Strike5150: so /var/lib/dpkg/status in the rootfs or you could query with apt/dpkg Jan 13 14:18:17 RP: ok i'll do that now, sorry was distracted by co-workers :| Jan 13 14:20:14 RP: According to /var/lib/dpkg/status the package was installed and status is ok Jan 13 14:21:24 Strike5150: ok, you really need to look inside the .deb and check it has what you think it has in it then Jan 13 14:21:26 RP: But in rootfs log it clearly shows its only installing package-dbg not package Jan 13 14:21:54 Strike5150: the status file is definitive, if it says its installed, I'd suspect it was Jan 13 14:21:56 RP: I know there are no files defined for the dbg package yet Jan 13 14:22:41 RP: I will also double check I got the right one Jan 13 14:23:47 RP: The image/rootfs/var/lib/dpkg/status date the file was touched is from yesterday Jan 13 14:24:00 RP: Not from todays build Jan 13 14:25:37 Strike5150: it would only rebuild if it thought something had changed and it needed to? Jan 13 14:27:27 RP: I've changed the contents yes, also the package definitely has the files in it Jan 13 14:30:21 RP: So just to verify, the image rootfs is in genericx86(machine type) and in there the rootfs/var/lib/dpkg/status says packageA is installed, but looking at temp/log.do_rootfs shows its only installing packageA-dbg which has no files in it Jan 13 14:30:49 Strike5150: are you sure there isn't a mention of packageA somewhere else in that log? Jan 13 14:30:59 RP: I will double check Jan 13 14:31:44 RP: Ok your are right, there is mention of packageA and packageA-dbg Jan 13 14:32:09 RP: I am using wic to make a direct image, is it possible I've made a mistake in that step? Jan 13 14:32:19 RP: Could be the image is fine Jan 13 14:33:10 Strike5150: its possible. somewhere along the line something is going wrong and that could be it Jan 13 14:34:22 RP: I've just dpkg --get-selections on the running image and it thinks the package is installed too but the files are missing, I need to think a bit :D. I will yell when I have something more sensible to say Jan 13 14:35:03 RP: Thanks for the help so far o/ Jan 13 14:36:50 Strike5150: np, something isn't adding up... Jan 13 14:59:40 Hm, I get InvalidGitRepositoryError when I'm trying to use buildhistory-diff. What am I missing? Jan 13 15:03:47 pohly: i see 1/2 rm_work: allow preserving additional context, but no 2/2 Jan 13 15:04:09 sveinse: check whether BUILDHISTORY_COMMIT = "1" ? Jan 13 15:04:21 I have confirmed that uim produces different result when being run in an image vs. when being run after wipe-sysroot; bitbake uim -C unpack. What should I do next? Jan 13 15:04:39 rburton: the other commit was for meta-swupd - I should have formatted the patch differently. Jan 13 15:05:01 sveinse: read the uim makefiles, configure.ac etc and figure out how to fix it. you may need to pass an option, or patch the configure.ac, or something else. Jan 13 15:05:25 sveinse: or, just add the dependency that causes the change to DEPENDS so its always present Jan 13 15:05:38 pohly: ok, thanks Jan 13 15:06:11 rburton: you mentioned a trick for configure that was applied on other oe recipes like this? Jan 13 15:07:18 rburton: about the earlier rm_work enhancements: you currently have those in your mut branch, but the implementation based on prioritized anonymous functions needs to be replaced. V2 just sent, but it depends on RP's event patch, which hasn't been finalized yet. Jan 13 15:08:00 sveinse: if you want to be able to control the dependency then you need to use AC_ARG_ENABLE in configure.ac to make it deterministic, but if you just want to have it on always then just add the dependencies that you find are not listed Jan 13 15:08:42 rburton: right, thanks. I'm investigating now Jan 13 15:09:14 pohly: yeah just noticed the v2, ripped out of the branch now Jan 13 15:43:36 mario-goulart: hi Jan 13 16:39:52 is gh clone working today ? Jan 13 16:40:12 git clone git://github.com/torvalds/linux ended with fatal: remote error Jan 13 16:40:45 khem: https://status.github.com/messages major service outage? Jan 13 16:41:49 08:14 PSTMajor service outage. Jan 13 16:42:01 thats like 1/2 hr back Jan 13 16:44:07 seems to ahve come back Jan 13 16:44:44 yeah Jan 13 16:45:19 if github disappears what will we do ? Jan 13 16:45:31 it will be like stone age Jan 13 16:45:52 well, we just find somewhere else to push our code Jan 13 16:51:41 khem: gitlab and bitbucket are both superior to github Jan 13 16:52:21 kanavin: why are they superior? :) Jan 13 16:52:48 aratiu: because I tried them both and liked them better than github :) Jan 13 16:53:30 ymmv Jan 13 16:54:04 gitlab is good as long as its not on my infra Jan 13 16:54:06 its a pig Jan 13 16:54:26 gitolite is the best Jan 13 16:55:17 lol Jan 13 17:33:04 OK locked python3 to 3.5 on arch for now Jan 13 17:36:39 ERROR: mtd-utils-2.0.0-r0 do_install: Function failed: do_install (log file is located at /home/balister/src/oe-core/build/tmp-glibc/work/cortexa9hf-neon-oe-linux-gnueabi/mtd-utils/2.0.0-r0/temp/log.do_install.14252) Jan 13 17:36:43 jsut blew up on me Jan 13 17:37:01 | make[1]: Leaving directory '/home/balister/src/oe-core/build/tmp-glibc/work/cortexa9hf-neon-oe-linux-gnueabi/mtd-utils/2.0.0-r0/build' Jan 13 17:37:01 | install: cannot stat 'ubi-utils/include/libubi.h': No such file or directory Jan 13 17:47:46 I see there will be a Yocto Developer Day North America 2017 on February 24th, an addon to the ELC/IOT NA conference. http://events.linuxfoundation.org/events/embedded-linux-conference/extend-the-experience/co-located-events - is this all the published information there is on it so far? Kind of curious who all will be there. Jan 13 17:49:01 Dev Day is more a training thing Jan 13 17:49:10 not sue if the schedule is up Jan 13 17:49:21 We also have http://www.openembedded.org/wiki/OEDAM_2017 on Monday Jan 13 17:50:10 this is the schedule from the last one https://www.yoctoproject.org/yocto-project-developer-day-us-2016-elc Jan 13 17:50:27 It's promised that there will be an Introductory Track (tracking) as well as the advanced track (which I'm interested in), which is partly training, demo, and reveal of features to come.... that's what interests me. Jan 13 17:50:35 https://www.yoctoproject.org/yocto-project-developer-day-europe-2016 Jan 13 17:50:47 ther are the actual schedules from the last two Jan 13 17:50:49 I did attend the 2017 NA one in San Diego. It was good. Jan 13 17:50:59 I would expect something similar Jan 13 17:51:04 yup Jan 13 17:51:06 unelss someone wants to correct me :) Jan 13 17:51:10 nope Jan 13 17:52:54 Agenda for Advanced Track (from http://events.linuxfoundation.org/events/embedded-linux-conference/extend-the-experience/co-located-events) Jan 13 17:52:56 - New Workflows in the Latest Release of the Yocto Project Jan 13 17:52:58 - Deep Dive and Class Exercises on Devtool and ESDK Jan 13 17:53:00 - WIC Jan 13 17:53:02 - Brief Tour of Meta-Security Jan 13 17:53:04 - Deep Dive Presentation and Demo on CROPS Jan 13 17:53:06 - BitBake Event Model - The Data Source for Toaster Jan 13 17:53:08 - New set-up.py Script Jan 13 17:53:10 - Kernel Roundup Jan 13 17:53:12 - User Space Jan 13 17:53:14 - Build History and Bitbake-diffsigs Jan 13 17:53:16 - Q&A Jan 13 17:53:26 rburton: the mystery of shell not finding things in PATH is solved :) Jan 13 17:54:16 rburton: rpm covertly chroots into image's rootfs Jan 13 17:55:43 gah! Jan 13 18:07:52 Crofton|work, thanks for the OE Dev meeting heads-up... I will see if I can arrange to attend. Jan 13 18:19:25 jamesp, I spammed the lists with an email Jan 13 18:19:42 and if you have dieas for the agenda, add them to the wiki Jan 13 18:19:48 we'll triage closer to the meeting Jan 13 18:20:01 I need to link the minutes from the last one to that page Jan 13 18:27:20 if I wanted to build selinux with gdb symbols so I could debug its usage in a devshell, how would I do that? Jan 13 18:27:43 im familiar with debug symbols on my target, but not debug symbols on my build system Jan 13 20:07:56 davis: are you looking for cross debugging ? Jan 13 20:08:19 otherwise just add selinux-dbg to IMAGE_INSTALL Jan 13 20:08:38 or use online package management to install it if you have configured it on target Jan 13 20:13:01 khem, no i am trying to debug why i have a failing bui8ld in one file system but not in another with identical source Jan 13 20:16:25 i always appreciate your help, so any advice is much appreciated Jan 13 23:44:31 is the autobuilder not publishing nightly images right now? Jan 13 23:45:59 I'm seeing that core-image-sato-sdk intel-core* wic images increased by ~1GB recently, and wanted to confirm whether its a meta-intel or generic change that caused that Jan 13 23:47:14 i don't make it publish mut builds at least Jan 13 23:47:24 id be shocked if that were a meta-intel issue Jan 13 23:47:28 when is "recently"? Jan 13 23:47:58 in the last month it looks like Jan 13 23:48:04 also, worth comparing the size of the tar.gz in case its the wic image and not the contents which is changing Jan 13 23:48:15 theres a wiki page with size metrics in that i can never find Jan 13 23:49:04 got to go to bed now, throw me a mail if you can't find anything :) Jan 13 23:55:21 clsulliv, Do you need a build with published artifacts? I could start one. Jan 14 00:00:11 halstead: No, its fine. I can start a local build before I leave today if need be. **** ENDING LOGGING AT Sat Jan 14 03:00:00 2017