**** BEGIN LOGGING AT Thu Jan 21 02:59:59 2016 Jan 21 03:15:01 that's interesting... I don't understand how in using 'journalctl' to capture outputs of a service I added, it is also capturing output from pppd Jan 21 09:10:27 good morning Jan 21 10:31:55 hello, I have some issue with jethro, and RPM4, is rpm4 is experimental, or somebody work with it ? Jan 21 10:33:27 yet, I saw 2 issue: The first is I need to add "probfilter |= rpm.RPMPROB_FILTER_IGNOREOS" in smart-rpm4-fixes.patch because, do_rootfs allways failed... Jan 21 10:34:28 The second, It seem RRECOMMENDS generate Suggests in the spec, but do_rootfs don't install suggests packages... Jan 21 10:52:17 I wrote 2 bug on filezilla: Bug 8968 and 8969 (I found a fixe for 8968, but I don't solve 8969 yet...) Jan 21 10:52:19 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=8968 normal, Undecided, ---, paul.eggleton, NEW , Issue with RPM4 and do_rootfs Jan 21 10:52:59 what's a good way to suppress variable expansion? i want to test for a string like "${foo}" in a QA test, but bitbake keeps expanding it. Jan 21 10:53:44 could hack around it with stuff like "${foo""}", but will e.g. \${foo} work? Jan 21 11:00:02 Hi. Is there a macro for sysroots location? Jan 21 11:00:35 Hi. Is there an environment variable for sysroots location? Jan 21 11:02:14 rt: search for STAGING_DIR in the reference manual Jan 21 11:02:32 STAGING_DIR* are defined in bitbake.conf Jan 21 11:13:12 @Ulfalizer How do I print an env variable in the shell Jan 21 11:13:49 ? Jan 21 11:13:54 wath is the best supported, recommended package format for Yocto ? RPM5, RPM4, DEB or IPKG ??? Jan 21 11:18:38 rpm5 or ipkg Jan 21 12:02:39 Is there a way to test all env variables using bitbake -c devshell? Jan 21 12:02:48 echo ${CC} is empty in my case Jan 21 12:03:31 Sorry echo ${CC} is fine but ${STAGING_DIR} is empty Jan 21 12:12:59 STAGING_DIR isn't exported by default Jan 21 12:17:33 rt: if you just want to check the value, do something like bitbake -e | grep ^STAGING_DIR Jan 21 12:17:50 bitbake -e | grep ^STAGING_DIR that is Jan 21 12:22:47 has anybody seen errors about .debug files while packaging gstreamer builds? Jan 21 12:23:05 I updated meta-gstreamer1.0 to gstreamer 1.6.2 , which is where I saw it. but that was in jethro. Jan 21 12:23:25 so I had to do this: https://github.com/dv1/meta-gstreamer1.0/commit/1a27f8467106da05da08a620633cc55ac3aa3789 Jan 21 12:23:28 dv_: not with gstreamer, but with another package that uses plugins like g streamer yes Jan 21 12:24:00 for some reason, oe-core master does not seem to have to add these FILES_ dbg lines Jan 21 12:24:09 dv_: because oe-core master is AWESOME and doesn't need them Jan 21 12:24:47 "... but superman does not need this. why?" - "because he is SUPERMAN, duh." :) Jan 21 12:24:48 oe-core da5ec06814e105451cca11cce76b5c5231110524 Jan 21 12:25:27 oh wow Jan 21 12:25:30 \o/ Jan 21 12:26:31 so don't backport master recipes to jethro without adding back the FILES lines Jan 21 12:41:07 Hi guys Jan 21 12:41:40 Is there a way to do an include or require conditionally in a bitbake recipe? Jan 21 12:42:26 I would like to include a ".inc" file based on a MACHINE_FEATRUE present or not Jan 21 12:43:08 set a variable that is either "" or the inc you want to include, then include the variabe Jan 21 12:51:56 @rburton: I tried: include ${@bb.utils.contains('MACHINE_FEATURES', 'dfu', 'dfu.inc', '', d)} and it works if dfu is in the features but it fails if not. Jan 21 12:52:16 hm, thought that would work Jan 21 12:52:21 maybe its only inherit that can do that Jan 21 12:52:41 This is the error I get: ERROR: ParseError in /home/projects/oe/build/../dfu/: not a BitBake file Jan 21 12:55:05 I got the following working: include ${@bb.utils.contains('MACHINE_FEATURES', 'dfu2', 'dfu.inc', 'no-dfu.inc', d)} where the no-dfu.inc is non existent Jan 21 12:57:30 @rburton: What I would really like is a conditional require. Where I get an error if the feature is definde but the file isn't found. But no error and nothing gets incuded if the feature is not defined. Jan 21 12:58:08 actually create no-dfu.inc but just put a comment in it, and use require Jan 21 12:58:35 or submit a patch so that bitbake's implementation of require/include can handle not having an argument Jan 21 13:00:15 bachp: what's so special in that file that it cannot have the conditionals inside of that .inc file? Jan 21 13:01:13 in other words, why not use single dfu.inc which adds or doesn't add dfu support based on MACHINE_FEATURES Jan 21 13:14:47 is anyone aware of a layer that has a recipe for evemu? Jan 21 13:21:58 CTtpollard: the layer index shows this http://layers.openembedded.org/layerindex/recipe/20519/ but I am not sure if that's what you are looking for … Jan 21 13:22:59 belen: http://people.freedesktop.org/~whot/evemu/ Jan 21 13:23:23 belen: looks like what I need :) Jan 21 13:23:38 so it's in meta-oe, great Jan 21 13:25:54 CTtpollard: the layer index kind of helps with these questions :) http://layers.openembedded.org/ Jan 21 13:26:11 belen: great stuff Jan 21 13:49:21 Hi. Did anyone ever use the openembedded-core/meta/recipes-kernel/linux/linux-dtb to build a set of dtbs predefined in the kernel-source/arch/${ARCH}/boot/dts/Makefile under dtb-$(CONFIG_BOARD)? I am trying to build the whole set of dtbs, but "oe_runmake dtbs" doesn't make the dtbs. However calling "oe_runmake specific_board_name.dtb" will create the dtb for specific_board.dtb. Jan 21 14:17:46 How to add meta layer git commit count/hash based version id to a package? It only has a few data files in meta repo. Jan 21 14:22:54 mcfrisk: i dont really understand what you want to do Jan 21 14:23:49 you want a recipe which fetches a specific git revision or a layer of a specific git revison? Jan 21 14:25:20 fl0v01: I have recipe with just file source, both in the same git tree. I want the PR of package to update automatically when the source file changes. Jan 21 14:25:34 ah yes Jan 21 14:25:36 moment Jan 21 14:27:47 you have to set Jan 21 14:27:48 SRCREV = "${AUTOREV}" Jan 21 14:27:54 and reason why I want to do this is that updating the PR is causing merge conflicts. Jan 21 14:28:14 fl0v01: thanks! I'll go and rtfm about that... Jan 21 14:28:23 mom Jan 21 14:28:24 check ${SRCPV} Jan 21 14:28:55 yeah you also need Jan 21 14:28:55 PV = gitr${SRCPV} Jan 21 14:28:59 or something like that Jan 21 14:29:04 and also meta-openembedded/meta-oe/classes/gitpkgv.bbclass if you want more deterministic revision cound Jan 21 14:30:06 I basically need the 'git describe' of the recipe and src file tree converted to PR. Jan 21 14:31:42 and not from a SW component git tree but the recipe git tree directly.. Jan 21 14:31:42 mcfrisk: don't use PR explicitly, set PV as fl0v01 said and use PR service if you need it Jan 21 14:33:21 sigh, can't deploy PR service that fast... Jan 21 14:43:44 mcfrisk: why do you want the revision of meta-layer? Jan 21 14:45:43 if you really want it then check meta/classes/metadata_scm.bbclass Jan 21 14:48:15 JaMa: because the recipe and the only data file live in the meta layer for easy reviews and modifications. There is no need for another git tree and indirections. Jan 21 15:26:10 oh, evemu looks handy, the record/playback capability in particular Jan 21 15:36:04 yeah Jan 21 15:36:23 there's a bug "input subsystem testing" in which i mulled that Jan 21 15:37:50 I'm using evemu recorder atm Jan 21 15:38:55 Hey all, can anyone point me to a good resource on getting python3 and pip3 running on yocto poky? Jan 21 15:39:00 my Google FU is failing me Jan 21 15:42:31 install python3-core python3-modules python3-pip Jan 21 15:42:45 *really* need python3 to install everything Jan 21 15:45:02 Does anybody know how to run custom installation scripts during a bitbake build? I've got problems finding the kernel headers even though I depend on virtual/kernel. Jan 21 15:46:31 what exactly are you asking Jan 21 15:46:41 "custom installation scripts" by definition are going to be different from one to the next Jan 21 15:46:46 we aren't psychic Jan 21 15:49:44 and with recent Yocto releases having virtual/kernel dependency isn't enough Jan 21 15:50:24 i just meant a script which installs a driver Jan 21 15:50:31 do_configure[depends] += "virtual/kernel:do_shared_workdir" Jan 21 15:51:30 thanks JaMa, will give that a shot Jan 21 15:52:39 rburton: thanks for your assistance, when I try to run 'bitbake python3-core python3-modules python3-pip' this firs two seem to succeed but when attempting to install python3-pip it returns this error 'Nothing PROVIDES 'python3-modules' any ideas what I might be doing wrong? Jan 21 15:52:48 Hairy: install those, not bitbake those Jan 21 15:53:03 the recipes if you want to bitbake them on their own are python3 python3-pip Jan 21 15:53:22 (i listed the packages to add to your iMAGE_INSTALL) Jan 21 15:53:33 thanks a lot, ill try that :) Jan 21 15:55:06 ERROR: Fetcher failure for URL: 'http://cache.ruby-lang.org/pub/ruby/2.3/ruby-2.3.0.tar.gz'. Checksum mismatch! Jan 21 15:55:07 File: '/home/ross/Yocto/downloads/ruby-2.3.0.tar.gz' has md5 checksum e81740ac7b14a9f837e9573601db3162 when 326e99ddc75381c7b50c85f7089f3260 was expected Jan 21 15:55:07 awesome Jan 21 15:55:13 thanks, ruby Jan 21 15:55:41 ugh, lovely Jan 21 15:55:53 better figure out why it changed, could have been compromised Jan 21 15:56:00 * kergoth kicks bitbake Jan 21 15:56:35 File: '/home/ross/Yocto/downloads/ruby-2.3.0.tar.gz' has sha256 checksum ba5ba60e5f1aa21b4ef8e9bf35b9ddb57286cb546aac4b5a28c71f459467e507 when 5ffc0f317e429e6b29d4a98ac521c3ce65481bfd22a8cf845fa02a7b113d9b44 was expected Jan 21 15:56:42 whereas Jan 21 15:56:43 Current stable: https://cache.ruby-lang.org/pub/ruby/2.3/ruby-2.3.0.tar.gz Jan 21 15:56:44 sha256: ba5ba60e5f1aa21b4ef8e9bf35b9ddb57286cb546aac4b5a28c71f459467e507 Jan 21 15:57:18 JaMa: The file contains all the files that are required for DFU in SRC_URI it also contains an additional task and additional installs. I have additional features like that and I would prefer to include these as a whole instead of distributing the conditions all over the file. Jan 21 15:57:32 if the tarball was compromised they also updated the web site Jan 21 15:57:33 RP: can you remind me of the behavior of ideptask? Jan 21 15:58:10 rburton, it's possible they rebuilt the package and didn't tell anyone.. I've seen that before.. Jan 21 15:58:16 yeah me too Jan 21 15:58:18 yep Jan 21 15:58:20 but if you have an older copy, definitely worth comparin Jan 21 15:58:29 that's the most common cause, really, but best to be cautious Jan 21 15:58:39 i don't, only hit this as i did a world Jan 21 16:03:16 rburton: latest ruby I have on jenkins server is 2.2.2 Jan 21 16:03:32 2.3.0 isn't in oe-core yet :) Jan 21 16:03:54 aah Jan 21 16:04:00 326e99ddc75381c7b50c85f7089f3260 downloads/ruby-2.2.2.tar.gz Jan 21 16:04:15 someone didn't update checksum in ruby upgrade Jan 21 16:04:22 not upstream ruby fault Jan 21 16:04:27 ahh there we go.. good Jan 21 16:04:58 no, the python-nose upgrade sneaked a rename in! Jan 21 16:07:17 JaMa: well spotted, thanks Jan 21 16:07:34 ah Jan 21 16:08:03 alejandro was obviously midway though the ruby upgrade and accidently committed the rename for an unrelated fix Jan 21 16:09:18 * kergoth likes to use separate project areas for every single task he works on, with isolated clones of all the repositories Jan 21 16:09:29 downside: i forget where the hell i put something i was working on at times :P Jan 21 16:10:02 agreed on both Jan 21 16:10:11 $ git branch | wc -l Jan 21 16:10:12 47 Jan 21 16:10:35 branches and workdirs instead of clones Jan 21 16:10:38 git workdir ++++ Jan 21 16:11:01 i'll have to think about using that a bit more Jan 21 16:12:05 rburton: got a bit further thanks to your help but I still keep running into these errors "UnicodeDecodeError: 'utf-8' codec can't decode byte 0xb0 in position 37: invalid start byte" my first thought was maybe the locale wasnt setup correctly but that doesnt seem to be it. Is there some sort of charset or something else I'm missing, sorry to keep pestering you Jan 21 16:13:11 probably missing locale/gconv packages Jan 21 16:13:30 I noticed that this '>>> print(sys.getdefaultencoding()) will return this 'utf-8' on yocto but on my macbook returns 'ASCII' Jan 21 16:13:47 thanks for that, ill have a little poke :) Jan 21 16:14:14 rburton: Thanks for the hint it seems to work. Jan 21 16:25:55 moin Jan 21 16:26:06 moin Jan 21 16:52:16 kergoth: you mean recideptask ? Jan 21 16:52:39 yeah Jan 21 16:52:59 kergoth: It solves a problem on knowing which task to use when recursing to look at other tasks. Jan 21 16:53:30 kergoth: for example if a depends on b, should it look at b:do_build or just the b: Jan 21 16:54:08 kergoth: recrdepends defaults to but the ideptask can promote it to some other task Jan 21 16:54:36 kergoth: This is having switched recrdepends between several behaviours... Jan 21 16:55:13 kergoth: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=5af68d42ba62a2a18d1b0cdacad8e32c2b119255 perhaps might explain it better Jan 21 16:55:41 ah! that helps, thanks Jan 21 17:12:37 khem: bruce has a 4.4 linux-libc-headers patch which obviously conflicts with your musl patch, can you rebase onto that? Jan 21 17:27:12 rburton: is it pushed ? Jan 21 17:27:59 khem: its in ross/mut in poky-contrib, or on the list. Jan 21 17:28:11 ok Jan 21 17:28:58 rburton: http://patchwork.openembedded.org/patch/113125/ Jan 21 17:29:03 is that the one ? Jan 21 17:29:15 yes Jan 21 17:29:40 ok Jan 21 17:29:57 poky-contrib:ross/mut-musl is your branch, rebased to mut (one conflict in libsolv), minus your linux-libc-headers patch Jan 21 17:30:05 *so close* :) Jan 21 17:44:32 rburton: https://github.com/kraj/openembedded-core/commit/fc1f8a24d2f4895c49fd4bb02e8c7b37208357a2 Jan 21 17:44:51 I have rebased the patches Jan 21 17:44:55 also Jan 21 17:45:15 but you can throw your old libc-headers patch and then git am this onw Jan 21 17:45:30 you can get raw patch from github like this Jan 21 17:45:35 wget https://github.com/kraj/openembedded-core/commit/fc1f8a24d2f4895c49fd4bb02e8c7b37208357a2.patch Jan 21 17:45:59 libsolv is rebased too Jan 21 17:46:05 its here Jan 21 17:46:41 wget https://github.com/kraj/openembedded-core/commit/2bbce92f40514dd145a3e693d8645d4b96ab0d1d.patch Jan 21 17:47:56 rburton: one of the kernel patches showed conflicts which were just in comments mostly pretty straight forward. Thats all fixed now Jan 21 18:37:40 JaMa: Thanks again for your suggestion. I've added a line do_install[depends] += "virtual/kernel:do_shared_workdir", but my install script still seems to be unable to find the /usr/src/kernel dir. Jan 21 18:38:00 I have the only StackOverflow bounty up under the ‘yocto’ tag if you or anyone else would like to see my issue in more detail for some (hopefully) easy rep. Jan 21 18:39:37 karobar: you shouldn't use /usr/src/kernel, it's in ${STAGING_KERNEL_DIR}/include Jan 21 20:34:11 khem: thanks! Jan 21 20:39:55 rburton: I also built an image meanwhile with 4.4 headers and seems to work fine Jan 21 20:50:17 RP: your new line numbering stuff has some bugs. it's pointing to completely wrong lines still for at least some of the entries in a traceback i'm seeing Jan 21 21:04:04 we need more QA tests for such things... Jan 21 21:06:16 kergoth: bug reports welcome Jan 21 21:06:42 kergoth: I would only claim to have made it more accurate than it was previously, there are still way too many corner cases :/ Jan 21 21:07:10 kergoth: e.g. the right file is a good starting point Jan 21 21:18:57 a headers only package when added as DEPENDS to a recipe ands that recipe appearing in image create a runtime dep on this headers package Jan 21 21:19:03 thats not expected Jan 21 21:19:38 obviously PN is empty for such a package Jan 21 21:20:19 so either ALLOW_EMPTY or a not empty PN fixes it Jan 21 21:20:35 but I wonder if depends to rdepends converter code should handle this Jan 21 21:35:00 RP: heh, fair enough :) Jan 21 21:51:28 kergoth: seriously though, please do file bugs with cases where its not right Jan 21 21:51:38 kergoth: its the only way we'll fix them Jan 21 21:51:39 WIll do Jan 21 21:52:10 in this case, it was pointing to a line in the bitbake metadata for a python exception, in the middle of a multiple python function callstack :) Jan 21 21:52:16 i'll try to get a reproducer Jan 21 21:52:29 erm, recipe metadata Jan 21 22:58:17 The license listed BSD-4-Clause was not in the licenses collected for recipe fts Jan 21 22:58:25 how to fix such warns Jan 21 22:58:38 the package itself does not have a LICENCE file Jan 21 23:02:20 khem: we did fix a problem relating to this a few months ago Jan 21 23:03:37 khem: http://cgit.openembedded.org/openembedded-core/commit/?id=ef7dc532e800d9b170246550cbc8703adf624beb Jan 21 23:04:07 not sure which branch you're using, but if it doesn't have that fix in it that may be the cause Jan 21 23:05:22 FYI it only got backported to jethro on Jan 12th Jan 21 23:20:16 i m running make and gettign following error Jan 21 23:20:24 make: W: Command not found Jan 21 23:20:32 anyone see this error before? Jan 22 00:11:22 sgw_, ping Jan 22 02:42:53 I've got a recipe with patches failing to apply, oddly enough I can't get into devshell and I think this is how I used manage patch queues Jan 22 02:42:57 has this workflow changed? Jan 22 02:43:19 nope, devshell has always run after patch, as otherwise people couldn't modify patched sources Jan 22 02:43:24 tried with bitbake master & 1.28 release Jan 22 02:43:30 :( weird Jan 22 02:43:52 there are really multiple use cases for devshell. fixing a broken patch is only one of them Jan 22 02:44:00 we should probably add a second task Jan 22 02:44:15 that's an interesting idea Jan 22 02:45:07 if you set PATCHRESOLVE = "user", a failed patch will automatically open a devshell for the user to resolve the failure, and fixes to the patch made with quilt (i.e. don't forget to run quilt refresh after fixing the rejects) are supposed to be pushed back to the patche in the layers Jan 22 02:45:14 not sure if it still works, it's not often used Jan 22 02:46:07 thanks I'll give that a go Jan 22 02:46:13 but we should still add a second devshell that runs after do_unpack instead of do_patch.. either that, or the ability to drop into a shell could be made implicit to the system, a bitabke feature, which could be run at any point in the process Jan 22 02:46:15 np Jan 22 02:57:27 Eric___: if you provide more context maybe someone can help. What configururation, version of yocto, which package has make complaining? **** ENDING LOGGING AT Fri Jan 22 02:59:58 2016