**** BEGIN LOGGING AT Mon Jun 20 02:59:59 2016
Jun 20 08:21:45 <CTtpollard> if I have incompatibility gplv3 selected, and a recipe has a non gplv3 / older version available, does bitbake know to select it/
Jun 20 08:39:58 <LocutusOfBorg> CTtpollard, yes
Jun 20 08:39:59 <LocutusOfBorg> it does
Jun 20 08:40:20 <LocutusOfBorg> it should exclude all the gplv3 recipes, and then select the higher version from non-gplv3 recipes
Jun 20 08:40:35 <LocutusOfBorg> I did bitbake tar, and it picked up the gpl3 one
Jun 20 08:40:51 <LocutusOfBorg> I did blacklist gplv3 and did again a bitbake tar, it picked the old one
Jun 20 08:41:02 <CTtpollard> cool, I'm seeing that with bash and gawk
Jun 20 09:56:40 <boucman_work> Hello everybody
Jun 20 09:56:56 <boucman_work> I have a slight problem with building multi-partition images with yocto
Jun 20 09:57:23 <boucman_work> I have a customer that wants a  RO rootfs with a RW partition mounted as /settings
Jun 20 09:58:06 <boucman_work> I know enough wic to create a multi-partition image, where /settings is an empty partition, but in this case there are some files (coming from various recipes) that need to end in /settings
Jun 20 09:58:58 <boucman_work> I can easily have my recipes install stuff in /settings but they will end up in the rootfs image how do I tell yocto (or wic, if it's wic's responsability) to remove the /settings content from rootfs and put it in a separate partition "as-is"
Jun 20 09:59:33 <boucman_work> or do I do two yocto images, one with rootfs, one with /settings, but how do I tell /settings that it doesn't need a kernel, compiler etc...
Jun 20 10:33:58 <lewiatan> Hi. I know lots of quiestions has been asked about this subject but one more time I have a question about PACKAGECONFIG
Jun 20 10:34:31 <lewiatan> If I want to change PACKAGECONFIG flags then where should I put the variable PACKAGECONFIG_pn_(some_recipe) ?
Jun 20 10:34:43 <lewiatan> will it work if I put it to bbappend file?
Jun 20 10:34:55 <lewiatan> will it work if I put it into my distro.conf?
Jun 20 10:35:48 <boucman_work> both would work but...
Jun 20 10:36:15 <boucman_work> if you put it in a bbappend, you don't need the _pn-<recipe> part, since a bbappend is already limited to a single recipe
Jun 20 10:36:53 <boucman_work> and distro.conf is probably not the cleanest place. it would work but this is not where you usually put it. local.conf is a more logicall place
Jun 20 10:37:03 <boucman_work> a bbappend is probably the best place, though
Jun 20 10:44:54 <srinivas__> Hi bluelightening
Jun 20 10:45:23 <srinivas__> bluelightning:Hi
Jun 20 10:45:40 <srinivas__> Hi All
Jun 20 10:45:50 <bluelightning> hi srinivas__
Jun 20 10:46:24 <srinivas__> How to override a class , in my custom layer
Jun 20 10:46:47 <srinivas__> in my case i have 2 bbclasses with the same name in different layers
Jun 20 10:47:13 <srinivas__> I would like to specify preference to one layer to pick bbclass
Jun 20 10:51:57 <lewiatan> thanks @boucman
Jun 20 10:53:57 <bluelightning> srinivas__: classes are picked up via bbpath, so the first one found will win
Jun 20 10:54:12 <bluelightning> srinivas__: that said I usually advise not trying to override entire classes
Jun 20 11:06:46 <srinivas__> Ok
Jun 20 11:06:57 <srinivas__> Thanks bluelightning
Jun 20 11:55:45 <Anticom> Hi all. According to environment-setup-* in my SDK installation, all *.sh files in environment-setup.d are getting sourced aswell. Now i was wondering how to include additional script files in those directories
Jun 20 11:55:52 <Anticom> Any clues?
Jun 20 11:56:42 <Mylene> /1/5
Jun 20 11:57:06 <Mylene> oops, sorry
Jun 20 12:01:12 <Anticom> I guess it's got to do with toolchain-scripts.bbclass but i can't find any proper documentation :(
Jun 20 12:20:27 <fredcadete> Anticom: I'm not sure this is the canonical way to do it, but meta-qt5 does that: https://github.com/meta-qt5/meta-qt5/blob/master/recipes-qt/qt5/nativesdk-qtbase_git.bb#L230
Jun 20 12:22:39 <Anticom> fredcadete: seems somewhat reasonable tho
Jun 20 12:22:40 <Anticom> thanks
Jun 20 14:26:27 <lkj> Hi. What is the reason for "installed vs shipped" error
Jun 20 14:26:29 <lkj> ?
Jun 20 14:28:02 <LocutusOfBorg> you install foo with make install
Jun 20 14:28:11 <LocutusOfBorg> but there is no package actually having it
Jun 20 14:28:22 <LocutusOfBorg> so, a mismatch between install and package-split contents
Jun 20 14:28:47 <LocutusOfBorg> and e.g. then no "rpm" contains foo
Jun 20 14:31:14 <lkj> I'm trying to extend bluez bitbake with do_install_append and I created a folder in my bbappend file. But bluez already has a do_install_append in its bitbake and I'm getting installed vs shipped error
Jun 20 14:32:34 <lkj> What might be going wrong. How do I relate "installed vs shipped" error with whats happening?
Jun 20 14:33:29 <lkj> Is do_install_append in my bbappend is getting called but not putting those folders into rpm?
Jun 20 14:34:48 <boucman_work> lkj: having multiple do_install_append should be possible
Jun 20 14:34:52 <rburton> your do_install_append is putting files into the staging directory but you didn't extend FILES to actually ship the files in a package
Jun 20 14:38:12 <lkj> rburton: Ohh. Thanks
Jun 20 14:39:12 <Circuitsoft> Hello - I'm trying to enable full-disk-encryption on an x86 install, and when installing my image, there seems to not be an initramfs.
Jun 20 14:39:24 <Circuitsoft> Does the kernel have built-in support for root=PARTUUID=... now?
Jun 20 14:39:48 <Circuitsoft> I guess that's a side question. How can I use an initramfs for normal boot?
Jun 20 14:41:40 <boucman_work> so... no way to build a multi-partition image easily with yocto (in the case where both partitions need to contain files generated by yocto, like a rootfs with a separate /usr)
Jun 20 14:41:42 <boucman_work> ?
Jun 20 14:42:45 <boucman_work> at this point I can only see a .tar.gz image with at post-rootfs command to remove the directory I want to transfer and save it separately in $DEPLOYDIR as another .tar.gz
Jun 20 14:42:57 <boucman_work> then use a wic image depending on that to rebuild it as separate partitions
Jun 20 14:43:24 <Circuitsoft> On mine, I'm just mounting a second FS at /var/lib/postgresql and re-running postgres setup on startup.
Jun 20 14:44:04 <boucman_work> Circuitsoft: that works because your files are not managed by the package manager, they are generated on the target
Jun 20 14:44:21 <Circuitsoft> Yup.
Jun 20 14:44:25 <boucman_work> thus, your second FS is empty when yocto generate images
Jun 20 14:44:27 <Circuitsoft> Yup.
Jun 20 14:44:34 <boucman_work> so... that's not my use-case :P
Jun 20 14:44:53 <Circuitsoft> Are you trying to support partial-system upgrades with disk images?
Jun 20 14:45:06 <Circuitsoft> boucman_work: Actually, I do know how to handle your use case.
Jun 20 14:45:23 <boucman_work> oh ?
Jun 20 14:45:26 <boucman_work> cool
Jun 20 14:45:34 <Circuitsoft> Make an initramfs-live-install_1.0.bbappend
Jun 20 14:45:57 <Circuitsoft> Find the part where it's partitioning the disk, setup your partitions there, and make sure your fstab has everything right.
Jun 20 14:46:08 <Circuitsoft> That way, all your folders will be mounted before it copies the rootfs files over.
Jun 20 14:46:36 <Circuitsoft> So while the rootfs on the install media will be a single ext4/archive, it'll extract to a tree of mounts.
Jun 20 14:46:54 <boucman_work> that will only work with the "live-images" which are x86 only
Jun 20 14:47:12 <boucman_work> i'm on an ARM and the system is factory-installed
Jun 20 14:47:30 <boucman_work> (but yeah, that's a good idea, I used it for another customer....)
Jun 20 14:48:40 <Circuitsoft> boucman_work: Seems you should start by doing that, then extract ubi images off your dev system to be written to the manufactured devices.
Jun 20 14:48:52 <lkj> rburton ERROR: QA Issue: bluez5: Files/directories were installed but not shipped   /usr/lib/firmware   /usr/lib/firmware/ti-connectivity   /usr/lib/firmware/ti-connectivity/TIInit_6.7.16.bts [installed-vs-shipped
Jun 20 14:49:05 <rburton> lkj: so set FILES
Jun 20 14:49:49 <lkj> That was the error I was getting. Now I added only "/usr/lib/firmware/ti-connectivity" and the error is gone. Wasn't I supposed to add everything?
Jun 20 14:49:54 <boucman_work> that's not a process that can be automated... I need to provide sources and build-instructions, I can't install on the machine then extract from the machine...
Jun 20 14:50:37 <lkj> FILES_${PN} += "${libdir}/firmware/ti-connectivity"
Jun 20 14:51:02 <lkj> Why isn't there an error even though I didn't add everything
Jun 20 14:51:12 <lkj> ?
Jun 20 14:51:25 <Circuitsoft> lkj: Since those items are all part of the same path, you could add either /usr/lib/firmware or /usr/lib/firmware/ti-connectivity/TIInit_6.7.16.bts and get the same result.
Jun 20 14:51:38 <Circuitsoft> boucman_work: How do you build your current flash images for manufacturing?
Jun 20 14:51:49 <rburton> so the actual file is /usr/lib/firmware/ti-connectivity/TIInit_6.7.16.bts and that matches the FILES, so it got packaged
Jun 20 14:51:50 <boucman_work> bitbake :P
Jun 20 14:52:08 <boucman_work> (we don't have multi-partition yet... thus my question)
Jun 20 14:52:15 <Circuitsoft> boucman_work: Oh, you mean like, with a computer?
Jun 20 14:52:24 <Circuitsoft> How do you do single-partition?
Jun 20 14:52:27 <Circuitsoft> What is the recipe like?
Jun 20 14:52:51 <boucman_work> there is a whole bunch of recipes that make an image that is i.MX6 compatible
Jun 20 14:53:20 <boucman_work> the rootfs itself is quite standard, then some FSL-provided classes will create a .sdcard file
Jun 20 14:53:44 <boucman_work> by assembling the rootfs, uboot, and a FAT partition with the kernel, dtb and a few other files
Jun 20 14:54:02 <boucman_work> that image will be passed to manufacturers
Jun 20 14:54:37 <boucman_work> the FSL scripts don't use wic, but I looked into it and there is nothing magic... it could be converted to wic if that's needed
Jun 20 14:57:28 <lkj> Circuitsof rburton Thankyou :)
Jun 20 14:57:42 <lkj> Circuitsoft
Jun 20 15:55:08 <fredcadete> hi, I'm having issues in the parsing step (https://gist.github.com/fredcadete/7fd42f8e2b23e66d71665de731d9060a) and I'm wondering how I can debug this
Jun 20 15:55:40 <fredcadete> my usual instinct is to use bitbake -e but if the problem is in the parsing it's no good
Jun 20 15:58:01 <khem> fredcadete: its giving you a good message "The SRCREV_FORMAT variable must be set when multiple SCMs are used."
Jun 20 15:58:18 <khem> so set SRCREV_FORMAT in the recipe
Jun 20 16:00:49 <fredcadete> khem: in fact this is happening for all efl recipes, and as far as I can tell they should have a single SCM set and SRCREV_FORMAT set
Jun 20 16:01:20 <khem> then check if the efl servers are online or not
Jun 20 16:02:10 <fredcadete> ... from the error message it fails even before going online ...
Jun 20 16:02:47 <fredcadete> is there a way to debug variable assignments when "bitbake -e" does not work?
Jun 20 16:14:30 <fredcadete> got it. the new machine set up multilib, which messed up this line: http://cgit.openembedded.org/meta-openembedded/tree/meta-efl/recipes-efl/e17/e-module.inc#n8 because ${PN} is lib32-package
Jun 20 16:15:32 <fredcadete> I got there by adding logger.warn in the bitbake python code, if anyone has a better method please let me know
Jun 20 16:16:06 <khem> replace PN with BPN
Jun 20 16:16:11 <fredcadete> yeah, I know
Jun 20 16:16:14 <khem> in the recipe
Jun 20 16:16:23 <khem> you can use bitbake -DDD
Jun 20 16:16:35 <khem> so get live parsing data
Jun 20 16:16:39 <khem> but thats pretty much
Jun 20 16:17:39 <kergoth> yeah, there's not a lot of ways to diagnose a parsing failure, it'd be nice to provide more, but the options are limited
Jun 20 16:18:37 <fredcadete> good to know, I can just add python code instead of looking for a better way :)
Jun 20 16:22:37 <kergoth> it'd be nice to allow -e to happen even if event handlers or anonymous python fail, but dump the failure messages as comments in the output, and return non-zero to indicate the error state. not sure if it'd be possible to do that for immediate expansions, though. we'd have to rework the parsing api to do a best effort, return incomplete metadata on error, and return the errors encountered
Jun 20 16:24:19 <khem> would it be possible for bb to dump out half cooked store ?
Jun 20 16:24:48 <khem> if it would be then would it be correct to a degree or just more confusing with partial expantions
Jun 20 16:26:17 <kergoth> yeah, that's what i was thinking. we'd ahve to make sure it's clear that it's incomplete
Jun 20 16:26:29 <kergoth> and the incomplete store could raise its own failures due to the incomplete state
Jun 20 16:26:35 <kergoth> (i.e. when vars are expanded to dump them in -e)
Jun 20 16:27:00 <kergoth> probably more trouble than its worth, but i've been in fredcadete's situation, and it's not fun to deal with
Jun 20 16:27:05 <kergoth> so doing something would be nice
Jun 20 16:27:18 <fredcadete> even unexpanded variables would be helpful
Jun 20 16:27:29 <fredcadete> although immediate expansions would make it complicated
Jun 20 16:27:39 <fredcadete> anyway, I'm talking out of depth here
Jun 20 16:27:41 <fredcadete> see you tomorrow
Jun 20 16:37:03 <Circuitsoft> How can I use an initramfs for boot of an installed image on x86?
Jun 20 16:39:23 <kergoth> first you need an initramfs image to use. most of the ones in oe-core are associated with the live image stuff. not many options for just a more typical use of an initramfs. once you have one that does what you want, you can use it by setting INITRAMFS_IMAGE = "image-name". there's an option to bundle it into the kernel as well. now actually using it is something else. if you use the live/hddimg/bootimg classes to build a disk image, it'll be used, but
Jun 20 16:39:23 <kergoth>  if you use wic or some other disk image construction mechanism, it won't, you'd have to arrange that yourself
Jun 20 16:39:31 * kergoth isn't very happy with the state of our initramfs support
Jun 20 16:40:14 <Circuitsoft> I created an initramfs and set INITRAMFS_IMAGE="my-image-name" in my image recipe, but I don't see where it's loaded. Maybe I need to add a line to init-install.sh?
Jun 20 16:40:19 <Circuitsoft> I've already customized that some.
Jun 20 16:40:43 <Circuitsoft> Where, on my dev system, can I browse what will be in /run/media/$1/ in that script?
Jun 20 16:41:31 <Circuitsoft> Found it.
Jun 20 17:06:49 * nerdboy doesn't have an answer for "conditional-on-host-toolchain-version"
Jun 20 17:17:16 <nerdboy> anybody?
Jun 20 17:19:25 <Circuitsoft> So, I have a system-image.bb that has INITRAMFS_IMAGE="system-image-initramfs".
Jun 20 17:20:14 <Circuitsoft> system-image-initramfs.bb has PACKAGE_INSTALL="initramfs-boot ${VIRTUAL-RUNTIME_base-utils} udev base-passwd ${ROOTFS_BOOTSTRAP_INSTALL}"
Jun 20 17:20:34 <Circuitsoft> But, when I boot my system, the initramfs seems to be starting init-live.sh rather than init-boot.sh
Jun 20 17:24:59 <Circuitsoft> To be exact, my system-image.bb has INITRAMFS_IMAGE = "${PN}-initramfs"
Jun 20 17:25:10 <Circuitsoft> But it seems to be pulling in the wrong initramfs
Jun 20 17:28:44 <rishabh> Hello ,just wanted to know whether it is possible to install mongoDb on yocto or not ?
Jun 20 17:30:43 <Circuitsoft> rishabh: There is a recipe for it in meta-oe.
Jun 20 17:32:38 <Circuitsoft> So, add the meta-oe layer to your build, and add mongodb to RDEPENDS_${PN} in the recipe for your app that uses it.
Jun 20 17:34:30 <rishabh> alright .is there any other database that can be used directly ?without any workarounds?
Jun 20 17:37:33 <rishabh> Circuitsoft ^
Jun 20 17:38:27 <Circuitsoft> That's hardly a workaround - that's the way Yocto is supposed to work.
Jun 20 17:39:38 <Circuitsoft> But, if you insist, there's BerkelyDB, GDBM, SQLite.
Jun 20 17:39:49 <Circuitsoft> None of them are comparable to MongoDB.
Jun 20 17:41:52 <rishabh> Thanks a lot  .'
Jun 20 17:48:11 <Circuitsoft> kergoth: So, with bitbake -e, I'm finding INITRD to be using INITRD_LIVE, when I have INITRAMFS_IMAGE set to my own image recipe. Where can I track this down?
Jun 20 17:49:44 <Circuitsoft> Also, to be clear, if in core-image-my-server.bb I have 'INITRAMFS_IMAGE = "${PN}-initramfs"', will that make INITRAMFS_IMAGE be core-image-my-server-initramfs.bb?
Jun 20 17:54:43 <Circuitsoft> Looks like I needed to set INITRD_IMAGE instead of INITRAMFS_IMAGE
Jun 20 17:58:10 <Crofton|work> zeddii, I am becoming more convinced linux-yocto-dev is broken on krogoth
Jun 20 19:37:30 <kergoth> hmm, lirc inherits pythonnative, but requires python 3, lirc-python rdeps on python2, but needs python3, and the main lirc package includes python 3 scripts but has no python rdep at all
Jun 20 21:07:45 <igor> guys, there is a way to resize the hddimg partition size?
Jun 20 21:08:21 <khem> kergoth: indeed that needs fixing
Jun 20 21:08:33 <khem> igor: are you looking for more free space ?
Jun 21 01:05:40 <nerdboy> okay, seems like git send-email is ignoring --subject-prefix=whatever and using the one from the patch header
Jun 21 01:05:57 <nerdboy> which is just [PATCH]
Jun 21 01:11:04 <nerdboy> aaaand...  i don't see that option in the man page anymore either
Jun 21 01:32:20 <kergoth> nerdboy: are you using send-email on a patch file?
Jun 21 01:32:34 <kergoth> send-email calls format-patch under the hood, options like subject-prefix are passed to that
Jun 21 01:32:48 <kergoth> if you operate against patch files, format-patch isn't called at all, so those options will have no effect
Jun 21 01:33:11 <kergoth> eitehr pass them to you own format-patch or create-pull-request, or use send-email on commits rather than patch files
Jun 21 01:33:40 <kergoth> if you see the man page to git format-patch, you'll see the options like that covered there
**** ENDING LOGGING AT Tue Jun 21 02:59:58 2016