**** BEGIN LOGGING AT Mon Aug 29 02:59:58 2016 Aug 29 06:31:13 hi guys Aug 29 08:20:12 hi Aug 29 08:24:41 hmz Aug 29 08:24:50 i got highlighted, but it's too far back in the scrollback Aug 29 08:31:12 is there a way to define conflicting packages within a recipe? Aug 29 08:33:07 how can i select different commit in a .bbappend ? a_git.bb (SRCREV="1" -> a_git.bbappend (SRCREV="2") Aug 29 08:35:10 SRCREV in .bbapend doesn't work Aug 29 08:37:06 if I edit a_git.bb and change the SRCREV to "2" works ok. However if i Set the SRCREV to "2" in a_git.bbappend it doesen't. Aug 29 08:45:08 Found the issue I was editing b_git.bbappend Aug 29 11:05:53 Hi, I am creating a recipe for a new package. I want to include some scripts (and config files for those scripts) so that users of my toolchain can do some stuff in an easy way. Aug 29 11:05:59 I am getting this QA message: ERROR: QA Issue: : Files/directories were installed but not shipped in any package: (...) Please set FILES such that these items are packaged. Alternatively if they are unneeded, avoid installing them or delete them within do_install. Aug 29 11:06:24 The thing is, I know that those files are shipped and they are not in any package. But that is totally fine. Is there a way to tell that to Yocto in order to avoid the warnings? Aug 29 11:10:53 tlk__: you can disable the check with INSANE_SKIP_ = "installed-vs-shipped". you might miss out on a good sanity check though. Aug 29 11:14:12 bit that error says something different. it says that the recipe installs a couple of files into its sysroot, but no package picks them up. its the exact opposite, they will *not* be shipped upon reproducing the build. Aug 29 11:26:41 LetoThe2nd: I want the files in the toolchain, but not in the SDK Aug 29 11:26:55 sorry, I want the files in the SDK, but not in the sysroot *** Aug 29 11:27:01 that is what I meant Aug 29 11:27:10 I guess I am doing something wrong in the recipe, then Aug 29 11:28:48 tlk__: probably, yes. that warning usually is a good hint that something in your install steps is not doing what you probably expect it to. Aug 29 11:31:03 is there some way to force a package to be deployed in a rootfs for no specific arch? actually my recipe inherits from 'allarch', but i figured out that if the package has previously been generated for a particular arch, then the package deployed is the one matching the machine arch. Aug 29 11:31:45 I have to clean everything to make it work properly, even the corresponding sstate-cache Aug 29 11:43:44 when using gitsm:// SRC_URI bitbake fails because it tries to clone the submodules to a non-existend directory "..../${PN}.tmp". Is that expected? How would I work around that? Aug 29 11:43:55 speaking of Dylan here :D Aug 29 12:43:17 hi all, can someone tell me how can I found the original package of a file present on my final image, for example I have got libasound.a and it come from alsa-lib-1.0.27.2-r0 Aug 29 13:00:59 hi everyone Aug 29 13:01:27 is it possible to build 2 sdk for 32 and 64 bits architectures Aug 29 14:37:00 How do you specify a partition without a mount point in wic? Aug 29 16:06:16 what's the best way of exporting all SRCREV / SRC_URI pairs of the packages included in building a specific target? bitbake -b and some custom logic is what I'm thinking Aug 29 16:30:03 blueCmd_, modify base class to output SRC_URI in a custom, newly created, do_*** method? Aug 29 16:30:29 then bitbake -c $new_method target Aug 29 16:33:58 darwish: ah, cool - thanks! Aug 29 16:38:47 blueCmd_, not an expert in OE though; others might have better-engineered solutions Aug 29 16:45:03 that'd just get you the info for the target, not the target's deps, unless you create a recrdeptask version, but that's a reasonable way to go. see do_fetchall for details on how to do it for its deps Aug 29 16:45:26 and please don't mangle base.bbclass pointlessly, just create a class in build/classes/ or your own layer/classes/ and add it to INHERIT in local.conf Aug 29 16:47:58 fwiw, my usecase is to detect all git repositories matching a certain string and check out at that revision to run the test suite of those repositories Aug 29 16:48:08 to get a build test status for the whole thing as far as possible Aug 29 16:48:27 thinking of it - surely this has been done? Aug 29 16:48:36 like a "do_check" thing Aug 29 16:49:16 ptest is how we run the upstream test suites of projects. many recipes already have support for it. https://wiki.yoctoproject.org/wiki/Ptest Aug 29 16:49:40 they're packaged for running on target or in qemu Aug 29 17:04:45 hello Aug 29 17:04:49 a question Aug 29 17:06:00 can the shared state cache be used in builds for different platforms e.g. beaglebone and raspberry pi ? Aug 29 17:54:44 i mean the same sstate cache Aug 29 17:59:33 iskander: yes, sstate can be used across different architectures Aug 29 18:00:12 great, thx Aug 29 18:00:48 is it worth the trouble ? Aug 29 18:01:07 iskander: yes, it saves quite a bit of time Aug 29 18:01:15 iskander: see here from more info: http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#shared-state-cache Aug 29 18:01:28 well, at least for the -native stuff I guess it saves quite a bit of time Aug 29 18:02:06 neverpanic: if only it could make the weekends last longer Aug 29 18:03:00 I've been running into a TaskHash mismatch error lately, and it *might* come from sharing a sstate cache between two loosely related builds Aug 29 18:03:40 I suspect that because it goes away if I do cleanall for the recipe with the issue Aug 29 18:04:13 but I haven't been able to put my finger on what exactly causes it Aug 29 18:04:48 I ran into with the other day when switching from MACHINE=raspberrypi2 to MACHINE=raspberrypi Aug 29 18:05:30 I mean, a build with the first value and a separate build (with the same sstate-cache directory) for the second Aug 29 18:08:06 the other weird thing I ran into with the same build is that I was adding something to /etc/systemd/network and it wasn't showing up in the final image, but when I put it in /lib/systemd/network/ it was there... I couldn't figure out what was special (from a recipe point of view) about /etc/systemd/network Aug 29 18:09:23 (basically I was trying to use systemd-networkd to obtain an IP address of DHCP for the wired interface, so I needed a wired.network file for that) Aug 29 18:11:31 Is this true of layer names?: Aug 29 18:11:43 so I can't have layer_awesome.2 ? Aug 29 18:11:53 (for example) Aug 29 18:13:50 RP: thoughts on adding a little method to DataSmart to set _defaultval? i.e. setVarDefault? Aug 29 18:14:02 RP: i think there's value in being able to set a ??= programmatically Aug 29 18:14:10 (without poking too far into internals) Aug 29 18:14:27 I can't find where bitbake validates the layer name, but I did find that help in toaster. Aug 29 18:30:56 jedix: Undercores are bad for a couple of reasons, I think Aug 29 18:31:08 sorry, the issue is actually the period Aug 29 18:31:12 bad example :) Aug 29 18:31:20 .2 Aug 29 18:31:31 jedix: Ah ok. I know one particular case that bit me was that Debian packages only allow alphanumeric and '-' Aug 29 18:31:49 This was fun because of course no one tests with Debian packages enabled in buildroot, only rpm :p Aug 29 18:31:57 I blazed that trail inadvisedly Aug 29 18:32:41 some people get all the fun? Aug 29 18:32:44 :) Aug 29 18:34:54 I'll rename the layer Aug 29 18:35:03 (and patch layerindex to not blow up) Aug 29 18:45:21 I read in a talk about Yocto that Poky should not be used for production. Does anyone know why this would be the case? Aug 29 19:27:39 hweaving: poky is a reference distribution. generally folks should create their own distros for production. Aug 29 19:28:15 kergoth: Or at minimum starting with Poky and customizing? Aug 29 19:28:23 I just wanted to make sure there weren't pitfalls / license issues / whatever in Poky Aug 29 19:28:26 that's what creating your own distro would be Aug 29 19:28:33 copy poky.conf to yourdistro.conf and go from there Aug 29 19:28:40 :) thanks Aug 29 19:28:44 or base on nodistro instead, really up to you Aug 29 19:28:58 or if you really want to, you could include/require poky.conf from your distro .conf and make your layer depend on meta-poky Aug 29 19:30:23 mentor graphics' distro directly includes poky that way, but mostly for maintenance reasons. we wanted certain bits from poky and wanted to avoid getting out of sync over the long term. most folks would be best off copying, i expect Aug 29 19:31:33 * Crofton|work avoids poky like the plague Aug 29 19:33:56 the poky distro, or the poky repo? :) Aug 29 19:35:44 Crofton|work: How come? Aug 29 19:36:38 both Aug 29 19:36:57 cant work upstream from poky repo Aug 29 19:37:30 and simple enough to make a clean distro conf Aug 29 20:50:52 kergoth: I agree we should have python api for it Aug 29 21:45:49 Has anyone got php-zmq working? **** ENDING LOGGING AT Tue Aug 30 02:59:58 2016