**** BEGIN LOGGING AT Tue Jun 19 03:00:04 2018 Jun 19 07:25:09 I think the git link is broken on the yocto website... git clone -b sumo git://git.yoctoproject.org/poky.git returns fatal: repository '...' not found Jun 19 07:27:38 Hello Yocto Community Jun 19 07:27:52 Hi Jun 19 07:29:04 how to use apt instead apt-get when selecting deb as package management feature ? Jun 19 07:29:21 no idea :/ Jun 19 07:29:31 i also just joined to ask a question -.- Jun 19 07:29:37 (y) Jun 19 07:29:58 sry Jun 19 07:30:31 Actually apt-get is used by default once we are going to work with deb package management Jun 19 07:31:00 so if someone else has an idea, do not hesitate to ping Jun 19 08:21:22 Hi. I have a problem that I try to solve and I could use some help :). I'm doing a new image fstype which is a signed archive containing u-boot, kernel and rootfs. How can I identify at the creation of the fstype, the names of u-boot binaries (SPL and U-Boot). I did identify the kernel by using KERNEL_IMAGETYPE var. However I'm not able to find a uniform solution for U-Boot. Any ideas? Jun 19 08:47:46 i7clock: can't confirm, the chackout seems to work fine here. Jun 19 08:48:05 distsys: actually, this is not how yocto respectively openembedded package management works Jun 19 08:48:25 distsys: if you actually want ubuntu or debian, then you're probably better off using ubuntu or debian Jun 19 08:49:32 Hi, i am building yocto images for Karo TX6S target (https://www.karo-electronics.com/856.html?&L=1). I would like to know how to create an qemu image for my target. Must i create a new build directory and specify a qemu machine target or on the same build as my Karo target machine build i can build a qemu image? Thanks in advance Jun 19 08:51:59 vladzouth: the former holds true, as you basically build a new MACHINE Jun 19 08:53:14 vladzouth: but no idea if qemu is able to emulate imx6. and it almost certainly is not able to reproduce board specific things (unless you implement that.) Jun 19 08:55:53 Letothe2nd: thanks you! Jun 19 08:56:45 LetoThe2nd: is there a way to know the list of available machine for qemu in yocto? Jun 19 08:58:20 the one i found is https://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#var-MACHINE . But the qemu arm emulate an armv5 but i would like to emulate a cortex-a9 Jun 19 08:58:39 vladzouth: http://layers.openembedded.org/layerindex/branch/master/machines/?q=qemu&search=1 Jun 19 08:59:04 vladzouth: those in openembedded-core are kinda "mainline" Jun 19 09:02:05 LetoThe2nd: Ok thanks! Jun 19 09:06:46 LetoThe2nd: I think you didnt understand me... Actually i was talking about runtime package managemt for the final image Jun 19 09:07:17 LetoThe2nd: once i add package-management to the image feature Jun 19 09:07:29 https://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#using-runtime-package-management Jun 19 09:09:06 distsys: i think i understood you very well. Jun 19 09:09:27 LetoThe2nd: so if the PACKAGE_CLASSES is set to deb, then is it possibe to managemt packages from the final image, using the apt command instead of apt-get Jun 19 09:10:47 distsys: of course you can, technically do that. but its a bad idea, imho. the deb file format is by far the least tested and supported one, so if you totally think you need runtime package management (which is almost never the case in my experience) and you are totally aware of the fact that you have to run and maintain all repository servers yourself then, at least pick either ipk or rpm Jun 19 09:11:30 oh Jun 19 09:11:37 distsys: rpm is the default choice of wind river (and poky) and sees a fair amount of testing, whereas ipk is the openemnbedded default. we hear about funny deb failures now and then, but to be honest, nobody really seems to care Jun 19 09:11:48 distsys: so, do you still think i did not understand the question? Jun 19 09:12:55 distsys: runtime package management is so much more than just installing $PACKAGEMANAGERFRONTENDOFYOURCHOICE. in fact, thats only the last 5% of the whole thing. Jun 19 09:13:20 LetoThe2nd: hmm... I think you did... Now i got your answer... may i didnt your first short answer :) Jun 19 09:13:27 :-) Jun 19 09:14:10 distsys: oh and for your reference, the apt recipe is here: http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-devtools/apt/apt_1.2.24.bb?h=master Jun 19 09:15:12 distsys: the really important point to understand is that just because you might be using deb/apt, neither ubuntu nor debian will feed you. its all up to you, yourself Jun 19 09:22:02 during do compile sysroot folder come's into existence and stores src files Jun 19 09:22:33 but i want to know why do we need sysroot folder, Jun 19 09:22:53 anyone please help. Jun 19 09:23:34 pk_: didn't you ask about the very same question already 3 times this week, and didn't you receive answers? Jun 19 09:24:02 no Jun 19 09:24:36 pk_: each recipe has its own sysroot you can check https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#new-sharing-files-between-recipes maybe Jun 19 09:25:07 pk_: so again: the host-signature sysroot contains all the tooling needed to run on the host, and the target signature sysroot contains basically all the libraries and headers that the toolchain needs (in later revisions of poky/OE per recipe, in earlier ones per build) Jun 19 09:25:44 pk_: here again, is the link to the explanatoin: https://www.quora.com/What-is-a-sysroot Jun 19 09:25:52 pk_: please don't make it another time. Jun 19 09:26:37 ok LetoThe2nd Jun 19 09:29:01 Anyone's using Intel arch? I'll receive a Skylake-S soon and I'm wondering which bootloader to use? (i mostly used arm) Jun 19 09:30:32 And do I need to use genericx86-64 or intel-corei7-64 ? Jun 19 09:30:34 LetoThe2nd: the deb file format is by far the least tested and supported one... so if you totally think you need runtime package management ... at least pick either ipk or rpm... Jun 19 09:30:45 LetoThe2nd: these are very sad news for me Jun 19 09:31:07 nayfe: use meta-intel Jun 19 09:31:36 LetoThe2nd: I just switched from the RPM world (suse, zypper) to the DEB world (debian, apt) Jun 19 09:31:52 distsys: what gain do you expect from deb over rpm, or more specifically ipk? Jun 19 09:32:08 distsys: of course the package manager tooling is mostly irrelevant to that discussion, especially when you're talking about neither suse or debian Jun 19 09:32:30 distsys: ipk is "kinda" compatible, and dpkg can be used to inspect the packages for example Jun 19 09:33:04 LetoThe2nd, rburton: Ok, fine :) Jun 19 09:33:27 distsys: really, the main point is understanding the whole package repository setup, maintenance, in short "burden". after that, picking a specific format is kinda... "hatever" Jun 19 09:33:40 *whatever, even Jun 19 09:33:46 rburton: thx, any advice about bootloader? Jun 19 09:34:46 nayfe: the default one works :) Jun 19 09:35:33 New news from stackoverflow: Difference between SRC_URI and FILESEXTRAPATHS_prepend in bitbake Jun 19 09:38:12 rburton: ok thanks, it seems i worry too much ;) Jun 19 09:52:34 hello Jun 19 09:55:41 switching from rocko to sumo for our local repo, a new error shows-up: 'CMake was unable to find a build program corresponding to "Unix Makefiles". CMAKE_MAKE_PROGRAM is not set' Jun 19 09:56:27 i have tried setting various variables (OECMAKE_GENERATOR and others) in various places (recipe, local.conf, cmake.bbclass) but it did not help Jun 19 09:56:41 is there any known fix for this problem? Jun 19 09:59:21 this is for some of our internal recipes Jun 19 10:00:36 cornel: cmake class defaults to ninja Jun 19 10:00:41 cornel: have you checked upstream if there have been recent fixes for cmake based recipes? Jun 19 10:01:01 why not just use ninja? Jun 19 10:01:12 rburton: yes, but setting OECMAKE_GENERATOR there to Unix Makefiles makes no difference Jun 19 10:01:21 rburton: fear of the dark, probably (tm iron maiden) Jun 19 10:03:12 rburton: LetoThe2nd: the test was like this: on the same local repo clone, i have used the 'sumo' cmake and the 'rocko' cmake. For the 'rocko' cmake it complained about some duplicate files ,for 'sumo' cmake it complained about Unix Makefiles Jun 19 10:03:36 except cmake, everything else was the same Jun 19 10:04:07 now i know that this does noty mean much but for the sake of simplicity i've assumed that the problem comes with the new cmake Jun 19 10:04:47 we are trying to migrate away from some jethro commit Jun 19 10:05:38 cornel: stupid question. have you checked that your OECMAKE_GENERATOR actually has effect? like, in bitbake -e Jun 19 10:05:38 so if i just let poky in the sumo state, our recipe complains about unix makefiles Jun 19 10:06:00 LetoThe2nd: i believe not but i have not actually check Jun 19 10:06:08 thanks for reminding me Jun 19 10:06:15 cornel: then start there :) Jun 19 10:08:07 LetoThe2nd: by default it uses ninja Jun 19 10:09:18 cornel: and does your value actually get applied? or is it overwritten again?? lik Jun 19 10:11:23 LetoThe2nd: just a moment Jun 19 10:13:14 trying to rebuild with changes Jun 19 10:15:36 Hi. I have a problem that I try to solve and I could use some help :). I'm doing a new image fstype which is a signed archive containing u-boot, kernel and rootfs. How can I identify at the creation of the fstype, the names of u-boot binaries (SPL and U-Boot). I did identify the kernel by using KERNEL_IMAGETYPE var. However I'm not able to find a uniform solution for U-Boot. Any ideas? Jun 19 10:16:00 so: OCMAKE_GENERATOR gets the recipe value, then i get a line where generator args is stet to unix makefiles and later i get a line where OECMAKE_GENERATOR_ARGS are again set to Ninja-specific Jun 19 10:16:11 LetoThe2nd: ^ Jun 19 10:16:47 cornel: usually you get the exact file and IIRC even line number where it gets reassigned Jun 19 10:17:08 cornel: maybe you are just setting it in your recipe ABOVE the inherit cmake? :-P Jun 19 10:17:17 ah Jun 19 10:17:23 it' an if-else Jun 19 10:17:48 if Unix set unix, if Ninja set ninja Jun 19 10:18:00 the py to set mojo is probably running a bit earlier so set OECMAKE_GENERATOR before the inherit Jun 19 10:18:38 i see Jun 19 10:18:42 let me try it Jun 19 10:18:47 it as at the end Jun 19 10:18:51 Was Jun 19 10:18:52 (or just fix your project so it also builds properly with ninja :) ) Jun 19 10:19:02 never thought order is important :/ Jun 19 10:19:04 yeah that's a better fix :) Jun 19 10:19:49 cornel: first things first applies to everybody, except doctor who Jun 19 10:23:17 it seems d.getVar("MAKE") is empty Jun 19 10:24:09 rburton,what was the inherit you were thinking of? Jun 19 10:26:42 * cornel doesn't know doctor who Jun 19 10:27:24 cornel: he citation is "first things first, but not necessarily in that order" - doctor who Jun 19 10:27:35 :) Jun 19 10:34:29 i was wrong Jun 19 10:34:47 MAke has 'make' value but maybe it's not found in the path Jun 19 10:37:32 no advice for my problem :) Jun 19 10:45:49 conf/bitbake.conf:export MAKE = "make" Jun 19 10:45:52 MAKE can't be empty Jun 19 10:46:44 what we're learning here is that maybe oeqa needs a pair of cmake tests, one for ninja and one for make Jun 19 11:49:09 rburton: thank you, gonna try, but i'm afraid our local build tools must be adapted to the new yocto Jun 19 11:52:42 hallo, has anyone tried to populate a sdk from "core-image-minimal" [poky, branch: sumo] image? I get a error that tell the file .../sysroots/x86_64-gaz_sdk-linux/usr/lib/locale is missing. That's true, the hole folder "x86_64-gaz_sdk-linux" is missing. Has this something to do with the "meta/lib/oe/sdk.py"? Jun 19 11:53:30 works for me, what bitbake command are you doing? Jun 19 11:53:42 ueni: my gut feeling is that its related to that "gaz" thing in your machine signature Jun 19 11:55:50 LetoThe2nd: Thank you, i will check this Jun 19 11:56:05 rburton: unfortunately this does not help. I'm afraid our local project specifics must be corrected Jun 19 11:56:26 ueni: i mean, tell us a little about your configuration? MACHINE, DISTRO, ...? Jun 19 11:57:23 rburton: I have an image recipe in my layer that includes the core-image-minimal.bb image. Jun 19 12:01:58 MACHINE: it's a custom machine similar to a Raspberry3 and its derived from the meta-raspberry layer; DISTRO: also custom; That is where the gaz signature come from. Jun 19 12:02:48 ueni: well then, good luck digging Jun 19 12:03:35 The DISTRO contains sume meta data such as maintainer, vendor, sdk-name, and so on Jun 19 12:07:02 rburton: change LICENSE and the recipe doesn't rebuild :( Jun 19 12:07:20 huh Jun 19 12:07:42 rburton: hmm, maybe user error, sorry Jun 19 12:07:47 yes, thankfully Jun 19 12:08:19 that had me worried for a minute Jun 19 12:10:02 rburton: here's the fix for meson that dealt with similar (or maybe even same) issue http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-devtools/meson/meson/0001-Linker-rules-move-cross_args-in-front-of-output_args.patch Jun 19 12:10:43 fun Jun 19 12:10:56 maybe the fix applies, but no longer has desired effect somehow, in any case randy should carefully check the command lines in both cases Jun 19 12:11:17 not sure what is his nick here Jun 19 12:11:21 vmeson Jun 19 12:11:25 vmeson: ^^^ Jun 19 12:12:09 i wonder if vmeson has identified that it is meson with a bisect, or whether the new gcc could be to blame Jun 19 12:12:36 rburton: reverting the meson update could clear that up Jun 19 12:12:39 yeah Jun 19 12:12:56 rburton: in any case ppc support is going away, right? Jun 19 12:13:11 just some bits, or something Jun 19 12:16:17 I wonder if we should elevate risc-v to first class at some point Jun 19 12:16:36 I fully support the idea of open hardware Jun 19 12:16:58 is there a qemu machine for riscv Jun 19 12:17:12 rburton: I think khem had something basic working Jun 19 12:17:34 kanavin_home: Its a question of maintenance and someone willing to step up and sponsor/drive it Jun 19 12:19:55 oh look https://github.com/riscv/meta-riscv/blob/master/conf/machine/qemuriscv64.conf Jun 19 12:21:27 khem is a star Jun 19 12:24:09 rburton: I've put a few things in -next, ok to test that or anything I should add/drop first? Jun 19 12:28:05 oh good you just pulled in most of what i just build tested Jun 19 12:28:46 clean up the bind message to remove the revision log? :) Jun 19 12:29:18 bind is the only one i've not reviewed already Jun 19 12:29:28 so go for it Jun 19 12:29:49 i've got a bundle with some meson tweaks in to sit on top of it too Jun 19 12:30:07 you know it would be nice if the AB had a 'build only' mode for speed Jun 19 12:30:29 rburton: I squashed maintainer removal into your omap ones Jun 19 12:30:36 cool Jun 19 12:30:36 * RP tweaks bind and fires Jun 19 12:31:04 rburton: leaving meson to you for now ;-) Jun 19 12:31:30 all because i discovered libdrm was doing the wrong thing and now i'm writing test suites for meson :) Jun 19 12:32:59 Hi , i would like to build yocto image for qemu cortex a9 machine. there is a layer meta-qemuarma9 (https://github.com/OverC/meta-qemuarma9/blob/master/conf/machine/qemuarma9.conf) which contain the machine conf file for cortex a9 but there is no image recipe for building yocto image. How can i do to inherit from other layer image recipes to build a yocto image? Jun 19 12:35:14 vladzouth, you should be able to use existing images like core-image-minimal Jun 19 12:36:50 Crofton|work: It's not working. i have the following error: "No recipe available for:" Jun 19 12:38:34 Crofton|work: And it lists me the available recipes that is in the layer meta-qemuamra9: linux-windriver_4.8 linux-yocto_4.12 Jun 19 12:50:33 you added the layer to an oe-core build? Jun 19 12:56:09 rburton & LetoThe2nd: thank you for your help, it was an misspelling by selecting the right distro, so that the folder in the sdk wasn't created and error messages was created by sdk.py Jun 19 12:58:30 rburton: do you know if derek the python guy is on this channel? Jun 19 12:58:45 ah, asteriusio I guess :) Jun 19 12:59:00 asteriusio: how is the python 3.6 update thing going? Jun 19 12:59:30 ah, i'll try and remember that nick :) Jun 19 13:00:24 rburton: I guess it'll fall back to me otherwise, but alex's python-sanity project will ruthless drop stuff that I don't undersand or care for :) Jun 19 13:03:42 kanavin_home, please a recipe for scipy :) Jun 19 13:04:06 is there not one already? Jun 19 13:12:55 Is Autobuilder still best for automated yocto build tests ? Jun 19 13:26:16 ftDev: its one option Jun 19 13:26:25 others use jenkins Jun 19 13:26:49 the new autobuilder-helper is builder-agnostic and should let you use any CI thing and reuse logic in autobuilder-helper Jun 19 13:48:30 rburton: Got it thanks. Jun 19 14:06:34 New news from stackoverflow: Yocto - how to create a repository with project specific changes Jun 19 14:36:40 New news from stackoverflow: bitbake: How to include .so build by one recipe into another recipe Jun 19 14:54:14 YPTM- armin is on Jun 19 14:59:06 YPTM: Zoom Meeting: https://zoom.us/j/990892712 Jun 19 14:59:15 YPTM: Stephen Joined Jun 19 14:59:52 what is a 'zoom meeting'? Jun 19 15:00:12 YPTM: Joshua Watt here Jun 19 15:00:29 YPTM: Bruce Ashfield joined Jun 19 15:00:42 sjolley, I can hear you Jun 19 15:01:12 armpit - I can't hear anyone Jun 19 15:01:17 I will dail back in Jun 19 15:01:32 k Jun 19 15:01:58 YPTM: Nick from Singlewire is on the call Jun 19 15:04:06 yes I can hear you Jun 19 15:04:11 sjolley, I can hear you Jun 19 15:04:31 I CAN HEAR YOU CLEARLY NOW! Jun 19 15:04:50 I can see clearly now the rain is gone! Jun 19 15:07:24 YPTM: Matt Hoosier (Garmin) Jun 19 15:07:26 YPTM: Alejandro joined Jun 19 15:08:17 https://docs.google.com/document/d/1Lr8KgkmmCZ84RF_ThIwnT3WWxMv_tUsHkpRV5PXTd0s/edit?usp=sharing Jun 19 15:09:44 yes Jun 19 15:13:20 YPTM: is over Jun 19 15:47:48 Sorry I missed the meeting, need to work on my calendering :/ Jun 19 15:48:29 better setup an exchange server... Jun 19 15:48:39 Jun 19 15:54:43 Hmm.... exchange server bitbake recipe ;) Jun 19 15:56:24 * RP shudders Jun 19 15:56:35 no no... Jun 19 15:56:40 bitbake exchange-server Jun 19 15:57:33 RP, off by one hour problem? Jun 19 15:58:13 Crofton|work: just didn't realise there was a call this week (mid month) Jun 19 15:58:30 Crofton|work: not on the calender so no reminder either Jun 19 15:58:55 RP, we are waiting Jun 19 15:59:01 heh, I just have to fix all the invite times I get Jun 19 15:59:16 RP, actually neither did I.. saw the meeting thing on the channel just before the call Jun 19 15:59:44 (I"ve come to terms with exchange ignoring meeting invites from sjolley.. no idea why it does that.. but it does) Jun 19 15:59:45 clearly we need a session at OEDEM "Calendaring" Jun 19 15:59:58 RP, it was short. You didn't miss much.... not a lot of discussion. Jun 19 16:00:57 Exchange used to perpetually think I was the same person as one of my co-workers.... it was an ongoing joke that we were different manifestations of the same entity Jun 19 17:12:01 hi people. i'm having issues with building an sdk. I want the sdk to include one of my python packages. It works just fine when adding it to DEPENDS when building a normal image, but I can't for the life of me get it to appear in my sdk Jun 19 17:12:31 let me put the recipe on gist.github Jun 19 17:15:22 DEPENDS is not used to add packages to either an image or an sdk, it controls what's build to satisfy build time dependencies Jun 19 17:15:27 https://gist.github.com/wesen/a0ee2892a094a859a43d285eb734bc7f Jun 19 17:15:49 IMAGE_INSTALL is used to add packages to an image, TOOLCHAIN_HOST_TASK and TOOLCHAIN_TARGET_TASK are used to add packages to an sdk Jun 19 17:15:53 yeah, DEPENDS makes the build-time dependency use that python package. the uavcan package isd required as a step build in our cmake build to generate some header files Jun 19 17:15:57 both of which are included in the yocto project documentation Jun 19 17:16:34 also depends on whether you're using a recipe dedicated to sdk construction (meta-toolchain, etc) or using populate_sdk okn an image recipe Jun 19 17:16:46 in the case of hte latter, the sdk starts with content mirroring what went into the image Jun 19 17:16:57 so i tried adding python3-uavcan-native to my TOOLCHAIN_HOST_TASK, but that didn't work Jun 19 17:17:11 i use populate_sdk on a dedicated "sdk" image Jun 19 17:19:58 i added BBCLASSEXTEND = "nativesdk" and use nativesdk-python3-uavcan right now, let's see what it does. i know i'm "pattern matching" to problem solve without really understanding what i'm doing Jun 19 17:25:24 ok, that worked Jun 19 17:32:40 -native can't be added to an sdk, as native isn't pckaged Jun 19 17:32:45 nativesdk is hte right approach, yeah Jun 19 17:34:15 RP: Hmm, wonder if it'd be worth adding support for git bundles to the git fetcher. ie. fetch any bundles, add them to the cloned repo as remotes, fetch them, and then and only then verify SRCREV is now available. would make it a lot easier to ship a good chunk of changes against i.e. an upstream repository without having to git format-patch a ton of patches into SRC_URI Jun 19 17:34:27 * kergoth ponders Jun 19 17:36:47 although, unless ls-remote would be viable, it'd limit options, so perhapps not Jun 19 17:36:57 just a random thought Jun 19 17:37:00 * kergoth gets more coffee Jun 19 17:37:04 Is "bundle" some specific git thing I am unfamilar with? Jun 19 17:37:42 s/thing/term/ Jun 19 17:44:09 JPEW: yes, it's a git command for distributing refs and objects between git repositories via archive files Jun 19 17:45:37 hmm, interesting Jun 19 17:47:02 kergoth: yeah its interesting indeed see https://www.zapcc.com/demo-webkit/ Jun 19 17:47:17 kergoth: combination of distcc+compiler in one Jun 19 17:47:43 9min vs 39mins I will take it anyday Jun 19 17:48:09 khem: Needs to be icecream+compiler ;) Jun 19 17:48:38 oh yes that too Jun 19 17:48:44 (which is probably much harder since it has a host side daemon) Jun 19 19:46:31 kergoth: i don't suppose i can convince you to make bb log work again, i'm too lazy to fix it up but it is so useful ;) Jun 19 19:48:16 agreed, its a good idea Jun 19 19:48:21 and probably trivial to fix Jun 19 19:49:35 yeah Jun 19 19:50:00 bb log? Jun 19 19:50:27 i kind of want bb-contents back too, even though its just a trivial wrapper around oe-pkg-data-util, oe-pkgdata-util list-pkg-files -p foo is awkward Jun 19 19:51:01 fray: just a little script that shows the console log if no args are passed, most recent task log for a recipe if recipe name is passed, or specific task log if recipe+task is specified Jun 19 19:51:32 rburton have you seen the WR live-logging patches? Jun 19 19:51:47 you can tell bitbake to just display the build logs while it's building and such.. Jun 19 19:52:26 fray: yeah saw them demod a while back. in this case i just wanted to see the log after the build though Jun 19 19:52:33 ok Jun 19 19:52:47 kergoth: bb contents works in my repo, it just calls oe-pkgdata-util for you :) Jun 19 19:52:51 https://github.com/WindRiver-OpenSourceLabs/bitbake/commits/wr-10.18-20180611 Jun 19 19:53:02 it's the 6 knotty patches.. (rebased on the 20180611 master version of bitbake) Jun 19 19:56:48 ah nice Jun 19 19:56:53 never did test those Jun 19 21:17:44 fray: hitting an issue where a glibc recipe (not oe-core's) is seeing filerprovides of libc.so(GLIBC_2.25) instead of libc.so.6(GLIBC_2.25), even though SONAME is libc.so.6 according to readelf. any thoughts? Jun 19 21:17:53 * kergoth isn't very familiar with the filedeps bits Jun 19 21:18:26 hmm Jun 19 21:18:48 * kergoth checks hte glibc patches in oe-core for anything applicable Jun 19 21:26:59 thinking Jun 19 21:27:45 check both readelf and objdump show the same SONAME info.. assuming it's right there, I'm not sure why it would display incorrectly through that path Jun 19 21:28:56 Hello everybody! I'm not sure if this is a right place but do we have a plan to upgrade the Python 3 recipe to Python version 3.6 or even 3.7? Jun 19 21:35:24 fray: yep, both show SONAME as libc.so.6. oddly this happens both with glibc-external and glibc-sourcery using this particular toolchain and associated glibc sources, so i'm assuming it's something about this particular glibc branch or how it's being built, but i haven't the slightest clue what's causing it to pick up the wrong name in the first place, so don't really know where to begin :) Jun 19 21:35:29 very odd Jun 19 21:35:33 * kergoth scratches head Jun 19 21:35:56 well i'll review the commits/patches we used on top of 2.27 in case it's one of those, or perhaps just drop them and try building stock 2.27 for comparison.. Jun 19 21:36:03 are you sure that the glibc-extern/sourcery classe/recipes aren't munging them? Jun 19 21:36:06 I know if the past they did.. Jun 19 21:36:35 ...but the last time I actively used them was 4? years ago now... so it's been a while Jun 19 22:41:21 hmm, we have multiple functions in oe.recipeutils that don't work with tinfoil's cookeradapter. that's irritating Jun 19 22:53:03 kergoth: bb.utils.contains_multi is the right thing when I have to depend on more than 1 variable being true or false Jun 19 22:53:11 is that correct ? **** ENDING LOGGING AT Wed Jun 20 03:00:08 2018