**** BEGIN LOGGING AT Tue Mar 27 03:00:03 2018 Mar 27 03:10:25 New news from stackoverflow: Yocto runs only one task inside at a time Mar 27 07:56:41 hello all! I'm trying to blacklist a kernel module in order to disable autoload but i don't have /etc/modprobe.d in the target rootfs, I have missed something but i don't know why :( any idea ? Mar 27 08:05:48 do you have the modprobe command ? Mar 27 08:08:04 if yes you can try to create the folder yourself. Otherwise i like to help you but i'm to inexperienced Mar 27 08:09:16 loulou2: are you pinksnake from yesterday ? Mar 27 08:09:59 @nayfe what ? Mar 27 08:10:51 loulou2: some guy asked same question yesterday Mar 27 08:10:57 @nayfe haha ok it's my coworker nickname ^^ Mar 27 08:11:57 @nayfe i have takenthis part because it's seems stuck :S so i'm trying to add a cmd linearg with modprobe.blacklist=modname, it' building ... Mar 27 08:12:00 maybe ask xilinx ML meta-xilinx@lists.yoctoproject.org ? Mar 27 08:41:30 New news from stackoverflow: Configure yocto to use protocol other than git Mar 27 08:41:49 hello guys, I'm compiling a library using a cutom recipe. During the do_package() I get this error: No GNU_HASH in the elf binary Mar 27 08:42:24 I've add the following: do_compile(){ Mar 27 08:42:25 oe_runmake LDFLAGS=--hash-style=gnu Mar 27 08:42:25 } Mar 27 08:42:44 but can't get rid of it Mar 27 08:50:50 make sure your library accepts LDFLAGS from the envrionment, then this should already work correctly without overriding do_compile() Mar 27 08:51:08 Usually, using a well-established build system is the easiest way to achieve this. Mar 27 09:03:05 @nayfe i sens a msg on yocto mailing list, there is probably an issue somewhere, because of the recipe inherit of kernel.bbclass but "install -d ${D}${sysconfdir}/modprobe.d" seems doesn't call ... Mar 27 09:04:36 neverpanic, solved thanks. Added TARGET_CC_ARCH += "${LDFLAGS}" Mar 27 09:09:55 loulou2 its due to kernel-module-split.bbclass, it removes empty folders Mar 27 09:27:53 @nayfe ok we can simply add blacklist.conf deployement in do_install_append(), that's should do the trick no ? Mar 27 09:32:30 loulou2 can't you upgrade yocto ? KERNEL_MODULE_AUTOLOAD should work on newest versions Mar 27 09:39:51 @nayfe no update possible we have to keep 2.1.2, do you think KERNEL_MODULE_AUTOLOAD is broken in old Yocto version ? Mar 27 09:40:54 loulou2: you received an answer on ML (module_conf_ = "blacklist ") Mar 27 10:11:06 Marex: there? Mar 27 10:29:23 rburton: yes, I see it, work in progress Mar 27 11:17:15 hello guys, I need glib-2.0-dev to compile a recipe, by the way bitbake says: glib-2.0 RPROVIDES glib-2.0-dev Mar 27 11:17:49 so I've added DEPENDS = "gettext glib-2.0 glib-2.0-dev" into my recipe Mar 27 11:18:02 by the way can't still compile Mar 27 11:18:29 fberg: DEPENDS is for recipe names. glib-2.0-dev is a PACKAGE name Mar 27 11:18:30 Any suggestion ? Mar 27 11:19:14 neverpanic: thanks, what am I supposed to do ? I'm not really an expert Mar 27 11:20:01 Add glib-2.0 to DEPENDS Mar 27 11:21:29 done, but the binary glib-gettextize provided by glib-2.0-dev provided by glib-2.0 can't still be found Mar 27 11:32:39 Does your build system try to run glib-gettextize at build time? Mar 27 11:32:47 If so, add glib-2.0-native to DEPENDS Mar 27 11:42:02 New news from stackoverflow: Pulseaudio build issue in yocto Mar 27 12:02:26 @nayfe, thx for info i have replied. But the trouble is probably about driver name Mar 27 12:28:32 neverpanic: thank you, your suggeestion fixed the problem. Mar 27 12:28:59 wht's the difference of using glib-2.0-native instead of glib-2.0 Mar 27 12:29:52 fberg: native is the native recipe which provides binaries you can run at build time Mar 27 12:29:58 such as glib-gettextize Mar 27 12:30:16 also glib-gettextize is deprecated so if this is an upstream you wrote, port it to plain gettext Mar 27 12:30:22 rburton: woha. I'm amused.. Mar 27 12:30:35 thanks Mar 27 13:06:27 Do I need any special switch to force symbolic (relative) links (usr/bin) being copied to recipe-sysroot? Mar 27 13:07:33 I do have image/usr/bin -> erl -> ../lib/erlang/bin/erl Mar 27 13:07:55 but the "erl" itself is not copied to recipe-sysroot for the dependent package Mar 27 13:12:19 New news from stackoverflow: rootfs folder in yocto Mar 27 13:37:47 Hi all, i'm still stuck... Someone here has a good method to prevent kernel modules loading ? I tried load_modules=off but udev still load all modules ... :( Mar 27 13:43:21 Hi, are there tools available to generate a document for GPL/OpenSource License compliance? I'd like to generate something like this: http://legato.io/legato-docs/latest/Legato_16_10_1_OpenSourceLicenses.pdf Mar 27 13:43:21 Im basically looking for something that takes an images license.manifest and makes a nice pdf Mar 27 13:47:49 hi guys, someone knows how to remove a inherited class via bbappend? Is that possible? Mar 27 13:50:39 @igor it's a strange question no? If you don't want to inherit from a recipe you can write your own no? Mar 27 13:59:06 hi i am losing my terminal history, only one session is kept, i haven't been able to find commands to change the history settings, anyone have a guide or can oint me towards the manual for the default shell in a tyocto build? Mar 27 13:59:16 yocto*** build Mar 27 14:01:45 @zazar just to be sure, you system is running on a persistent storage ? Mar 27 14:04:28 loulou2: yes it seems to only keep about 40 commands, HISTSIZE was not set, could that be the issue? Mar 27 14:05:19 hello guys, I'm trying compiling aravis using a custom recipe. During the compile process bitbake return with error: No rule to make target '/usr/bin/g-ir-scanner-wrapper', needed by 'Aravis-0.6.gir'. Stop. Mar 27 14:05:19 from openembedded I see that g-ir-scanner recipe is available from meta-gir. By the way it's pretty obsolete Mar 27 14:05:19 a newer recipes by the way seems to compile g-ir-scanner-wrapper and it is gobject-introspection Mar 27 14:05:19 from gobject introspection recipe: Mar 27 14:05:19 SYSROOT_PREPROCESS_FUNCS_append_class-target = " gi_binaries_sysroot_preprocess" Mar 27 14:05:20 gi_binaries_sysroot_preprocess() { Mar 27 14:05:20 # Tweak the binary names in the introspection pkgconfig file, so that it Mar 27 14:05:21 # picks up our wrappers which do the cross-compile and qemu magic. Mar 27 14:05:21 sed -i \ Mar 27 14:05:22 -e "s|g_ir_scanner=.*|g_ir_scanner=${bindir}/g-ir-scanner-wrapper|" \ Mar 27 14:05:22 -e "s|g_ir_compiler=.*|g_ir_compiler=${bindir}/g-ir-compiler-wrapper|" \ Mar 27 14:05:23 ${SYSROOT_DESTDIR}${libdir}/pkgconfig/gobject-introspection-1.0.pc Mar 27 14:05:23 } Mar 27 14:05:24 Si I've added gobject-introspection into the DEPENDS field of my aravis recipe. by the way bitbake is still complaining that No rule to make target '/usr/bin/g-ir-scanner-wrapper', needed by 'Aravis-0.6.gir'. Stop. Mar 27 14:07:57 @zazar maybe, according to the doc -> So for an infinite history list, make: HISTSIZE=(some number less than 0) Mar 27 14:08:35 @zazar you can try export HISTSIZE=50 HISTFILESIZE=100 and try to save 50 cmds Mar 27 14:10:13 loulou2: it keeps 40-50 commands already, i thought it was only one session but its only a few Mar 27 14:10:48 loulou2: i set HISTSIZE to 100000, will set file size also Mar 27 14:12:17 @zazar 50 was only for example ;) Mar 27 14:13:15 loulou2: i know, but HISTSIZE seems to be set only for one session Mar 27 14:14:03 loulou2: those variables are only for a single session Mar 27 14:14:27 loulou2: what doc? Mar 27 14:15:02 @zazar man bash Mar 27 14:15:28 i tried bash commands and they failed, i thought the shell was ash Mar 27 14:16:21 @zazar ok so maybe ash as a different history managment Mar 27 14:27:32 @loulou2 yes, but I rather change some lines than rewrite the whole recipe. It's easier to maintain Mar 27 14:29:07 well, I just wanted to know if threre is simple way, but I can undo what the class did via bbappend anyway Mar 27 14:29:17 thank you @loulou2 Mar 27 14:36:54 Hi! I'm trying to figure out why our new Yocto/OE-core based OS has a loadavg of about twice that of our old Angström/OE-classic based OS. Any leads? Kernel versions: 2.6.34.14 on classic and 4.9.28 on core Mar 27 14:38:38 RyanMeulenkamp, I saw that on list Mar 27 14:38:42 interesting question Mar 27 14:38:53 try things like perf top and see what jumps out? Mar 27 14:41:15 Nice, thanks! I hadn't heard of that one yet. Mar 27 14:41:41 not sure if you can run that on old image Mar 27 14:43:04 jsut old school top might help Mar 27 14:43:45 I'll give it a try Mar 27 14:45:33 I also tried logging /proc/[pid]/stat[14 + 15] but this gives the opposite result (much higher numbers on the OE-classic OS) Mar 27 14:47:21 that is a kernel thing not an OE things Mar 27 14:50:20 You mean the change in result is caused by the change in kernel? Mar 27 15:00:58 well the files in /proc are directly populated by the kernel Mar 27 15:02:12 pretty thin info, so can only talk in general terms Mar 27 15:15:31 loadavg is probably higher since we have defaults of BB_NUMBER_THREADS and PARALLEL_MAKE based on cpu cores. using the system more thoroughly by default Mar 27 15:15:32 * kergoth yawns Mar 27 15:17:11 kergoth: Can I have the link to that image licensing class you posted earlier? I lost it Mar 27 15:26:52 nm, found it in the archives. I was interested in giving it a try if you can answer some questions Mar 27 15:34:40 sure. it's made with a pretty specific purpose in mind, i was pretty curious on whether anyone would actually use it for anything Mar 27 15:40:15 Ya, was the purpose to have seperate development/release images that allow GPLv3 packages and not respectively? Mar 27 15:42:07 yeah. it's not perfect, but that's the idea. you'd set INCOMPATIBLE_LICENSE = "GPL-3.0", WHITELIST_GPL-3.0 += "gdb", then inherit the class in both images, set ALLOW_ALL_INCOMPATIBLE_WHITELISTED = "0" in the production image, which makes inclusion of gdb in that image fatal. so you can add gdb to the other image but aren't allowed to add it to the production image Mar 27 15:42:42 can also use ALLOWED_INCOMPATIBLE_WHITELISTED to allow specific incompatibly licensed packages in, rather than blocking all Mar 27 15:42:49 New news from stackoverflow: meta-virtualization rocko brach unable to locate package docker Mar 27 15:42:58 How does one get permission to post to 'openembedded-core@lists.openembedded.org'? I tried to subscribe but haven't received a password or any confirmation. Mar 27 15:43:00 mainly a proof of concept. at worst it's a final sanity check for the paranoid Mar 27 15:43:51 * JPEW looking up how we are currently doing that... Mar 27 15:59:29 laplante: if your email is hosted by microsoft then there's a problem where it is rejecting mails, email the admin and you'll get added manually Mar 27 16:01:39 rburton: ah, thank you. that's probably it. is the admin the mailman@lists.openembedded.org address? Mar 27 16:49:47 So I found a bug in the Python makewrappers script in pseudo. Mar 27 16:49:51 Which I've never hit before. Mar 27 16:50:10 Because I've never tried to wrap a "..." function before. Mar 27 16:50:58 > except Exception(e): Mar 27 16:51:00 this should have been Mar 27 16:51:04 except Exception as e: Mar 27 16:51:32 this has been through i don't know how many code reviews, but since we never actually hit a parse failure... Mar 27 17:04:48 Does anyone have an arm64 EABI machine that I could run some sample/test code on? Mar 27 17:10:40 try those ? https://www.scaleway.com/armv8-cloud-servers/ Mar 27 17:10:56 qemuarm64? Mar 27 17:13:13 New news from stackoverflow: Yocto Toaster Compatible image recipes is zero empty Where all recipes? Mar 27 17:25:36 huh. qemu's probably fine, yeah. Mar 27 17:25:47 Wow, renameat2 is horrible to implement. Mar 27 17:25:57 Specifically, the exchange case is hard. Mar 27 17:30:07 seebs: Indeed. Are you actually trying to get pseudo to "do the right thing", or just return ENOSYS? Mar 27 17:32:22 I'm trying to do the correct thing, because so far as I can tell, if renameat2() is actually showing up as a call, it should be expected to work. Mar 27 17:32:44 It turns out I need to add an exchange op, because no existing ops can express an atomic swap of two paths. Mar 27 17:33:20 But, say you rename a to b, and a wasn't in database. pseudo creates a database entry for a, using b's dev/ino, to do the "rename". Mar 27 17:33:54 Ah Mar 27 17:33:55 If you rename a to b and b to a simultaneously, I have to do that with exchanged dev/ino values, and so on. Mar 27 17:33:57 wheeeeeee. Mar 27 18:13:41 hey all. Struggling with building a golang project via bitbake. any diehard go experts? Mar 27 18:26:01 seebs: yeah i noticed that, got a patch here too. i also have a trivial renameat2 wrapper which just does ENOTSUP... Mar 27 18:31:14 I'm staring at the thing, it turns out that I probably have to change the pseudo IPC structure to allow passing more than one dev/inode. Mar 27 18:31:26 You actually *can't* do this safely/reliably without both names *and* both dev/ino pairs. Mar 27 18:31:54 My concern is mostly that I don't think renameat2() should be present in libc and yet returning ENOTSUP. I think. Mar 27 18:32:13 Crofton|work: just saw your message from the other day, probably better to direct PetaLinux stuff to the Xilinx forum (https://forums.xilinx.com/t5/Embedded-Linux/bd-p/ELINUX) ;) Mar 27 18:33:33 Heh, I am no petalinux expert :) Mar 27 18:34:49 seebs: tbh glibc git master doesn't expose it yet so thats mostly academic right now Mar 27 18:36:07 seebs: it uses syscall(renameat2...) if present, but no wrapper is present Mar 27 18:37:42 Crofton|work: neither :), but i think directing petalinux users to the meta-xilinx list would just uncover the thinly veiled cloth protecting them from the wonders of OE :) that they are probably not ready to handle. Mar 27 18:39:13 Yeah, I didn'y know about the other path :) Mar 27 18:39:39 Feel free to correct my answer and we will both get dinged by SO mods for not answering the question Mar 27 18:43:13 Huh. Mar 27 18:43:34 Okay, fair enough, if glibc's not actually exposing it, we can probably avoid doing the thing yet. Mar 27 18:45:13 Hmm. Mar 27 18:45:31 So I have a man page for renameat2() on my system, which is why I assumed it was an actual function I could call. Maybe I should check this out more. Mar 27 18:47:29 ... oh, right. kernel's documenting syscalls, glibc doesn't always provide them. Eww. Mar 27 18:50:30 The deeper this goes, the more I think that article was right.... there should be a liblinux-syscalls (or whatever he called it) Mar 27 18:51:26 na, apps that call syscalls directly should have their maintains drawn and quartered.. and the libc(s) should prioritize adding proper interfaces for syscalls.. :P Mar 27 18:53:11 I mean... I've never used syscall()... *ducks* Mar 27 19:01:52 seebs: last time i fell for that it took me a few hours to discover that lots of projects were copy/pasting the same code from a manpage for a function which glibc never exposed and had a mistake in anyway, so even if they did add that wrapper it wouldn't work Mar 27 19:02:27 ew Mar 27 19:03:16 so, yocto creates linux images, right? like if i were to install windows, do the updates, remove a bunch of crap, and then install it on a bunch of computers, yocto is like a linux version of that? Mar 27 19:03:51 erm, sort of? Mar 27 19:04:37 i'm just sort of trying to understand this conceptually because the guides just sort of dive right in Mar 27 19:05:00 its a tool to build a custom linux-based operating system Mar 27 19:05:28 so, i'm trying to create a linux image for this embedded device we have, so why am i installing ubuntu onto a virtual machine and doing a bunch of stuff on it first in order to accomplish that? Mar 27 19:05:46 no idea, you tell me Mar 27 19:05:55 because the yocto tools don't work in windows? Mar 27 19:06:41 ok, so i'm configuring this ubuntu environment right now on this virtual machine--when does that turn into configuring the environment that i want on the embedded device? Mar 27 19:06:59 never Mar 27 19:07:13 i.e. i'm not sure what stuff is working towards getting yocto working, vs getting the finished image configured Mar 27 19:07:17 The Yocto Project 'build system', is a tool that runs on a Linux OS of your choice.. (which is ubuntu based on your description) Mar 27 19:07:36 from there you would configure a new -custom- Linux distribution that is YOUR embedded Linux distribution. Mar 27 19:07:47 You build it, and then get something out of it that can be deployed into your custom environment Mar 27 19:08:06 so my ubuntu machine is going to download a bunch of packages and things, but not for it--for the custom environment? Mar 27 19:08:10 hello, if i feel like changing the init system to something else, how to make the link /sbin/init to somewhere else? if i do it in FILES_${PN} i have an error about someone else owning the link Mar 27 19:08:18 The Yocto Project build system requires a Linux based operating system to run on.. but it is purely a 'tool'. Ubuntu is only used as an OS to run the tool.. Mar 27 19:08:53 The Yocto Project tooling, running on your Ubuntu OS will download what it needs to compile that applications, that you configured, to create an image to run on your configured target. Mar 27 19:09:27 hastake, usually you would do this by defining a different initscript 'system'. I beleive currently sysvinit and systemd are defined.. you could define your own.. Mar 27 19:09:30 how come people don't just emulate the target hardware, install linux on it, get the stuff they need, and then save that as an image instead? Mar 27 19:10:04 Do you have an emulator for all available hardware? If so I'd like it.. baring that, we do build and use QEMU for some things.. but we're limited to the hardware (and CPUs) that qemu is capable of emulating.. Mar 27 19:10:14 plus that's slow Mar 27 19:10:33 well, what if i have the hardware itself? Mar 27 19:10:34 AbleBacon: the point is yocto is building the linux *you want*. not what ubuntu is, or redhat. Mar 27 19:10:36 There are severe limitations to the emulation on various IA, ARM, and Power platforms. (I think MIPS is fairly well emulated, unless you end up with a variant like the Cavium Octeon) Mar 27 19:10:59 AbleBacon, with some exceptions, USUALLY your host system is much faster at compiling then the target hardware.. Mar 27 19:11:01 AbleBacon: if you have the hardware yourself and eg ubuntu is provided for it and it does what you want already then feel free to use ubuntu Mar 27 19:11:04 so everything is cross compiled. Mar 27 19:11:35 i wouldn't even know how to install linux on this ARM processor in the first place. but yocto takes care of that--right? Mar 27 19:11:56 well, arm is comedy because none of them boot the same Mar 27 19:12:08 which is kind of why you can't just get "ubuntu for arm" Mar 27 19:12:12 If your host system is not already faster -- or you don't want to cross compile -- or as rburton said, you have an existing OS to run on that.. you don't have to do this..... BUT there is another penalty.. reproducibility. The Yocto Project can repduce the same image using the same configuration over and over.. in a way that you can PROVE the sources you have are being used.. this is VERY important from a commercial perspective.. Mar 27 19:12:58 Ya, YP will be able to build the kernel and OS for a known board.. but most likely the install process (for most 32-bit ARM) is so custom, you will have manual steps to deploy it.. the BSP/machine layer usually includes instructions on how to deploy) Mar 27 19:13:00 so if i want to share my linux image with others, i save a small recipe file and send it to them and they can compile it themselves? Mar 27 19:13:17 AbleBacon: yeah. or just share the image. Mar 27 19:13:30 You would share custom (code) content via layers.. configuration either via layers or simply 'here use this configuration', etc.. Mar 27 19:13:45 but absolutely, the layers+image recipe is all they'll need to reproduce what you built. Mar 27 19:14:01 (You can of course share the binary, but again -- in a commercial setting that isn't a good idea for lawyers and such.. they want proof the source you said you used, you used.. so they know what the licensing agreements are) Mar 27 19:14:04 again, useful in commericial environments where you want to do an update two years after shipping something. Mar 27 19:14:08 but they'd have to go in and do some fiddling to get everything set up, even if i told them what layers and recipes to use, right? Mar 27 19:14:27 No, if they use exactly the same layers and configuration as you do -- there is nothing further to 'setup'. Mar 27 19:14:37 Remember the Yocto Project builder is a tool that runs on your host.. Mar 27 19:14:50 and the host system must be a Linux based host.. (with a few other requirements) Mar 27 19:15:01 how would i send them the layers and configuration? is there like a makefile sort of thing? Mar 27 19:15:10 it just happens to be that a lot of people do this on Ubuntu.. but that isn't required.. it can be CentOS, Arch, RHEL, SuSe, even some Yocto Project based systems Mar 27 19:15:25 in your project directory you will have a conf subdirectory.. That is your configuration. Mar 27 19:15:54 the layers referred to by the configuration are almost always git repositories. Use whatever method you wish to share your git repositories (an current commit) with others Mar 27 19:16:08 (the build log always starts with a list of layers and the commit they are on for tracking purposes.. Mar 27 19:16:17 so the conf directory is everything... assuming another user has checked out the same version of yocto, if i send them my conf directory, their resultant image will be an exact match of mine? Mar 27 19:16:24 yes Mar 27 19:16:54 so why not have a single file that contains all of the information about the build? like how a makefile might work Mar 27 19:17:14 i feel like i could get going with an XML sort of deal a lot faster Mar 27 19:17:28 bitbake (the utility you are running) is roughly an equivalent to 'make'.. but way way more extensible and (for the purposes of this build) better able to handle the demands of the system. Mar 27 19:17:52 There are various (not YP) programs that can take YAML, XML, or others and generate the .conf files.. Mar 27 19:18:00 but the .conf format ont he Yocto Project is fixed and has a history.. Mar 27 19:18:06 AbleBacon: probab;ly easiest if you just try a build Mar 27 19:18:14 so if i want to have, say, GCC 6.10 on my system, i need a... layer for that? Mar 27 19:18:18 there is a learning curve.. Mar 27 19:18:54 the 'meta' layer (a.k.a. oe-core) contains the basics for the core system.. meta-yocto contains various stock distribution configurations.. and meta-yocto-bsps contains various sample BSPs.. Mar 27 19:19:04 there are additional layers you can download (see layers.openembedded.org) Mar 27 19:19:21 As far as GCC, if you want to use GCC.. then you would most likely use the one from oe-core (included) Mar 27 19:19:28 AbleBacon: you generally get the gcc that comes with the version of yocto you use. changing compiler is *not easy* Mar 27 19:19:39 oh excellent! i was wondering how to find the specific thing i need for this chip Mar 27 19:19:40 if you wanted to use some alternative version, you may have to implement something yourself.. this is definitely an 'advanced' operation Mar 27 19:20:07 search for your COTs dev system on layers.openembedded.org.. be sure to swithc the 'branch' to whatever branch of the YP you are using.. Mar 27 19:20:17 (Rocko is the last released version).. this changes every 6 months... Mar 27 19:20:44 if you truely have something 'new', then you will need to learn how to do multiple parts of the system -- but usually you can start with one of the QEMU bsps, and then adjust from there Mar 27 19:22:33 so the ones on that site are the open source ones? i'm seeing my manufacturer has their own BSP, but it's not listed on that site Mar 27 19:24:00 then your manufacturer has not published it to the index -- or they only support it for a different version. You may have to look at older releases. Mar 27 19:24:26 Rocko is current, but it's not unusual to support latest (master) or ancient versions (way past support) Mar 27 19:24:52 using a several year old version is a bad idea. No good source for updates Mar 27 19:24:57 httpstr://wiki.yoctoproject.org/wiki/Releases Mar 27 19:25:14 ... https://wiki.yoctoproject.org/wiki/Releases ... that gives you an idea of how old something is by the code name (branch name) Mar 27 19:25:33 so if they're mentioning the "krogoth" release in this documentation, i should not try to use rocky? Mar 27 19:25:34 anything older then 1 year should not be used.. and even something as old as one year should be carefully considered.. Mar 27 19:25:41 Krogoth is 2.1, April 2016 Mar 27 19:26:00 the Yocto Project does community (best effort) support for about 1 year.. so that is no longer supported after April 2017 Mar 27 19:26:21 but if i try to use a newer version, it definitely won't be compatible? Mar 27 19:26:24 (if you have a commercial provider, they certainly can and do extend support longer then 1 year.. but the community generally does not) Mar 27 19:26:39 most likely it will not be directly compatible -- which means you will have to port the kernel and/or other pieces (say an SDK).. Mar 27 19:26:57 if it's a common board, others may have already done it.. if it's not a common board.. beat up the board maker Mar 27 19:27:42 yeah it's like an ancient board i'm pretty sure Mar 27 19:27:58 pretty cool, tho. size of a coaster. technology these days! Mar 27 19:28:24 You can use krogoth, and it will probably work.. but it will be harder to ask general questions when you have a problem.. Mar 27 19:28:43 if this is just a hobbiest system.. that may be good enough for you (especially if you are just learning) Mar 27 19:29:20 eh, not exactly--it's a class 3 medical device :-/ Mar 27 19:29:54 for that, I'd definitely recommend either beating up the board vendor for something newer -- or finding a commercial provider.. Mar 27 19:30:27 (commercial providers often have 3, 5, 10, or even 15 year support... plus can advise some additional options when working in the medical or similar spaces) Mar 27 19:30:55 (i.e. from a system licensing perspective, medical devices often can't allow an end user to modify them -- but the GPLv3 license may require that.. which often means don't use GPLv3 software..) Mar 27 19:31:14 these are things that the Yocto Project can help you do, that other systems (like Ubuntu) can not.. Mar 27 19:31:30 (nothing wrong with Ubuntu, it's just not built for that environment) Mar 27 19:31:47 right--i wasn't thinking i could use it on here, but i was confused about why i needed it on a virtual machine. makes sense now though Mar 27 19:31:50 Commercial OSVs, as well as consultants and such often have a lot of experience in these areas to help you get started or support you long-term Mar 27 19:32:32 (I work for an OSV, as do others on this channel -- as well as semi makers, ISVs and consultants...) Mar 27 19:32:50 what are OSVs and ISVs? my google is failing me Mar 27 19:33:42 OSV -- Operating System Vendor.. ISV - Independent Software Vendor (generally applications, sometimes more) Mar 27 19:33:57 Semi's (the guys who make the CPUs).. board vendors (you already know that).. etc Mar 27 19:33:58 ah Mar 27 19:34:16 yeah, we got vended some of this qt framework stuff from one of those Mar 27 19:34:18 OSV - Wind River (who I work for), MontaVista, Mentor Graphics, Enea and many others.. Mar 27 19:34:46 I'd consider QT software to usually be an ISV.. Mar 27 19:34:54 right Mar 27 19:37:39 i should probably just turn off my brain and follow these instructions and see where it gets me Mar 27 19:37:52 ya.. run through it first that way.. check out what you get.. Mar 27 19:38:11 I think you will have a better idea of difficulty (or ease of use) at the end.. and then can focus questions on what ya need.. Mar 27 19:43:42 New news from stackoverflow: Yocto bitbake ninja: command not found Mar 27 20:45:55 I wan't to change /etc/init.d/lighttpd file Mar 27 20:46:27 but it's in poky. I could modify poky but changes to poky is not tracked by our VCS Mar 27 20:46:39 how do I go about doing this? Mar 27 20:53:42 noway96: You could make a new layer and change it in a .bbappend file Mar 27 20:53:52 (or in one of your existing layers) Mar 27 21:11:20 when building a golang app it can't find all.bash file, although some go stuff is in there. most of it,actually... im wondering which package has it... Mar 27 21:42:21 very general question. how can i reference `cortexa8hf-neon-linux-gnueabi/golang/1.8.3-r0/go/` directory from any recepie? Mar 27 21:42:53 you can't, thats the per-recipe workdir Mar 27 21:43:25 kergoth: first, thank you for your reponse. appricate it Mar 27 21:43:52 kergoth: then, how can i get same contents that i have in this dir into my recipe? Mar 27 21:44:24 in `cortexa8hf-neon-linux-gnueabi/telegraf/1.4/telegraf-1.4 ` for example Mar 27 21:47:34 kergoth: i have a feeling im having a wrong dependency or inheritance, but not quite sure what exactly. I was able to build other golang project with very similar settings, it just didn't use gdm Mar 27 21:50:09 fray: thank you. is there any docs or pointer on how to define your own initscript? Mar 27 21:55:39 i did find a patchset for openrc, but it seems to change things in oe and poky, it's not obvious how to change that in top layers only Mar 27 22:06:52 what if instead of a recipe file to create the root filesystem, i have a tar file that i want to use instead? Mar 27 22:41:55 where can I find all the drivers that my image uses? **** ENDING LOGGING AT Wed Mar 28 03:00:01 2018