**** BEGIN LOGGING AT Tue Jun 30 02:59:57 2020 Jun 30 06:48:31 Has anybody tried building python in Yocto warrior. I get the errors - Jun 30 06:48:32 errors found, failing task. Jun 30 06:48:33 (/home/shreyas/temp/custom_yocto/build-custom-zynq/../poky/meta/recipes-devtools/python/python3_3.7.6.bb:do_configure) failed with exit code '1' Jun 30 06:48:34 Previously configured separate build directory detected, cleaning /home/shreyas/temp/custom_yocto/build-custom-zynq/tmp/work/cortexa9t2hf-neon-poky-linux-gnueabi/python3/3.7.6-r0/build Jun 30 07:04:17 Guys, I am trying to add meta-java layer and everytime it comes up with the following error message: Jun 30 07:04:19 ERROR: glibc-2.30-r0 do_package: SYSTEMD_SERVICE_nscd value nscd.service does not existERROR: Logfile of failure stored in: /opt/graylog-poky/build/tmp/work/core2-64-poky-linux/glibc/2.30-r0/temp/log.do_package.712ERROR: Task (/opt/graylog-poky/meta/recipes-core/glibc/glibc_2.30.bb:do_package) failed with exit code '1' Jun 30 07:04:44 Where as nscd is already installed. Jun 30 07:24:15 Srijan, what is the error the log file is highlighting ? Jun 30 07:29:15 Srijan, kroon seems like the nscd.service file is missing Jun 30 07:30:45 Yeah it seems the nscd.service file is missing. Jun 30 07:31:59 I have added CORE_IMAGE_EXTRA_INSTALL_append = " nscd" and it gets installed without issues. But not via the meta-java layer Jun 30 07:36:47 If I remove the meta-java layer, everything gets built without issues. Jun 30 07:41:31 * Letothe2nd is manning the booth for cases of emergency. Will direct folks mostly to slack Jun 30 07:45:10 Srijan, how're trying to install nscd from meta-java? Jun 30 07:47:58 I add meta-java using bitbake-layers add-layer and then try to do bitbake core-image-full-cmdline, it it here I get the error Jun 30 07:51:47 I am not trying to add nscd, it is a dependency for glibc I presume...I am trying to add meta-java as a layer. And then when I do a bitbake core-image-full-cmdline, I get the error. Jun 30 07:56:20 can you try a clean build for glibc? keeping the meta-java layer added Jun 30 07:58:26 Ok Jun 30 08:10:31 So I did a bitbake -c cleanall glibc and then ran bitbake glibc...glibc was installed successfully. Now I am trying bitbake core-image-full-cmdline to check it the issue persists or if it has been resolved. Jun 30 08:11:23 Hi Jun 30 08:11:56 Now when I did a bitbake core-image-full-cmdline, I get a whole lot of error messages: Jun 30 08:12:03 ERROR: linux-yocto-5.2.17+gitAUTOINC+bb2776d6be_25b14cdf96-r0 do_package: Error executing a python function in exec_python_func() autogenerated:The stack trace of python calls that resulted in this exception/failure was:File: 'exec_python_func() autogenerated', lineno: 2, function: 0001: *** 0002:split_kernel_module_packages(d) Jun 30 08:12:04 0003:File: '/opt/graylog-poky/meta/classes/kernel-module-split.bbclass', lineno: 171, function: split_kernel_module_packages 0167: # avoid warnings. removedirs only raises an OSError if an empty 0168: # directory cannot be removed. 0169: dvar = d.getVar('PKGD') 0170: for dir in ["%s/etc/modprobe.d" % (dvar), Jun 30 08:12:04 "%s/etc/modules-load.d" % (dvar), "%s/etc" % (dvar)]: *** 0171: if len(os.listdir(dir)) == 0: 0172: os.rmdir(dir) 0173:} 0174: 0175:do_package[vardeps] += '${@" ".join(map(lambda s: "module_conf_" + s, (d.getVar("KERNEL_MODULE_PROBECONF") or "").split()))}'Exception: FileNotFoundError: [Errno 2] No such file or Jun 30 08:12:05 directory: '/opt/graylog-poky/build/tmp/work/genericx86_64-poky-linux/linux-yocto/5.2.17+gitAUTOINC+bb2776d6be_25b14cdf96-r0/package/etc/modprobe.d'ERROR: Logfile of failure stored in: /opt/graylog-poky/build/tmp/work/genericx86_64-poky-linux/linux-yocto/5.2.17+gitAUTOINC+bb2776d6be_25b14cdf96-r0/temp/log.do_package.16472ERROR: Task Jun 30 08:12:05 (/opt/graylog-poky/meta/recipes-kernel/linux/linux-yocto_5.2.bb:do_package) failed with exit code '1'ERROR: volatile-binds-1.0-r0 do_package: SYSTEMD_SERVICE_volatile-binds value var-volatile-lib.service does not existERROR: Logfile of failure stored in: Jun 30 08:12:06 /opt/graylog-poky/build/tmp/work/all-poky-linux/volatile-binds/1.0-r0/temp/log.do_package.16475ERROR: Task (/opt/graylog-poky/meta/recipes-core/volatile-binds/volatile-binds.bb:do_package) failed with exit code '1'ERROR: glibc-locale-2.30-r0 do_package: Error executing a python function in exec_python_func() autogenerated:The stack trace of python Jun 30 08:12:06 calls that resulted in this exception/failure was:File: 'exec_python_func() autogenerated', lineno: 2, function: 0001: *** 0002:package_do_split_gconvs(d) 0003:File: '/opt/graylog-poky/meta/classes/libc-package.bbclass', lineno: 189, function: package_do_split_gconvs 0185: dot_re = re.compile(r"(.*)\.(.*)") 0186: Jun 30 08:12:07 0187: # Read in supported locales and associated encodings 0188: supported = {} *** 0189: with open(oe.path.join(d.getVar('WORKDIR'), "SUPPORTED")) as f: 0190: for line in f.readlines(): 0191: try: 0192: locale, charset = line.rstrip().split() 0193: except Jun 30 08:12:07 ValueError:Exception: FileNotFoundError: [Errno 2] No such file or directory: '/opt/graylog-poky/build/tmp/work/core2-64-poky-linux/glibc-locale/2.30-r0/SUPPORTED'ERROR: Logfile of failure stored in: /opt/graylog-poky/build/tmp/work/core2-64-poky-linux/glibc-locale/2.30-r0/temp/log.do_package.16469ERROR: Task Jun 30 08:12:08 (/opt/graylog-poky/meta/recipes-core/glibc/glibc-locale_2.30.bb:do_package) failed with exit code '1' Jun 30 08:13:17 The logs are here, sorry for pasting it in the channel....https://pastebin.com/fcDDihwA Jun 30 08:14:18 Hi, which tool is the standard on yocto-Linux to create an incremental backup against a factory reset snapshot version? Jun 30 08:14:43 fbre: there is none. Jun 30 08:15:13 ...or maybe a good hint which one is useful :-) Jun 30 08:15:48 fbre: other than "it all depends"? Jun 30 08:16:18 fbre: from rumours, i would judge that some overlayfs construction is an appraoch that many are taking these days. Jun 30 08:17:23 overlayfs? Is this a term I can google for further details? Jun 30 08:17:30 fbre: hum, try and find out? Jun 30 08:19:31 OK, thanx! I think this is a good trigger for more precise investigations *thumbsup* Jun 30 08:19:57 fbre: but seriously, the question is both very specific while at the same time lacking about all necessary content, so that pointer is probably all one can give. Jun 30 08:21:06 I am tying to add a new recipe for adding a new lib and facing the below error while baking the new recipe Jun 30 08:21:32 ```| ../sysdeps/arm/start.S:119: error: undefined reference to 'main'| collect2: error: ld returned 1 exit status| make: *** [all] Error 1```\ Jun 30 08:22:04 Well, the first thing is always to look at the community if there is a standard way everyone uses :) Jun 30 08:22:36 viswa: that sounds very strange, as it seems to refer to baremetal startup code Jun 30 08:22:59 yep Jun 30 08:23:45 any reference links for adding a lib Jun 30 08:23:47 ?? Jun 30 08:23:51 viswa: well then the usual techniques don't apply, as the recipes are meant to build things for linux applications. Jun 30 08:24:28 viswa: what is this library meant for then? this is certainly not a classic yocto build. Jun 30 08:25:09 viswa: did you hand-code the makefile? Jun 30 08:25:17 working on easy-mesh lib for rdk platform Jun 30 08:25:29 rburton: i guess we rather have a conceptual problem here. Jun 30 08:25:49 like, trying to apply a baremetal mindset to a yocto build. Jun 30 08:27:09 @xtron - If i remove the meta-java layer, everything builds perfectly fine. Jun 30 08:28:18 Srijan: send patches to meta-java then. it has always been problematic and is basically kept alive by people fixing it. Jun 30 08:29:05 Ok. Jun 30 09:34:06 hello, I am having trouble building a C99 CMake project via bitbake (problem locating pthreads) Jun 30 09:34:10 Could NOT find Threads (missing: Threads_FOUND) Jun 30 09:34:39 does compile locally on Ubuntu 18.04 x86 PC Jun 30 09:34:49 bu not via bitbake Jun 30 09:35:49 the usual find_package(Threads REQUIRED) approach is used Jun 30 09:36:23 target_link_libraries(app PRIVATE Threads::Threads) Jun 30 09:37:02 I am using Yocto 3.1 Dunfell Jun 30 09:37:26 eduardas: sounds like you'll need to dig into the logs by cmake itself, they should be in the build directory of your package in the work structure. Jun 30 09:39:18 Letothe2nd: log hardly gives anything except Could NOT find Threads (missing: Threads_FOUND) Jun 30 09:40:11 eduardas: well at least autotools provide soemthing like config.log under the hood, which tells how it treid to do things. you'll need to find the cmake equivalent, methinks. Jun 30 09:41:45 eduardas: is pthread headers/lib found in WORKDIR/recipe-sysroot of the recipe complaining? If no, missing DEPENDS. If yes, probably CC, LDFLAGS or CFLAGS being overridden in your Cmake? Jun 30 09:42:17 qschulz: shouldn't pthreads come as an implicit depends thourhg the glibc? Jun 30 09:42:34 I'm getting this weird error ever since upgrading to dunfell, "ERROR: coreutils-8.31-r0 do_package_qa: QA Issue: /usr/bin/stdbuf contained in package coreutils-stdbuf requires /usr/bin/coreutils, but no providers found in RDEPENDS_coreutils-stdbuf? [file-rdeps]" Jun 30 09:43:17 And I seem to be alone, can't find any references on any mailing list or google search, can anybody point me in the correct direction in trouble-shooting it? Jun 30 09:43:31 hi all Jun 30 09:43:34 sh00p: reproductible with pure poky? Jun 30 09:44:12 Letothe2nd, well I haven't made any patches to poky if thats what you mean Jun 30 09:44:34 sh00p: no, rather without any layers on top that might mess with it via appends or such Jun 30 09:44:52 Letothe2nd, ah I see, that I haven't tried Jun 30 09:45:10 sh00p: also, see: http://cgit.openembedded.org/openembedded-core/commit/meta/recipes-core/coreutils?id=992cec44ad2073cfe05cb2c58936032af189748a Jun 30 09:45:11 sh00p: e.g. a bbappend on coreutils in a layer :) Jun 30 09:45:44 Letothe2nd, yeah I found that commit, and it says he moved coreutils-stdbuf to an own package, but I can't find the bb for it Jun 30 09:45:45 Letothe2nd: re pthread, does not hurt to check. Don't know where it comes from :) Jun 30 09:46:08 sh00p: there isn't a specific bb for it, it's just a separate package in the same recipe Jun 30 09:46:18 qschulz: nope certainly doesn't hurt. Jun 30 09:46:43 magicmoon: hi Jun 30 09:47:06 qschulz, okay I see, so possibly somewhere some rogue recipe is modifying my coreutils Jun 30 09:47:11 anyone working with meta-ros? Jun 30 09:47:31 magicmoon: nope, but we have metaquestion experts galore! Jun 30 09:48:27 sh00p: bitbake-layers show-overlayers, or show-appends might help Jun 30 09:51:41 hmm i am running into trouble when trying to compile ros sources with catkin_make... roscpp seems not be confiugred correctly... Jun 30 09:51:42 show-overlayed sorry Jun 30 09:52:17 magicmoon: no error log, no help to expect from us :) Jun 30 09:52:34 where to post ? here? Jun 30 09:53:00 i don't want to get banned :D Jun 30 09:53:12 qschulz, nope no coreutils overlaying listed... but I did found some suspicious lines when grepping, "meta-pil/conf/distro/basic.inc:72:PACKAGECONFIG_append_pn-coreutils = ..." Jun 30 09:53:23 magicmoon: some pastebin of your least distrust Jun 30 09:53:33 sh00p: bingo, thats it. Jun 30 09:53:52 magicmoon: pastebin somewhere on the internet :) Jun 30 09:55:34 https://pastebin.com/GHNsSKGv Jun 30 09:56:30 magicmoon: see, you didn't say you were using an SDK :) Which is a very important piece of information to give :) Jun 30 09:56:45 i was going to tell Jun 30 09:57:17 i sourced it of coarse Jun 30 09:58:48 * Letothe2nd would bet on the sdk being somewhat incomplete. but then, 2.3 is also massively outdated, and all that. Jun 30 09:59:25 I'm not knowledgeable on SDK stuff so just wild guessing now. Have you added roscpp to your SDK? Have you checked the searched file is available somehwere in one of the SDK sysroot? Yes? Is the path passed to your cmake correct? Jun 30 09:59:48 and i would advise to avoid the SDK at all cost for stuff liek that. Jun 30 09:59:48 Letothe2nd: true, don't know why people are sticking to old unmaintained versions Jun 30 09:59:59 qschulz: because industry best practise. Jun 30 10:00:03 * qschulz hides (using 2.6.4) Jun 30 10:00:14 where can I find the "env" definition in Yocto? Jun 30 10:00:33 I mean python object "env" Jun 30 10:00:49 magicmoon: what exactly is your usecase? Why are you trying to use the SDK? (devtool or devshell for example might be a good tool depending on the usecase) Jun 30 10:01:10 i have a yocto build for a stereocamera Jun 30 10:01:35 * Letothe2nd has a look, and concludes that if he ever is in for massive pain, he'll give that meta-ros a try. Jun 30 10:01:36 and i need to integrate ROS Nodes into it Jun 30 10:02:15 "It is no longer under active development and is not being tested, but will be retained until it is no longer needed by the community. The only criterion for acceptance of a pull request for it is that it's based off its current HEAD and merges cleanly." Jun 30 10:02:16 at the moment the camera gives out unrectified pictures... so i need to get the algorithms on the platform Jun 30 10:03:01 i want to use the sdk to build my ros nodes on my build system and then install them afterwards on the yocto linux Jun 30 10:03:46 the yocto is pyro because of meta-ros Jun 30 10:04:36 magicmoon: devtool with a basic recipe should do the trick, since anyway you'll probably want to integrate it to your Yocto image :) Jun 30 10:05:23 the point is that you won't be using the SDK, which isn't as widely used as... well Yocto itself. And you don't lose time fixing something that you anyway don't need that much Jun 30 10:05:33 qschulz: ++ Jun 30 10:06:00 qschulz: thank you for the advice. It really helped. Ultimately it seems my problem was overwriting CFLAGS in the CMake project. Was able to fix it. Jun 30 10:07:30 Two questions: 1) where can I find yocto "env" python object definition? 2) is there anyboddy who is building Node modules with Dunfell? Jun 30 10:08:22 eduardas: a sadly common mistake :) (Been there :) ) Jun 30 12:42:17 zeddii: around? Jun 30 12:42:30 yup. Jun 30 12:43:16 trying to build perf, but the virtual/kernel is a linaro kernel if that makes a difference. failing with: Jun 30 12:43:17 | /usr/src/debug/perf/1.0-r9/perf-1.0/tools/perf/util/srcline.c:200: undefined reference to `bfd_get_section_flags' Jun 30 12:43:22 anything obvious? Jun 30 12:43:30 yah. I know the fix for that. there's an upstream one. sec. Jun 30 12:43:40 s/one/patch/ Jun 30 12:43:42 awesome, thanks Jun 30 12:43:52 i'll buy you a beer this evening at ELC Jun 30 12:44:00 :D Jun 30 12:44:40 make sure this patch is the kernel: https://github.com/torvalds/linux/commit/0ada120c883d4f1f6aafd01cf0fbb10d8bbba015 Jun 30 12:45:17 perfect, thanks Jun 30 13:01:55 rburton: re: the docker go build issue. how exactly do we specify to use go-cross ? Those recipes have always just inherited 'go' and I can't find an example. Jun 30 13:07:20 zeddii: i thought it just worked Jun 30 13:07:29 it is supposed to. yes. Jun 30 13:07:42 but all of these 32bit arm go recipes are blowing up now. Jun 30 13:08:43 "inherit go" and it sets up everything else. I've commented out bits of the recipe that were setting the same as the defaults, and it still blows up. So I don't think it is purely the recipes being of an older style. Jun 30 15:22:44 hi all, really silly beginner question, if I've run a `bitbake core-image-minimal` *where* does that image go? The build appeared to succeed (880/2200 recipes ran, exit status was '0') but I just can't find where the new firmware I just built is? Jun 30 15:23:09 leehambley: build/tmp/deploy/images/$YOURMACHINE/images :) Jun 30 15:23:33 leehambley: and let me warmly recommend https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj :) Jun 30 15:23:54 ah , last /images was probably superfluous Jun 30 15:25:18 cool looking you-tube series, thanks! Is there some "copy firmware to a real directory" bitbake command I'm missing, it seems weird that build artifacts go to build/tmp! Jun 30 15:25:33 leehambley: what is unreal about that directorY? Jun 30 15:26:03 leehambley: no, it makes perfect sense: all artifacts are considered temporary as they can be recreated upon the desire to do so. Jun 30 15:26:24 just seems strange that the thing I built intentionally is in a tmp/ directory, if I think about a proberbial "make ..... right, your 2nd message arrived before I was finished thinking, but yeah, makes good sense :) Jun 30 15:26:41 :) Jun 30 15:27:42 maybe my other question is not 100% yocto related, but I'm kinda going in at the deep-end here. (background, fairly experienced linux developer, but embedded is new to me) Jun 30 15:27:43 I'm using an i.MX6 ULL module from MYiR-Tech, they provide BSP and most of a working Yocto build, etc - all works great (I'm about to flash the firmware I built from the CDROM sources they provided) Jun 30 15:28:33 the question I have, ... for configuring device pinouts, the .dts file, doesn't change, right? Seems like most of the pinout configuration happens at kernel build time, via options in menuconfig like... enabling SPIdev? (this is one of those typical SOMs that makes you choose between sdio/spi because the pins are reused) Jun 30 15:28:48 mope, thats all dt Jun 30 15:29:26 kconfig enabling it just builds the drivers/software. the DT then connects it to the real (HW) world Jun 30 15:29:28 but I guess I'm asking, do I need to go find the NXP pinout config tool and generate something new that goes into the bsp recipe... seems like not, right, device tree has a _complete_ description of all the hardware onboard, and I can activate anything that doesn't conflict through kernel config Jun 30 15:29:48 nope. Jun 30 15:29:53 ah, so not _exactly_ menuconfig (I know, I for e.g I have no userspace spidev on thieir firmware) Jun 30 15:30:03 leehambley: pinout configuration is done in the Device tree. Then you compile in (or compile kernel modules) through menuconfig Jun 30 15:30:18 but, awesome, thanks for the help I'll go dig more into how to actually activate/configure stuff based on DT Jun 30 15:30:33 qschulz: kernel level is your trade :) Jun 30 15:31:05 haha, yeah, I'd prefer writing a kernal module for this, userspace spidev seems like asking for trouble, I'd like to have a "file" I can read with my sensor data in, and just make the kernel deal with the details, but .. babysteps, you know :) Jun 30 15:31:23 but then the .dts for this board, from the manufacturer is "complete" and doesn't get modified to enable/disable hardware.. dt will get configured, somehow once I learn how to do that, and it'll reference the parts of the .dts file Jun 30 15:31:41 leehambley: see iio :) Jun 30 15:31:52 leehambley: maybe your spi device is already supported. Otherwise, if it's a rather simple spidev, reading the documentationin the linux kernel for the device tree bindings should jsut give you enough :) Jun 30 15:32:23 thanks, tbh my spi device is just a ADXL345 so .. it's a line protocol of IMU data Jun 30 15:32:27 it's about as 'unixy" as it gets Jun 30 15:32:37 leehambley: yeah, there's a bunch to learn about and #yocto isn't really the good place to get good infomration :) Jun 30 15:32:59 leehambley: probably using IIO subsystem as Letothe2nd said then Jun 30 15:33:28 :) understood @qschulz - just trying to find out what's the "right" layer for this stuff, the pointers to where to find my firmware, and "confirmation" that I don't need to use NXPs gui tool already saves me a huge amount of work <3 Jun 30 15:33:28 leehambley: I can suggest this training: https://bootlin.com/training/kernel/ Jun 30 15:34:29 (free access to all slides/labs etc.. without a trainer, you pay for in-person (or remote) training. I was an employee there so biased but I don't earn anything for this ad :) ) Jun 30 15:35:22 haha, that training looks like amazing value for money, I'll definitely consider that Jun 30 15:35:48 a week off "work"( yocto/embedded is a hobby thing with a friend one evening every week at the moment) working on a linux course sounds like a great week off :) Jun 30 15:39:09 zeddii: I confirmed that beaglebone build works after your patches, fails before on f32, thanks Jun 30 15:39:34 leehambley: (they have other trainings, specifically targetting the yocto one :) but it's presenting the basics for a lot of stuff and not going deep into details so might not fit everyone's skills/desires) Jun 30 15:39:39 cool. I'll still keep my debug patch handy, but that means we don't need any flag gymnastics Jun 30 15:40:52 zeddii: can you remote print in marks office? :P Jun 30 15:41:07 honestly, I already wondered that. Jun 30 15:41:16 zeddii: GOGOGO! Jun 30 15:41:36 if the whiteboard was a fancy smartboard, that would be awesome. Jun 30 15:41:50 more than awesome. Jun 30 15:42:24 his slides had terminals, I looked, but couldn't glean any routing info to try and get into the office :P Jun 30 15:42:42 dang Jun 30 15:56:21 * kergoth yawns Jun 30 16:04:49 zeddii, you trying to hack fray ? Jun 30 16:04:57 of course! Jun 30 16:05:03 super classy! Jun 30 16:05:29 he's doing a presentation, and there's a printer in view. would be awesome if it started spewing pages. Jun 30 16:06:07 a 1990s postscript fail, where you get 500 pages with one char on each. Jun 30 16:06:26 absolutely. staircase to hell. Jun 30 16:21:04 zeddii: evil :) Jun 30 16:21:39 RP - ask zeddii what I did to his office phone one day.... Jun 30 16:22:06 *that* was evil. B-) Jun 30 16:24:04 paulg: I'm sure I've heard of this before :) Jun 30 16:30:52 tgamblin: do I take alex's revert? Just want to be sure I understand your reply? Jun 30 16:34:20 RP: yeah, his revert is fine. I looked at why my patch wouldn't apply to setuptools, and it was because their devs had made changes to do a less-aggressive version Jun 30 16:34:44 Still a significant improvement Jun 30 16:34:48 tgamblin: ok, cool. Just want to ensure we're all on the same page Jun 30 17:20:34 is it possible to a do_install_myfeature and expect the function to run if myfeature is present in MACHINE_FEATURES (or anywhere ...) ? I'm trying to understand the how much I can push the limit of do_install. Jun 30 17:23:21 What's the magic to get utf-8 locale in my image? Jun 30 17:24:20 I have GLIBC_GENERATE_LOCALES = "en_US en_US.UTF-8" and IMAGE_LINGUAS = "en-us" in my machine conf, but locale -a only lists C and POSIX Jun 30 17:38:33 fullstop: can you do a find in build/tmp/deploy/ for "*locale-base*" Jun 30 17:39:16 khem: I was just doing that.. I now have en_US and en_US.ISO-8859-1, but no utf8 Jun 30 17:39:45 There were some glibc-binary-localedata-* packages Jun 30 17:39:58 aleblanc: do_install_myfeature will override standard do_install if myfeature is in OVERRIDES so please ensure if its in yur override list Jun 30 17:40:57 There is no glibc-binary-localedata with utf8 encoding.. hmm. Jun 30 17:43:38 do you have LOCALE_UTF8_IS_DEFAULT set ? Jun 30 17:45:31 Wait.. Jun 30 17:45:39 they are actually utf-8 even if not specified Jun 30 17:46:07 right so your default is utf-8 I guess Jun 30 17:46:16 what is GLIBC_INTERNAL_USE_BINARY_LOCALE set to ? Jun 30 17:47:43 looking.. waiting for bitbake.. :-) Jun 30 17:48:08 and perhaps you want to set ENABLE_BINARY_LOCALE_GENERATION="1" Jun 30 17:48:20 then it will build them from source during build Jun 30 17:50:02 If I wish for these locales to be added to an image, do I need to explicitly add glibc-binary-locale-blah-blah or is there a better way to do it? Jun 30 17:52:59 I think IMAGE_LINGUAS should do it Jun 30 17:56:19 that built them but did not include them in my case. Jun 30 17:57:08 hmm I see Jun 30 17:57:56 meta/classes/image.bbclass:RDEPENDS += "${PACKAGE_INSTALL} ${LINGUAS_INSTALL} ${IMAGE_INSTALL_DEBUGFS}" Jun 30 17:58:07 meta/classes/image.bbclass:LINGUAS_INSTALL ?= "${@" ".join(map(lambda s: "locale-base-%s" % s, d.getVar('IMAGE_LINGUAS').split()))}" Jun 30 17:58:13 so it should have Jun 30 17:58:19 khem: tks ! Jun 30 17:58:40 aleblanc: np Jun 30 17:59:07 :-/ Jun 30 18:12:16 I see the package in oe-rootfs-repo but it's not getting pulled into the image. odd. Jun 30 18:14:38 My image inherits core-image and not image, but that should not matter Jun 30 18:16:49 oh, wow, I had IMAGE_LINGUAS = " " just chillin in my recipe. Jun 30 18:37:03 well there you go then :) Jun 30 19:01:04 RP: I think we need to think about renaming our target tuple for arm+hf to match other distros Jun 30 19:02:45 I have patches atleast 5 different packages to make it understand that we do not pass float ABI via compiler tuple, and some of them are rust compiler, clang, chromium etc. which are not simple to fix and they keep changing and the fixes require enahncements and forward porting all the time Jun 30 19:03:26 if we just following the gnu convention a lot will simplify Jun 30 19:21:10 khem: what is the difference? Jun 30 19:23:22 fullstop: sounds like you copied from core-image-minimal instead of a less minimized image recipe :) Jun 30 19:23:31 huh, didn't know arm+hf was a thing, interesting Jun 30 19:23:47 kergoth: I probably started there.. ages ago. :-) Jun 30 19:24:17 ugh, I'm trying to support non-multilib builds of archs that were built as a multilib in the external oe sdk. ex do a x86_64 base + i686 + x32 sdk and then use the i686 multilib build to build for qemux86 Jun 30 19:24:49 doesn't work due to conflicts with bindir, it either fails to pick up the i686 binaries or just picks up the wrong arch ones. think i'll have to override bindir and base_bindir when building an sdk intended for external use Jun 30 19:25:07 iirc the sourcery toolchain put the multilib binaries in ${libdir}/bin or so Jun 30 21:25:57 RP: these tools trigger on adding hardfloat/softfloat based on tuple Jun 30 21:26:21 so internally they expect the triplet to decide the ABI not cmdline options Jun 30 21:26:54 and many checks are made based on tuple name containing 'eabihf' for arm32 e.g. Jun 30 21:27:18 they are fail for OE, even though we are building a hf ABI toolchain Jun 30 21:34:28 RP: Can you pull in master-next of meta-mingw when you get a chance? Jun 30 21:34:40 Test it on the AB that is Jun 30 21:53:42 JPEW: ok Jun 30 21:56:10 JPEW: https://autobuilder.yoctoproject.org/typhoon/#/builders/89/builds/2106 Jun 30 21:56:17 khem: we could try testing a patch Jun 30 22:45:24 RP: the trouble is we did HF before debian and others came along with agreeing on the new tuple so I guess a bit of legacy there Jul 01 00:00:28 khem, rules are meant to be broken Jul 01 00:53:12 hello. is there a way I can change SRCREV inside the recipe to checkout a particular commit SHA . Jul 01 00:53:50 one idea is to set an environment variable to the commit SHA and set SRCREV to this environment variable inside recipe. Jul 01 00:54:17 How do I set SRCREV to this environment variable inside the recipe Jul 01 01:51:16 bluelightning: are you still in NZ? Jul 01 01:51:31 denix: I am indeed Jul 01 01:52:13 bluelightning: is there MS office down there or do you work remotely? Jul 01 01:52:57 there is one here (in fact there are at least two) but I still work from home Jul 01 01:53:10 bluelightning: and how's MS treating you? are you in Azure/IoT division? Do you like it? Jul 01 01:53:41 yeah it's pretty good - I certainly can't complain :) Jul 01 01:54:19 I'm not working on Azure Sphere, I can say that much Jul 01 01:54:32 (not that I'd mind working on it, of course) Jul 01 01:55:40 I'm trying to increase my external contributions now having had some time to settle in Jul 01 01:55:56 bluelightning: great Jul 01 01:56:11 denix: how's life with you? still in DC? Jul 01 01:56:39 bluelightning: are you in the same group with alejandrohs or different? Jul 01 01:57:29 bluelightning: yeah, I'm fine, still in DC Jul 01 01:58:17 denix: it depends on how high up the organisation tree you go :) but I'm mostly in a different group, though he and I do meet regularly Jul 01 01:58:23 denix: cool Jul 01 02:04:39 bluelightning: with the current sitatuation it feels like we are on different planets... Jul 01 02:04:55 yes it is very strange Jul 01 02:05:39 bluelightning: I wonder if it ever goes back to normal and we get to go to conferences again and meet old friends in person... Jul 01 02:05:40 we have been quite lucky down here, being isolated is usually a problem but has had its benefits in this situation Jul 01 02:05:47 I sure hope so Jul 01 02:06:41 I was certainly looking forward to going to more conferences and meeting people in person again :) Jul 01 02:07:13 yeah, that would be great! Jul 01 02:08:16 I suppose the silver lining is that remote working is much more widely accepted now Jul 01 02:09:41 does MS normally allow remote working, or only temporarily for now? Jul 01 02:11:03 in here some companies embraced remote working, but some can't wait to round up everyone in the office again, once it's over... Jul 01 02:13:13 I think they have been pretty open to it even before COVID - at least there was no concern in my group with being remote Jul 01 02:13:46 I'm not sure but perhaps being a vendor of remote collaboration tools helps :) Jul 01 02:16:02 I've got to step out unfortunately - great to chat with you if only briefly Jul 01 02:16:04 back later Jul 01 02:17:20 bluelightning: take care! Jul 01 02:30:12 hmm apparently I'm not going out after all... will leave getting the car tested until tomorrow Jul 01 02:31:26 I think the thing I miss most about working in the office is the social part, I really liked being able to go to the pub some evenings **** ENDING LOGGING AT Wed Jul 01 03:01:14 2020