**** BEGIN LOGGING AT Tue Jun 23 03:03:14 2020 Jun 23 07:55:28 Looks like meta-sancloud has disappeared from the layer index again. I can't re-add it as it says the name is already taken Jun 23 07:56:37 Looking at https://layers.openembedded.org/layerindex/updates/30073/ I can guess the issue, we depend on meta-rtlwifi which does not have a dunfell branch Jun 23 07:57:10 So it's saying this dependency isn't met and so refusing to process the dunfell branch of meta-sancloud Jun 23 07:58:04 The master branch of meta-rtlwifi supports dunfell though: https://github.com/EmbeddedAndroid/meta-rtlwifi/blob/master/conf/layer.conf#L7 Jun 23 08:22:21 layerindex iirc does not know to look in branch != dunfell ... Jun 23 08:24:10 dl9pf: I've moved meta-rtlwifi from LAYERDEPENDS to LAYERRECOMMENDS, will see if it works on the next layerindex update Jun 23 08:24:28 Also opened an issue on meta-rtlwifi to add stable branches to make the layerindex happy Jun 23 08:29:15 are there any metrics in a build on how mmuch of sstate and downloads have been reused? other than the rather coarse "xxx tasks not needed to run"? Jun 23 08:38:29 Letothe2nd: there is buildstats that you can enable Jun 23 08:38:46 INHERIT += "buildstats-summary" Jun 23 08:39:33 dl9pf: let me check Jun 23 08:39:56 INHERIT += "buildstats" Jun 23 08:39:57 INHERIT += "buildstats-summary" Jun 23 08:43:45 dl9pf: that did the trick, thanks! Jun 23 08:52:22 anybody knows how to add `strace` debugging and diagnostic utility to my linux yocto image ? Jun 23 08:52:38 gaston53: yes. Jun 23 08:52:54 Letothe2nd how please ? Jun 23 08:53:24 gaston53: add strace to IMAGE_INSTALL in your image recipe. http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-devtools/strace/strace_5.7.bb?h=master Jun 23 08:54:24 Letothe2nd thanks Jun 23 08:59:35 gaston53: or even add in local.conf EXTRA_IMAGE_FEATURES_append = " tools-debug" Jun 23 09:00:00 mckoan: *sigh* Jun 23 09:01:19 mckoan okey! Jun 23 10:10:24 Maybe a bit dumb question - but I would like to ask if Yocto project uses containers to test all supported Linux distributions? Jun 23 10:11:20 If yes, then are there somewhere available dockerfiles, which create those containers? Jun 23 10:11:20 To be more specific - each supported distro requires some specific set of packages installed on host Jun 23 10:11:29 (by host I do mean the container) Jun 23 10:12:44 lukma: we use actual installs to test but there are crops containers for most distros Jun 23 10:13:03 lukma: the manual documents the things that need to be installed and we try to keep that in sync with our autobuilder Jun 23 10:16:39 RP: Do those "container" use any docker files for automatic creation ? Jun 23 10:16:49 RP: Or is it solved in another way? Jun 23 10:17:07 Like manually prepared image and then saved/exported? Jun 23 10:18:36 lukma: building Yocto in containers can be done with kas or pyrex AFAIU. That might be something to have a look at if "how to build in a container" is your question and not "how to build for all distributions" Jun 23 10:18:39 lukma: I just searched for "yocto crops" and found https://www.embeddeduse.com/2019/05/06/yocto-builds-with-crops-containers/ which will probably help answer your questions Jun 23 10:26:31 RP: Thanks for help :) Jun 23 11:11:56 RP, oe-core/meta/conf/distro/include/distro_alias.inc contains some entries for grub-efi-{i586/x86-64}, which I believe stem from a time when the target arch was encoded in the native package names, grub-efi-${TRANSLATED_TARGET_ARCH}-native (more info in oe-core commit 53d3f1273983dfce2a907b39768978afe99aab1a). Is that something you can verify, and if so should we remove them ? Jun 23 11:36:31 kroon: yes, we should remove Jun 23 11:37:11 RP, ok Jun 23 11:44:06 how to install libgpiod-dev in my yocto image ? Jun 23 11:45:03 gaston53: http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/libgpiod/libgpiod_1.4.3.bb?h=master Jun 23 11:45:21 gaston53: generic procedures apply, as well as the usual remark about compiling in-target being stupid. Jun 23 11:48:05 lukma: maybe this may be helpful as well https://koansoftware.com/yocto-linux-lab-2018-florence/ Jun 23 11:58:00 mckoan: thats really a nice slidedeck1 Jun 23 12:32:28 Letothe2nd After installing libgpiod , I just find libgpiod-tools into packages.manifest files Jun 23 12:32:36 I don't find libgpiod-dev Jun 23 12:34:12 besides, when I run a program that needs libgpiod , it complains that libgpiod driver is not installed Jun 23 12:34:19 gaston53: like i said: generic procedures apply. if i install X, why should it bring along X-dev? so, install X-dev. plus, compiling in-target is stupid. Jun 23 12:34:45 gaston53: and whats more, libgpiod seems to be quite configurable. you might want to look into it. Jun 23 12:34:46 Letothe2nd: thanks Jun 23 12:35:17 mckoan: i might refer to it some day. do you happen to have a more recent version, or are you fine if i link back to the '18 version? Jun 23 12:35:50 Letothe2nd: you're free to use it, however I don't have plans to update it Jun 23 12:36:09 mckoan: great! Jun 23 12:43:33 hi, in meta-raspberrypi is a line that starts with "CMDLINE_append += '". does the usage of _append in combination with += make sense? I can not find an explanation for it in the Bitbake User Manual. Jun 23 12:46:28 polaris-, some people use += with _append to avoid the extra space, e.g. CMDLINE_append = " something" Jun 23 12:48:13 kanavin_home: is something going on with vulkan headers? (https://autobuilder.yoctoproject.org/typhoon/#/builders/23/builds/2290) Jun 23 12:48:33 that is not logical. Jun 23 12:48:36 * zeddii runs Jun 23 12:49:26 mihai-, thanks! Jun 23 12:54:37 zeddii, :-D :-D :-D Jun 23 12:57:45 kanavin_home: I think your test builds have updated the source mirror and upstream removed the commits we were using :( Jun 23 13:12:52 kanavin_home: can't do the devtool patch refresh thing on ICU Jun 23 13:13:08 $ devtool finish --force-patch-refresh icu ../poky/meta Jun 23 13:13:08 ERROR: Source tree is not clean: Jun 23 13:13:08 ?? source/patches/ Jun 23 13:20:38 quick question. I spent already few hours with it and it gets really frustrating. Jun 23 13:21:09 how can I pull a file from an https URL that requires authentication? Jun 23 13:22:49 I tried the FETCHCMD_wget_append = "auth_args" and FETCHCMD_wget += "auth_args" and both seem to rewrite the FETCHCMD_wget entirely Jun 23 13:23:07 wasn't this suppose to work? Jun 23 13:23:36 https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#http-ftp-fetcher Jun 23 13:24:25 * zeddii is starting to regret the slack signup for ELC Jun 23 13:26:03 nucatus: Have you tried using a url like `https://username:password@site.com/path/to/file`. It's not something I've used myself in a long time but I think that's how it's intended to work Jun 23 13:27:23 paulbarker: in case I store the user and the pass in env variables, can those be expanded in that URL? Jun 23 13:27:57 nucatus: That should also work. Are they set in bitbake or externally in your shell environment? Jun 23 13:28:17 in the shell env Jun 23 13:28:37 nucatus: You'll need to add the variable names to the whitelist for bitbake Jun 23 13:28:56 hmmm... how can I do that? :) Jun 23 13:29:18 nucatus: https://www.yoctoproject.org/docs/3.1.1/bitbake-user-manual/bitbake-user-manual.html#var-bb-BB_ENV_EXTRAWHITE Jun 23 13:30:09 paulbarker: cool. I'll try this. Thanks! Jun 23 13:33:10 kanavin_home: seems we need a ".0" on the end of the branch name Jun 23 13:35:16 otavio: unless there is something open, can you merge meta-freescale-distro, too? Jun 23 13:42:21 kanavin_home: nearly, they've only left tags. I've just referred to master which contains the same commit Jun 23 13:43:53 What's the best way to report a layerindex issue? Post to the mailing list or open a bugzilla issue? Jun 23 13:44:38 Looks like there's now some invalid data being returned by the API which is leading to a backtrace if you run `bitbake-layers layerindex-fetch meta-sancloud -b dunfell` Jun 23 13:44:50 Not sure if it should be fixed on the client side or on the server side Jun 23 13:54:31 paulbarker: bugzilla entry and cc the appropriator people (PaulE, fray)? Jun 23 13:55:24 RP: Will do Jun 23 14:15:20 RP: thanks, I just looked into it - yeah they deleted branches, but we can take commits from master branch which have the needed tag Jun 23 14:15:40 rburton: use -f Jun 23 14:17:33 Bugzilla always confuses me. "New Case" is not the button to raise a new issue haha Jun 23 14:19:05 paulbarker: we'll drop the testopia bits when we upgrade Jun 23 14:19:40 RP: That would probably reduce confusion :) Jun 23 14:21:20 Looks like I can't add more than one attachment Jun 23 14:21:34 paulbarker: probably one at a time Jun 23 14:42:09 find yourself a drink and get comfy, folks: live coding session coming up in 20 minutes! https://www.twitch.tv/letoatreidesthe2nd Jun 23 14:43:24 RP: I've filed it, thanks for the pointers there. https://bugzilla.yoctoproject.org/show_bug.cgi?id=13954 Jun 23 14:49:30 paulbarker: I added mark/paul to its cc Jun 23 14:49:57 paulbarker: we might have to email to get pauls attention but at least the info is collected in one place Jun 23 14:53:30 A dumb question, since I never set up a http download cache before - all the files I have to upload are those ${f} which have a ${f}.done, haven't I? Jun 23 14:56:12 find yourself a drink and get comfy, folks: live coding session coming up in 5 minutes! https://www.twitch.tv/letoatreidesthe2nd Jun 23 14:58:56 YPTM: Scott Murray is on Jun 23 14:59:39 YPTM: Trevor Woerner is on Jun 23 15:15:53 Letothe2nd: kernel-module-hello is the name of the package (most of the time, all packages are prefixed with kernel-module-) Jun 23 15:32:17 Letothe2nd: S = "${WORKDIR}/git" ! Jun 23 15:34:17 Letothe2nd: since you're using AUTOREV, please state also that PV should have SRCPV in it as specified in the documentation here: https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-AUTOREV Jun 23 18:14:22 /whois dreyna Jun 23 18:14:28 lol Jun 23 18:15:00 indeed, who is dreyna? Jun 23 18:15:09 my wife asks all the time Jun 23 18:16:02 my keyboard keeps inserting spaces and other chars Jun 23 18:18:21 I've had the same problem after I spilled water on mine Jun 23 18:18:48 bluelightning: a beer may have been involved Jun 23 18:19:16 although it is a model of Mac Book that has the "keyboard issue" Jun 23 18:19:29 * moto-timo realizes the beer didn't help Jun 23 18:32:25 moto-timo: lies! beer always helps.. or you at least think it does :) Jun 23 18:36:07 mranostay: I have code that proves you are correct! Jun 23 18:46:28 RP: rburton: I think I might have a lead on icudata fails Jun 23 18:55:33 I think there is a race between: Jun 23 18:55:33 LD_LIBRARY_PATH=../lib:../stubdata:../tools/ctestfw:$LD_LIBRARY_PATH ../bin/gencnval -s . -d ./out/build/icudt66l mappings/convrtrs.txt Jun 23 18:56:00 which writes out a file, and these: Jun 23 18:56:03 LD_LIBRARY_PATH=../lib:../stubdata:../tools/ctestfw:$LD_LIBRARY_PATH  ../bin/genrb -s ./misc -d ./out/build/icudt67l -i ./out/build/icudt67l -k -q numberingSystems.txt Jun 23 18:56:05 which read it Jun 23 18:57:55 moto-timo: berr++ Jun 23 18:58:40 Letothe2nd: Habensie bier? Jun 23 18:59:00 oh no, I have forgotten German grammar Jun 23 18:59:03 moto-timo: ja :) Jun 23 18:59:26 moto-timo: as you know, ethanol is not a solution, but a solvent :) Jun 23 18:59:30 * moto-timo resets the cache flushing algorithm Jun 23 18:59:45 Letothe2nd: ++ Jun 23 19:00:21 Tapped a fresh homebrew keg of Oatmeal Cream Stout on Sunday. Jun 23 19:00:27 It is delicious. Jun 23 19:01:12 * Letothe2nd loves stout. Jun 23 19:01:18 send one over, please. Jun 23 19:03:50 * Letothe2nd needs to do post-shoe housekeeping. Jun 23 19:03:59 post-show, even. Jun 23 19:34:41 finally, got meta-external-toolchain working with an oe-built meta-toolchain. just need to add the setup environment parsing to ease usage Jun 23 19:35:24 did we change the compression type for ipk ? Jun 23 19:35:43 https://github.com/kergoth/meta-external-toolchain/compare/master...kergoth:external-sdk + https://gist.github.com/kergoth/b799ce2ce8c9b97c5c64e1587bc2d4ba gets a core-image-base built, but only if i add the locale packages to TOOLCHAIN_TARGET_TASK on the sdk build, otherwise glibc-locale explodes Jun 23 19:37:23 course, most wanting an oe built sdk toolchain probably will just use ext sdk or manually lock down and distribute toolchain sstate archives, but what the hell Jun 23 19:37:31 figured it makes sense to use it for testing the layer if nothing else Jun 23 19:46:47 kanavin_home: its genrb racing Jun 23 19:47:24 Any meta-swupdate folks here? Jun 23 19:47:36 kanavin_home: the easy fix is to just turn off parallel for the data/ directory. i ended up ratholing on improving the rest of the build time to make up for the performance hit Jun 23 19:47:59 kanavin_home: i did look briefly at trying to identify what the actual race is but didn't get very far Jun 23 19:48:26 as i said the icu build is so horrible i rewrote large chunk of the recipe instead Jun 23 19:51:12 RP: can we fix the msg functions to str() the first argument? bb.warn(str(foo)) over and over gets old when debugging Jun 23 20:12:08 kergoth: we probably should although I seem to remember some non trivial overhead on the performance graphs last time I looked at it :/ Jun 23 20:12:24 kergoth: I think I did rip out the debug on the critical paths since then, hopefully Jun 23 20:41:16 rburton: mercy me, rules.mk is written by a custom python library Jun 23 20:41:38 and of course they got it wrong, as made clear by make races Jun 23 20:43:14 kanavin_home: that sounds like a very promising lead! Jun 23 20:44:00 RP: proper fix is made extra complicated due to the makefile containing the issue being generated by a custom in-tree set of python modules! Jun 23 20:44:38 RP: I am tempted to just remove the parallelization, and set the whole data generation thing as an optional PACKAGECONFIG, off by default Jun 23 20:44:58 it was Khem's idea and implementation, so probably not broadly useful Jun 23 21:09:15 RP: I went ahead with that, icu patches on the list Jun 23 21:23:52 Anyone else seeing SSL Cert errors on layers.openembedded.org? Jun 23 21:24:36 'bitbake-layers layerindex-fetch meta-oe meta-python' is failing hard due to that for me Jun 23 21:24:47 website too Jun 23 21:26:20 kanavin_home: I'm not sure that option is going to be well accepted :/ Jun 23 21:26:53 jonmason: getting a NET::ERR_CERT_DATE_INVALID? Jun 23 21:26:55 On another topic, how many people realise ASSUME_PROVIDED entries are also regexs :/ Jun 23 21:27:43 seems the Let's Encrypt cert expired Jun 23 21:31:36 RP: filtered icu data generation is a very recent thing coming from Khem, and is only there to save a few megabytes of space. I'd say it is not commonly used. Jun 23 21:32:10 I can put up with fixing musl failures, but this really doesn't need to be on by default. Jun 23 21:35:53 its more than a few, you can save tens Jun 23 21:36:04 which is non-trivial Jun 23 21:37:47 and the make in icudata doesn't need to build everything, just the data Jun 23 21:42:57 yes, but the default behavior is to not filter anything and just rebuild the full set Jun 23 21:43:19 and the patch I sent does not remove the feature, just not enable it by default Jun 23 21:53:39 mranostay: ssl.SSLCertVerificationError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: certificate has expired (_ssl.c:1056) Jun 23 21:54:41 rburton: does kanavin_home's fine match what you've been looking at? Jun 23 21:55:01 rburton: incidentally my last -next build failed with the icu issue :( Jun 23 21:59:35 vmeson: I put a possible lead into #13911, might be enough of a hint to reproduce Jun 23 22:05:15 RP: that patch is a good workaround, it shunts the function, rburton was trying to fix it in a better wy Jun 23 22:06:29 RP: perhaps we should apply a filter instead of building with no-filter defaults Jun 23 22:06:42 since I do not see the build failures with filter applied Jun 23 22:07:50 khem: according to kanavin_home's analysis it won't matter, the race is there Jun 23 22:12:26 RP: try a build with https://gist.github.com/kraj/59592787effe9613757cedef7038f5f5 Jun 23 22:12:46 RP: we do plenty of builds per day and it has never shown Jun 23 22:48:27 RP: similar enough Jun 23 22:48:45 there's definitely a race, it's just not common Jun 23 22:49:18 RP: i can rebase my patches on top. i shouldn't have got distracted boiling the ocean Jun 23 22:51:52 ah, new qemu still explodes everywhere on aarch64 **** ENDING LOGGING AT Wed Jun 24 02:59:57 2020