**** BEGIN LOGGING AT Wed Apr 28 02:59:56 2021 Apr 28 03:24:39 moto-timo: It does need a patch to meta-oe; I'll try to remember to post it tomorrow Apr 28 05:54:25 is there a common pattern for when say I'd like one recipe/package built with a certain set of defines or dependencies in one case (initramfs) but with others in another case (rootfs)? my problem in particular is I have a custom util written in C I'd like to build with minimal amount of dependencies for use in initramfs, but then build a more full-fledged version for inclusion into the main rootfs Apr 28 07:36:39 xantoz: i would do recipea.bb, recibea-mod.bb: "include recipea.bb, change some stuff.." Apr 28 07:42:04 ernstp: right. I think I will attempt something like this. Thanks Apr 28 07:57:24 yo dudX Apr 28 09:56:33 At last our release is announced on LWN.net: https://lwn.net/Articles/854647/ Apr 28 09:56:50 It took a bit of time because they overlooked the message I sent to them Apr 28 09:57:38 Don't hesitate to post your comments there, like what the most important features are for you (especially the ones I forgot to mention) Apr 28 10:16:06 Morning all. I am trying to include JAVA on a Raspberrypi 3B. As machine I am using raspberrypi3-64. Specifically, I want to build openjdk-8. I am getting the following warning (that in consequence trigger at an immediate later step an error) in the recipe openjdk-7_99b00-2.6.5.bb: WARNING: ... : Exception during build_dependencies for LLVM_CONFIGURE_ARCH. I have included the dependencies for meta-java Apr 28 10:16:12 (non of them as far as I can ... Apr 28 10:16:15 ... tell provides LLVM) and I also have meta-clang in my bblayer config for another feature of my image which is non-java related. Apr 28 10:17:04 Is this a conflict for the target architecture? Apr 28 10:18:42 nemgti-og: what happens if you remove meta-clang? Apr 28 10:20:12 LetoThe2nd: if I remove meta-clang then the build will fail because meta-clang is a dependency for my meta-browser (I am building chromium as part of my image). Apr 28 10:21:29 But.... at least it could help me rule out if that is conflicting with openjdk-8 (dont know why didnt think of that before).... now even if that is the case, I have an issue for my building..... Apr 28 10:23:49 nemgti-og: go and find out. Apr 28 10:26:54 LetoThe2nd: Tanks. I got the same WARNING... something I forgot to mentioned: when including openembedded-core (listed in meta-java dependencies) I got an error due to the fact that openembedded-core conflicts with poky's init script and the error gave me the hing to remove one of these layers (either poky or openembedded-core). I ended up removing openembedded-core... could it be possible that this is Apr 28 10:27:00 the actual root cause of this ... Apr 28 10:27:02 ... issue? Apr 28 10:27:16 the build gave me the hint* Apr 28 10:27:45 nemgti-og: poky/meta == oe-core Apr 28 10:30:18 LetoThe2nd: I didnt get your comment. Is that a question? I am including poky/meta in my bblayer.conf (i dont know if that was the actual question). Apr 28 10:30:46 it is not a question. it is a statement. poky/meta is openembedded-core Apr 28 10:30:57 oh.... ok Apr 28 10:31:04 Thanks Apr 28 10:54:29 Hi there. I'm having a problem installing shared libraries in my system. I'm cross compiling for arm so the libraries get the foo_arm_le.so name. I'm wondering what shall be done so there is also a link to this library that does not contain the 'arm' string. I.e, foo_le.so -> foo_arm_le.so or something like that. Anyone can help with this? Apr 28 10:56:05 jsandman97254: why is the filename reflecting the arch? Apr 28 10:57:02 i mean, you can always whack the do_install stage of your recipe in shape, but it sounds like a bad design decision to me. Apr 28 10:58:45 Don't know. That's what the crosscompiler produces in my configuration. Maybe I am missing something for the buildsystem config. I am even getting /lib/ld-linux-armhf.so.3 but not a link named /lib/ld-linux.so.3. That is giving me trouble as well and I haven't touched any recipe related to that ld. Apr 28 10:59:26 what library are we talking about Apr 28 11:02:10 and, are we talking about something that really runs through a bitbake build? Apr 28 11:08:05 The library creating the foo_arm_le.so is a recipe that Im putting together for a piece of software. I can probably create the link myself in do_install. But the one I mentioned in /lib is what realy gives me trouble. I get this '/lib/ld-linux-armhf.so.3 -> ld-2.31.so' but not a /lib/ld-linux.so.3. Some binaries fail because they don't find the Apr 28 11:08:05 correct interpreter. That ld one is the one I don't know what is the proper way of having an arch independent link for it. Apr 28 11:10:18 the only thing i can say is, look at other recipe libraries and build processes to find out what they do. Apr 28 11:12:13 ok. Will do. I'll see if I can find whatever I am missing. Thanks Apr 28 11:19:10 does site.conf support a form of system-wide setting? like dropping something into /etc/bitbake/site.conf and it being applied everywhere? Apr 28 11:46:35 morning all ;-) Apr 28 11:47:08 getting a new error configuring efl : ERROR: Cross dependency 'openssl' not found Apr 28 11:47:22 even though openssl is listed as DEPEND Apr 28 11:47:26 any idea? Apr 28 12:12:11 intera_91: wrong flags, hardcoded variables in Cmake/Makefile Apr 28 12:52:34 intera_91: but it's strange that it says "cross dependency" check what exactly it's checking for, are you building efl from enlightenment? I've removed meta-efl years ago Apr 28 12:53:14 Aug 2017 Apr 28 14:37:06 RP, ndec: someone's asking to add details to the YP documentation about the generation of container images. See https://bugzilla.yoctoproject.org/show_bug.cgi?id=14368. Do you think that's within the scope of our manual(s), or is it too specific? Apr 28 14:39:00 michaelo: i would say it is a common use case these days, but usually requires some form of custom distro/machine combination. i've presented it like that: https://youtu.be/jPbcQEffzJo Apr 28 14:41:38 michaelo: looks like a good thing to add to our documentation.. assuming we can find a way to document it properly :-) Apr 28 14:55:22 LetoThe2nd, ndec: thank you for your advice. I'll re-open the bug then. Apr 28 14:58:19 michaelo: thanks for taking care of it. its not exactly perfect, but actually https://github.com/LetoThe2nd/meta-containerization gives a good impression of what one has to do in order to make a simple container. Apr 28 15:00:22 LetoThe2nd: thanks! Apr 28 15:37:05 hmm, ELC is now in Seatlle Apr 28 15:37:29 not Dublin Apr 28 15:42:00 Hmm... I guess I don't have to renew my passport then :/ Apr 28 15:46:40 LF just sent out that message. I was planning on a vacation over that time. maybe next yr Apr 28 15:48:44 Seattle != dublin Apr 28 16:09:34 awwww Apr 28 16:11:15 Hi all, I am trying to point the SRC uri to git repo with the tag which was recently migrated (https://github.com/protocolbuffers/protobuf/commit/c9d2bd2fc781fe67ebf306807b9b6edb4a0d2764) when I specify the tag to v3.12.4 it fails to find the commit. Apr 28 16:14:16 prabhakarlad: You have to sepecify the SHA1, not the tag name (I forget why) Apr 28 16:21:34 could you please point me to an example Apr 28 16:23:07 JPEW: you mean set the SRCREV ? I did that and yocto complains the commit doesnt exists in the branch. Apr 28 16:23:08 https://github.com/protocolbuffers/protobuf/commit/c9d2bd2fc781fe67ebf306807b9b6edb4a0d2764 Apr 28 16:23:08 as even github says "This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository." Apr 28 16:23:25 prabhakarlad: SRC_URI is for the git repo and the branch if the commit you're going to list is not in master branch. Then SRCREV for the commit hash Apr 28 16:23:52 prabhakarlad: well... that's an issue isn't it? Apr 28 16:26:10 qschulz: if I manually clone the repo and do a checkout of commit c9d2bd2fc781fe67ebf306807b9b6edb4a0d2764 this works OK Apr 28 16:26:45 with yocto I get this error "ERROR: protobuf-3.12.4+gitAUTOINC+c9d2bd2fc7-r0 do_fetch: Fetcher failure: Unable to find revision c9d2bd2fc781fe67ebf306807b9b6edb4a0d2764 in branch master even from upstream" Apr 28 16:27:08 prabhakarlad: Right, I think the SRCREV must be reachable from a branch head to be used by the fetcher Apr 28 16:28:50 prabhakarlad: Ah, do they just tag releases with no branch head? Apr 28 16:29:45 Im not sure Google just migrated the repo and screwed it up! Apr 28 16:30:48 prabhakarlad: In that case, you can add the nobranch option in SRC_URI, e.g.: SRC_URI="git://....;nobranch=1" Apr 28 16:36:08 JPEW: thanks for the trick, that is working fine :) Apr 28 16:36:15 armpit: hmm, Seatle. Might be virtual for me then... Apr 28 16:53:22 I guess it kinda makes sense since vaccination is going much faster in the US than in Europe? Apr 28 16:54:27 They seem to prefer US timezones for virtual conferences, so perhaps they are expect it to be mostly virtual? Apr 28 17:07:26 seeing this ptest failures on master/gcc11 Apr 28 17:07:28 {'bash': ['run-heredoc', 'run-read'], Apr 28 17:07:28 'libpam': ['tst-pam_start_confdir'], Apr 28 17:07:28 'valgrind': ['memcheck/tests/leak_cpp_interior']} Apr 28 17:08:05 I know the valgrind one is due to the testcase itself and have reported it upstream and someone is trying to fix it but not sure about others Apr 28 17:08:08 any ideas ? Apr 28 17:08:38 i saw a bug on libpam one as well so perhaps thats also not related to gcc11 Apr 28 17:08:47 but not sure about bash ones Apr 28 17:12:57 khem: Did you run that MinGW patch through the AB when testing GCC? Apr 28 17:16:18 not through AB Apr 28 18:05:02 qschulz: depends which bit of Europe but I can understand moving it to US due to vaccination rates Apr 28 18:05:57 JPEW: is that in master-next already? khem reminded me I should be using -next with gcc 11 Apr 28 18:07:39 RP: Yes, it is in master-next Apr 28 18:17:30 JPEW: I think I remember testing it previously then Apr 28 18:18:39 RP: is there a way to see w/o running the full build: a) what sstate reuse would be (stats) - b) what the first package in the chain is that does NOT use sstate ? Apr 28 18:19:26 stuck with debugging sstate re-use atm and wondering how to approach this Apr 28 18:20:49 i know once I can point to a package, I can use diffsigs et al. but I wonder what the equivalend of bitbake-whatchanged is if you do a build from scratch with an sstate-mirror essentially Apr 28 18:25:56 dl9pf: maybe a -n (dry-run) would tell you? Apr 28 18:26:14 hmmm Apr 28 18:26:17 dl9pf: this can be tricky Apr 28 18:26:24 indeed Apr 28 18:27:37 dl9pf: the setscene-only option may help too? Apr 28 19:31:11 vmeson: https://events.linuxfoundation.org/cloud-native-rust-day/ Apr 28 20:13:27 Hi! Could someone please point me towards a light please? I'm trying to build a YoE distro (based on Dunfell) without (L)GPL3 (using meta-gplv2), and some recipes (cairo and libidn2 from oe-core) are failing the do_package() step with "QA Issue: Excluding blabla-{src,dev,dbg...} from packaging [... because bad license]". Whaaat? Why don't the Apr 28 20:13:28 recipes themselves fail to get pulled into the build if their licenses are bad in the first place? Apr 28 20:13:31 I've tried IMAGE_INSTALL_remove and PACKAGE_EXCLUDE-ing the offending entries, but it doesn't work. Apr 28 20:14:29 https://www.openembedded.org/pipermail/openembedded-core/2016-September/126681.html and https://patchwork.openembedded.org/patch/168802/ imply that there have been adjustments to these recipes because of this reason, but they still don't package Apr 28 20:30:49 all, OpenEmbedded Happy Hour is in 1.5 hours - https://www.openembedded.org/wiki/Happy_Hours Apr 28 20:31:29 moto-timo: especially for you ^^^, as you keep on being late :) Apr 28 20:33:00 denix: for once I put the correct time on my calendar!! But thank you Apr 28 21:51:16 denix: haha, lol Apr 28 21:53:03 denix, moto-timo is better then me as I have missed that few meetings Apr 28 21:54:18 armpit: ok, then a personal invite for you - see you in 5 minutes Apr 28 22:49:15 for the record: YoE has a WARN_TO_ERROR_QA var in meta-yoe/conf/distro/yoe.inc which raises a lot of warnings to errors. I removed this incompatible-license from the list so it stays a warning, as it does in Poky. Apr 29 00:08:41 On my system, a base-image-minimal with systemd boots much faster than with sysvinit. On yours too? **** ENDING LOGGING AT Thu Apr 29 02:59:56 2021