**** BEGIN LOGGING AT Wed Feb 05 03:00:09 2020 Feb 05 03:12:29 heh shoot yourself in foot and head perhaps Feb 05 03:13:20 I was asked same question today so its still confusing people Feb 05 03:18:32 when I say think of debian and ubuntu, immediate question was so which company provides support for yocto, I ran away screaming Feb 05 03:39:42 khem, you don't scream.. you just give them the look : ) Feb 05 03:40:03 you know that ;) Feb 05 03:40:39 hehehe OE tasers .. we can train them Feb 05 03:41:42 these guys are working on alpine so they understand distros well Feb 05 03:44:55 perhaps I should have said as alpine is to archlinux ... hmm too late Feb 05 03:45:10 learning experience Feb 05 08:31:25 New news from stackoverflow: howto find which recipe provides a RPM (or DEB or IPK) package? Feb 05 08:41:47 LetoThe2nd: I didn't expect to give you the last push to give up on SO. /me sad Feb 05 08:42:21 qschulz: more opportunities for you to shine. Feb 05 08:42:25 * LetoThe2nd shrugs. Feb 05 08:43:15 as you have probably noticed, when it comes to community work i'm the "lets try it first and judge later" kind of person. sometimes it works: irc, live coding. sometimes it does not: SO. such is life. Feb 05 08:43:31 i'm happy to have finally made a conclusion, and thats it. Feb 05 08:46:25 LetoThe2nd: Whatever suits you :) Feb 05 08:47:11 qschulz: metal attire, if you ask me. wife disagrees. Feb 05 08:47:53 LetoThe2nd: she's afraid of not having one, give her some. Feb 05 08:48:40 qschulz: tried that too. lets count it as another failed community work attempt, mkay? Feb 05 08:48:54 (she happily wears linux shirts, though) Feb 05 08:50:24 LetoThe2nd: who wouldn't it's a damn cute penguin! Feb 05 08:52:02 qschulz: i had my blast from the past this morning, when she picked a tizen one.... Feb 05 08:53:13 LetoThe2nd: everyone has their flows. Feb 05 08:53:20 flaws* Feb 05 08:53:22 well both works Feb 05 08:53:25 work Feb 05 08:53:29 damn wednesday morning Feb 05 09:04:50 good day all Feb 05 09:15:47 * LetoThe2nd does not exactly agree so far. Feb 05 09:16:04 * mranostay scrolls up Feb 05 09:31:23 Hi guysss, I tried to build a musl based distro with systemd Feb 05 09:32:20 I'm currently on sumo, and systemd compilation breaks mercilessly with systemd Feb 05 09:32:27 kriive: correct Feb 05 09:32:41 later releases build, but there's a warning that upstream doesn't support it Feb 05 09:33:12 systemd specifically targets glibc Feb 05 09:33:56 Didn't know that! Feb 05 09:34:24 Thank you Feb 05 09:35:02 Guess I will have to drop musl :c what a shame Feb 05 09:35:30 move to a newer release of yocto and it will build but you get told it isn't a supported configuration Feb 05 09:36:11 kriive: https://github.com/systemd/systemd/issues/10130#issuecomment-423893926 Feb 05 09:36:34 Not a problem at all, I will stay safe with glibc, don't wanna deal with unexpected behaviours Feb 05 09:36:46 musl was like a little play experiment of mine Feb 05 09:38:04 it has its usecases and it is good we are aware of it, but in my personal opinion it is massively overrated. just like initramfs. Feb 05 09:40:38 rburton: thank you for the link, I get the point of systemd maintainers and I think it's fair Feb 05 10:07:37 Well, at least the lists of deb and ipk reproducibile differences now match! Feb 05 10:09:23 :-D Feb 05 10:15:29 How can I easily extend the meta-ide-support to have all libraries that my image also uses? I would like to have an in-place SDK to prevent having to rebuild a separate SDK. Feb 05 11:05:09 Where should I put a Feb 05 11:05:21 PACKAGECONFIG_remove_pn-qtbase = " blah blah"? Feb 05 11:05:32 IMAGE.conf? Am I correct? Feb 05 11:06:01 there is no image_conf :) Feb 05 11:06:25 the .conf of your DISTRO is the most obvious place for this. Feb 05 11:08:41 yeah, right, there is image.bb, the whole machine, distro and image is still a bit foggy lol Feb 05 11:08:51 thank tou Feb 05 11:12:24 kriive: well i take it that you have already watched the video that explains it? :) Feb 05 11:13:29 Yup! recipe data is local conf data is global Feb 05 11:13:34 kriive: in this specific case it is rather simple, too: if you stick it in the image recipe, it cannot affect what happens in another recipe. but you want to influence qtbase, which comes from another recipe. so, the image.bb file is certainly wrong. Feb 05 11:13:38 :) Feb 05 11:14:27 Yeah, that makes sense Feb 05 11:15:17 Image is a recipe, not a config, I need a post-it somewhere ahah Feb 05 12:04:28 rust takes a long time to compile! Feb 05 12:08:53 rust-native-1.39.0-r0 do_compile - 1529s and counting Feb 05 12:09:48 Ad0: tail /proc//fd/2 and check if it's still doing something? Feb 05 12:10:13 it does, and in true spirit, it finished just as I pasted this! Feb 05 12:10:14 (or tail ${WORKDIR}/temp/log.do_compile would hopefully be doing the same) Feb 05 12:10:17 :D Feb 05 12:10:38 Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz Feb 05 12:10:45 6 cores, 12 threads Feb 05 12:12:14 Ad0: are you setting paralell_make somewhere and making it not use the default (which is nproc)? Feb 05 12:12:27 no Feb 05 12:12:31 it uses all cores Feb 05 12:12:38 it just takes time I guess Feb 05 12:14:33 Ad0: did you check if swapping was used? Feb 05 12:15:25 doesn't it only do that if I run out of RAM? Feb 05 12:15:34 all cores were @ 100% Feb 05 12:23:01 rust-native is just massive I think Feb 05 13:25:46 Hello, is there any way to have a recipe that is present in the target image but not in the sdk? Feb 05 13:27:21 RP: looks like there weren't any logrotate-related failures on the AB this time? Feb 05 13:44:34 tgamblin: no! :) Feb 05 13:44:44 tgamblin: I think it was the other change confusing things, sorry Feb 05 13:45:01 tgamblin: interpreting things is proving tricky atm :( Feb 05 13:45:26 RP: No worries. I think my patch is still sort of a workaround. I know where the issue seems to be, but I'm unsure if there's anything to be done that is a better fix than what the patch already does Feb 05 13:45:57 tgamblin: I think I'll end up merging it and then we can take any further issues from there Feb 05 13:46:09 RP: k Feb 05 13:46:45 (catching the other half of the population...) i've started a gsoc ideas page. feel free to edit, or email me suggestions -> https://wiki.yoctoproject.org/wiki/GSoC/Ideas Feb 05 13:47:39 tgamblin: in fact I'll do that Feb 05 13:48:07 tlwoerner: if i had any real ideas, i would happily add. Feb 05 13:48:19 RP: Nifty, thanks Feb 05 13:48:36 LetoThe2nd: thanks! Feb 05 13:58:43 RP: for the record, the issue seems to be that when meta/recipes-core/initscripts/initscripts-1.0/populate-volatile.sh runs, it is sometimes unable to create the wtmp log (filesystem hasn't come out of readonly before the script runs, maybe?). The v3 patch just makes sure to create the file before the test Feb 05 14:00:30 Ad0: rust compilation is slow and rust is built in rust Feb 05 14:00:39 so you're building a complete compiler in a slow-to-compile language Feb 05 14:00:42 s/before the test/as the first step in the test/ Feb 05 14:00:48 hehe ok Feb 05 14:02:16 there are people moaning about how slow rust is, but the tradeoff it makes is lots of analysis at build time so Feb 05 14:04:00 rburton: who cares about sane discussion when complaining is so easy. Feb 05 14:04:28 Clearly Rust needs a quick blurb about the time-bandwidth product :) Feb 05 14:04:56 * rburton sees that RP signed off a patch properly and faints Feb 05 14:13:32 rburton: haha :) Feb 05 14:14:50 tgamblin: perhaps its a parallel race in volatiles? the mount point/location the log is may be racing with something which creates that path? Feb 05 14:16:35 RP: Possibly. I'll keep looking at it in the background Feb 05 14:18:23 RP: do the rest of the reproducible builds people understand just how much further we're taking the concept i wonder Feb 05 14:18:47 rburton: perhaps not Feb 05 14:19:05 debian mandates a stable host and build paths etc Feb 05 14:19:47 did your latest patch series sort out all of the reprod issues that were recently filed or are there any left Feb 05 14:20:10 rburton: we're about 50% through the issues Feb 05 14:20:19 rburton: There are open bugs and a number I haven't filed yet Feb 05 14:20:30 i'l grab one of the open ones once i've written a report Feb 05 14:20:33 rburton: good news is the list of issues looks pretty defined now Feb 05 14:20:49 rburton: the last builds were quite specific in the issues they showed Feb 05 14:21:09 rburton: any help is much appreciated! Feb 05 14:22:20 the perl-cross maintainer is quite responsive, worth copying them to perl-cross and seeing if they know the cause Feb 05 14:23:06 hi all, is it possible to specify a local directory tree as SRC_URI in a recipe? Feb 05 14:23:45 nameclash: not directly. you need to use externalsrc for the desried effect Feb 05 14:28:36 nameclash: also depends on why you want to do such a thing. Might be a bad idea :) Feb 05 14:28:46 rburton: happy for someone to ask but I think we may be on our own with some of the issues Feb 05 14:28:53 rburton: I've sent a summary to oe-core and cc'd you Feb 05 14:28:56 RP: just sent one to perl-cross we never know Feb 05 14:29:54 rburton: the openssl/libidn2/sysvinit ones look simpleish, the others trickier Feb 05 14:30:05 the idn one involves *libtool* Feb 05 14:30:14 rburton: simple then! Feb 05 14:30:16 * RP runs Feb 05 14:41:16 LetoThe2nd, qschulz sorry, was interrupted by a collegue... the reason I ask is we maintain our yocto project in a super repo that contains our proprietary layer, among others. our layer points to a bunch of repositories in our code base and we have consistency problems when working on branches Feb 05 14:42:43 nameclash: that does not exactly change the fact that you cannot refer to a directory in SRC_URI Feb 05 14:46:10 the idea of referring to a directory came along with the idea of adding the component repo's as submodules to the meta repo to resolve the consistency problems Feb 05 14:46:48 nameclash: no worry, i get the idea. it just doesn't work like that, sorry. Feb 05 14:47:25 right now, we have recipes pointing at the master branch of the repo and that causes constant headache when the project is under heavy development... Feb 05 14:48:05 nameclash: like i said: there is externalsrc to be used for that, we even have the magic there in devtool Feb 05 14:49:32 I'll look into externalsrc, thanks Feb 05 14:50:42 nameclash: I fail to see the consistency problems if one's using shared branches or tags? My company uses externalsrc but I'm not super fond of it Feb 05 14:51:29 our src uri's look like this: Feb 05 14:51:41 Using subnmodules + externalsrc, aren't you going to move the consistency problem to the repo conf instead of Yocto? Feb 05 14:52:00 SRC_URI = "git://HOST/REPO;protocol=http;branch=master;" Feb 05 14:53:05 so the recipes are fetching master, even if we're on a feature branch on the yocto repo Feb 05 14:54:17 suppose the feature that's being worked on involves changes in both the yocto repo and the component repo. This meant that the recipe had to refer to: Feb 05 14:54:22 nameclash: I still don't see it. If you're not using AUTOREV, there is no reason for your sources to change Feb 05 14:54:34 SRC_URI = "git://HOST/REPO;protocol=http;branch=FEATURE;" Feb 05 14:56:19 I need to read about autorev to understand your comment, hang on Feb 05 14:56:43 you want all your recipe to switch to this FEATURE branch? what exactly is your problem? what's your worry? what do you want to improve in your workflow? Feb 05 14:56:51 nameclash: you most likely don't want autorev so forget about it Feb 05 14:57:00 ok Feb 05 14:57:16 nameclash: for reference anyway: https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#var-AUTOREV Feb 05 14:57:49 RP: diffoscoping the gtk differences now Feb 05 14:57:53 used to make Yocto always fetch the last commit in a specific branch (or git repo). breaks reproducibility and deterministic builds Feb 05 14:58:45 nameclash: you want all your recipe to switch to this FEATURE branch? what exactly is your problem? what's your worry? what do you want to improve in your workflow? Feb 05 14:59:48 kindof, yes Feb 05 15:00:24 we thought about injecting the branch to get via environment variable but that's likely to cause headache, too Feb 05 15:01:14 there was a discussion in the ML in the last month or so on how to do this Feb 05 15:01:15 what is the default task when you run bitbake foo? is it do_build in foo? Feb 05 15:01:17 it's a recurring topic Feb 05 15:02:17 fray: yes Feb 05 15:02:56 I thought so, that's why I wanted to ask how its done the right way before walking the dark path.. Feb 05 15:03:16 can you provide more detail on how to find that ML thread? Feb 05 15:03:23 nameclash: http://lists.openembedded.org/pipermail/openembedded-core/2020-January/291082.html Feb 05 15:03:49 thanks man, I'll look into it Feb 05 15:10:15 RP: either try.diffoscope is dying or the gtk libraries just OOMd the tool Feb 05 15:10:56 rburton: nice :/ Feb 05 15:11:05 rburton: try on just the src to start with? Feb 05 15:28:58 rburton: build/gdk/gdkresources.c differing it seems Feb 05 15:29:31 rburton: I suspect you have a better idea than I do on that one Feb 05 15:30:01 if the source differs, the binaries will differ so its no doubt from that file Feb 05 15:41:03 rburton: found the tar issue Feb 05 15:41:13 i think the gtk issue is just sorting Feb 05 15:42:55 rburton: sometimes its that simple Feb 05 15:43:07 rburton: although the size did differ? Feb 05 15:43:10 its the embedded gresource section Feb 05 15:43:18 the tool reports all the same files are in it Feb 05 15:43:22 so i'm wondering if its sorting+padding Feb 05 15:43:28 rburton: could be padding Feb 05 15:44:50 the gresource section was the only one that grew Feb 05 15:51:18 rburton: you were so close with the idn2 fix Feb 05 15:51:28 RP: gar Feb 05 15:51:41 should i just have chopped more away Feb 05 15:52:03 hm i forgot to wipe potential whitespace didn't i Feb 05 15:52:09 rburton: you removed the -L but not the whitespace Feb 05 15:52:12 rburton: right Feb 05 15:52:24 thats quite frustrating! Feb 05 15:52:33 rburton: yes, so close! Feb 05 15:53:01 Who is the likely culprit if I get a reboot: not found error after installing the hddimg via EFI? Feb 05 15:54:46 rburton: I am slightly concerned it adds/removed the lib path though :/ Feb 05 16:03:06 * zeddii can see the light at the end of his build tunnel. Feb 05 16:33:05 zeddii: is it daylight or a train though? :) Feb 05 16:33:49 bug 13776 is a bit of a mystery. How can a group appear/disappear :/ Feb 05 16:33:50 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=13776 normal, Undecided, ---, unassigned, NEW , sysvinit reproducibility issue (shutdown user leakage) Feb 05 16:47:58 RP: if you want to look at perl instead of sysv the perl-cross guy replied to my bug Feb 05 16:57:20 rburton: thanks, will have a look Feb 05 16:59:45 hey. is there anyone available with knowlege about emmc devices, jdec5.0 Feb 05 17:17:52 RP: was the gtk repro fail reliable? Feb 05 17:18:57 rburton: yes Feb 05 17:19:09 rburton: in that we've seen it regularly Feb 05 17:19:42 ok i think i can replicate the failure with constructed files, so i've a simple fix Feb 05 17:25:34 RP: patch on the list. let it sit in next for a bit Feb 05 17:25:54 i'm about 80% confident Feb 05 17:32:35 i should write a wrapper for git-send-email to point out i have staged but uncommitted changes before it lets me send Feb 05 17:33:40 rburton: thanks! In the context of this commit? Feb 05 17:33:46 yes :) Feb 05 17:34:10 .../gtk+/gtk+3/sort-resources.patch | 19 +++++++++++++++++++ Feb 05 17:34:10 1 file changed, 19 insertions(+) Feb 05 17:34:16 helpful to commit the .bb change too Feb 05 17:34:36 rburton: the commit on list is ok though? Feb 05 17:34:45 RP: i sent a v2 Feb 05 17:34:58 rburton: which I'm looking at hence the confusion! :) Feb 05 17:35:27 rburton: I think the vairadic thing is dash vs bash Feb 05 17:35:35 rburton: trying to confirm now Feb 05 17:36:09 we should bind-mount our own /bin in place so we can really force sh to be bash Feb 05 17:43:42 rburton: confirmed its dash vs bash Feb 05 18:25:23 might be doable without root with a new mount namespace? Feb 05 18:25:24 * kergoth yawns Feb 05 18:37:32 root privs would be the main thing stopping us Feb 05 18:43:33 I poked around a couple of weeks ago with the thought of trying to prototype something, and there's not obviously a common cross-distro story on unpriviliged namespace / containers that I could find Feb 05 18:47:16 hi all, how to add a command in IMAGE_PREPROCESS_COMMAND after some other command e.g prelink_image? I'm trying to sign rootfs and I want to do it after prelinking but right now it is execute before prelinking with changes executables Feb 05 19:45:31 evening creich Feb 05 19:45:38 hope you've had a good day? Feb 05 20:56:35 https://www.irccloud.com/pastebin/LkiEUMgs/ Feb 05 20:56:50 when trying to run bitbake -c menuconfig Feb 05 21:08:03 Hi, I have one question related to dmcrypt and bitbake Feb 05 21:09:15 It is still not possible to open luks image due to missing sudo called from recipe ? Feb 05 21:10:26 I explore this topic a little and I found only unanswered questions or answers that it is not possible to use fakeroot because of device mapper subsystem Feb 05 21:10:59 someone something ? Feb 05 21:24:45 you can't use sudo in a recipe, unless you're really evil Feb 05 21:25:19 if you set up sudo so that it doesn't prompt for a password it might work... Feb 05 21:28:11 but If I want to do this in generic way it seems it is not possible Feb 05 21:28:36 so not as a root Feb 05 22:08:50 rokm: have you looked in to utilities to access such images in user-space instead? Feb 05 22:43:05 attieg: hi there. any progress on your problem? have been busy all day and was disconnected quite some time, so might have missed some messages Feb 05 22:43:20 palate: how's you're project going on? any progress? Feb 05 22:52:15 Crofton|road: doesn't menuconfig only work on linux-yocto and derivatives like linux-intel? Feb 05 22:53:17 maybe Feb 05 22:53:35 this is linux-yocto + bbappends, but mayb esomething fussy broke it Feb 05 23:01:34 haven't seen the history of your conversation, but if you guys talk about menuconfig for the kernel, that one should work for every linux kernel Feb 05 23:01:45 as far as i can tell Feb 05 23:01:58 menuconfig depends on the terminal support which is temperamental and poorly tested :( Feb 05 23:02:07 * RP wishes someone would write proper tests Feb 05 23:23:55 * armpit wish I could read and write Feb 05 23:25:02 armpit: just use write() and read() Feb 05 23:44:41 khem: xlocale.h was dropped in later glibc versions right? Feb 06 00:05:14 * RP -> sleep Feb 06 00:08:25 criech: hi, unfortunately not yet... thanks for asking - I've got another yocto environment & board I'm going to try it on tomorrow... at least that should prove something Feb 06 00:08:42 for the moment I'm just patching it into the tree and building it there (not ideal) Feb 06 02:06:19 other than -v and -DDD options for bitbake is there any other way to single step inside recipes files. Feb 06 02:06:39 I see -c devpyshell option however, I do not see more documentation on how to use it Feb 06 02:06:52 it will be good to single step in the recipe file Feb 06 02:13:09 RP: yes its dropped Feb 06 02:14:58 kiwi_29: you can do bitbake -e that dumps the pre-processed output that bitbake will see for that recipe Feb 06 02:15:14 its quite handy to see how variables are expanded etc. Feb 06 02:47:07 quit **** ENDING LOGGING AT Thu Feb 06 02:59:57 2020