**** BEGIN LOGGING AT Tue Feb 12 02:59:59 2019 **** BEGIN LOGGING AT Tue Feb 12 04:22:34 2019 **** BEGIN LOGGING AT Tue Feb 12 05:13:22 2019 **** BEGIN LOGGING AT Tue Feb 12 06:36:21 2019 Feb 12 07:03:31 hello Feb 12 07:10:21 Hello all Feb 12 07:13:10 Could someone help me, I'm wondering if I can take "snapshot", ie. save all upstream sources from certain build I'm trying to make, so that I can replicate it offline.. I have tried to go trough the manuals, and found BB_GENERATE_MIRROR_TARBALLS, but it makes quite a lot of data, is it all needed actually to build? any help is appreciated =) Feb 12 07:48:07 Jaakko: I think that's the way to go. You can get some ideas from e.g. https://pelux.io/software-factory/PELUX-3.0/swf-blueprint/docs/articles/baseplatform/reproducible-yocto-builds.html Feb 12 07:49:00 It's not all relevant for you, but parts of the info is probably useful Feb 12 08:11:11 erbo: very interesting project thanks for sharing ;-) Feb 12 08:48:49 Hey thanks very much erbo Feb 12 09:03:10 Greetings! I'm trying to build a GTK3 app within an embedded build, and even if the m4 macro for GTK has been added, I still get gtk/gtk.h: No such file or directory within bitbake Feb 12 09:44:22 Jaakko: the easiest method is probably a clean download directory and set your original downloads directory as a "own-mirror". Then a "bitbake --runonly=fetch" Feb 12 09:45:10 Jaakko: The MIRROR_TARBALLS variable creates tarballs of any source control repos it checks out so its easier to use the end result as a mirror. You can likely discard the git2 and other scm work directories if you do that Feb 12 09:56:11 good morning Feb 12 10:07:13 mckoan: yeah it's a interesting effort, and although it's mainly targeted towards automotive it contains some good stuff that's generic Feb 12 10:07:35 armpit: done Feb 12 10:07:52 I used to work on it in its early days Feb 12 10:29:08 RP: thanks, I'll take a look Feb 12 10:34:22 is there a way to dynamically set/extend VARDEPS so that it triggers changes in sigdata? Feb 12 10:35:37 RP: ^^^^^^ Feb 12 10:37:10 rationale: we have a recipe that modifies a sourcefile based on variable replacements. now we would like to rebuild whenever one of the variables change, but we do not know which ones are relevant in advance, e.g. we parse the template file during a custom task after do_fetch Feb 12 10:38:36 RP: that library (libEGL) needs to be installed on the host Feb 12 10:38:57 LetoThe2nd: the premise of the code is that you can calculate the signature in advance Feb 12 10:39:21 RP: in terms of, without executing any tasks? Feb 12 10:39:22 kanavin: ah, right. So we have inconsistent installs on the builders Feb 12 10:39:33 LetoThe2nd: yes, else how can it know to fetch from sstate? Feb 12 10:39:59 RP: humkay. any pointers on the problem described, maybe? Feb 12 10:40:10 RP: yep, I think so. The same test passed fine on ubuntu's Feb 12 10:40:29 LetoThe2nd: best you can do is depend on the hash of the upstream source control Feb 12 10:40:37 LetoThe2nd: when it changes, it rebuilds Feb 12 10:41:05 kanavin: sounds like I need to talk to halstead and we need to audit the dependencies in the manuals Feb 12 10:41:44 LetoThe2nd: the sstate hash equivalency changes are related to what you're asking Feb 12 10:42:08 RP: yeah i noticed there was something going on, but didn't follow the details. Feb 12 10:42:17 * RP would love to work on that but other problems right now Feb 12 10:44:09 RP: i see. well in a nutshell it means for my given situation: does work as expected, we either need another evil hack, or manually fill vardeps. agreed? Feb 12 10:44:42 LetoThe2nd: I think so Feb 12 10:45:48 kanavin: can't help wonder if we could kill off FuncFailed, its a horrendous idea. Patch looks good though Feb 12 10:45:57 RP: okay, thanks for the explanation! Feb 12 10:47:39 RP: I can't really offer an opinion, but yes, there's a lot of baggage in the exception handling code Feb 12 10:49:11 kanavin: I didn't know what I was doing with some of that stuff :( Feb 12 11:05:47 Out of curiosity: the last release for the ADT that I can find is 2.0.3 Feb 12 11:06:05 *is within yocto 2.0.3 Feb 12 11:06:13 Is it now deprecated? Feb 12 11:08:10 sk_tandt: the SDK and eSDK are the current toolchain offerings Feb 12 11:08:34 Good, supposed as much Feb 12 11:16:19 RP: If I may ask - is the glibc-init package only to build and install crtX.o and allow further development/build of e.g. gcc ? Feb 12 11:16:31 Mh, and yet: I've installed the Yocto SDK Eclipse plugin, but I can't find the toolchain built with bitbake meta-ide-support: isn't it within my usual build/tmp ? Feb 12 11:16:33 RP: Or is there any other gotcha? Feb 12 11:17:03 New news from stackoverflow: How to avoid giving autoreconf in yocto? || How to securely update configuration for root password in yocto Feb 12 11:34:34 hi Feb 12 11:34:54 is it possible to access barebox environment from within yocto linux? Feb 12 11:35:30 I would like to change the next boot target from within linux. Feb 12 11:35:44 (change of default target) Feb 12 11:47:08 New news from stackoverflow: How to securely update configuration for root password in yocto? Feb 12 11:53:52 BuddyButterfly1: here you go: http://layers.openembedded.org/layerindex/recipe/67489/ Feb 12 11:56:47 lukma: we got rid of it in master Feb 12 12:07:42 LetoThe2nd: Oh, wow, cool! Thanks for pointing it out. Feb 12 12:08:36 RP: one more thing (TM). is there a way or best practise to put a multiline string into a bb variable, e.g. into the recipe verbatim? Feb 12 12:09:04 BuddyButterfly1: have fun Feb 12 12:17:00 LetoThe2nd: our escaping sucks :/ Feb 12 12:17:14 New news from stackoverflow: How to add python-xbee receipes for yocto Feb 12 12:17:17 LetoThe2nd: functions are multiline strings with the func varflag set Feb 12 12:18:11 RP: can you elaborate on the last? Feb 12 12:19:13 LetoThe2nd: somevar[func] = "1" Feb 12 12:19:48 Hm... If I've got a couple of old boards to re-build & support (from danny & dylan) is there any "best practice" to follow? Most of our boards are build on an ubuntu 16.04 box as they're pyro - I could either have another box / VM running ubuntu 12.04 or I could try and port up the BSP's to pyro... Feb 12 12:20:09 RP: so if i abuse that for a standard variable, it will basically be multilined? Feb 12 12:20:26 Is that a heinous thing to do / likely to break the BSP builds in exciting and unexpected ways? Feb 12 12:20:50 no_such_user: go and find out (TM) Feb 12 12:20:59 Ha :-D Feb 12 12:22:03 no_such_user: AFAICS there is no other generic answer. depending on your projects in can be painful or painless to forward it, and painful or painless to build on a newer ubuntu. Feb 12 12:22:03 LetoThe2nd: OK, if you're not saying "youre insane to consider giving that a go" that's kind of a good starting point :-) Feb 12 12:22:31 no_such_user: not insane at all. Feb 12 12:23:13 LetoThe2nd: Thanks - I still haven't quite worked out how Yocto works philosophically speaking! Feb 12 12:24:44 It feels like Im a bit at the mercy of upstream layer maintainers (ie meta-freescale and meta-ti) and what targets / kernels they support in their releases... I wasnt sure if trying to keep versions going in my (later release) layers that theyve dropped for their layers is considered to be criminal or not! Feb 12 12:25:04 and dont even get me started on the PPC targets I have... :-/ Feb 12 12:25:55 no_such_user: its the same as everywhere. if you expect something specific, insert coin. Feb 12 12:27:00 LetoThe2nd: Oh sure - thats not a problem, more trying to make sure I dont start off a significant porting effort by heading off in what turns out to be a completely daft direction :-) Feb 12 12:29:37 RP: Ok, I will look on it (as now I do use 2.6.1) Feb 12 12:34:25 lukma: you can remove glibc-initial if you drop the siteconfig pieces basically Feb 12 13:34:03 Using m4 in an autotools chain, getting configure: error: The pkg-config script could not be found or is too old (but it is present in the Host system). How do I add it to Yocto? Feb 12 13:40:11 Hi all. I have a little question. With bitbake it is possible to create some variables flag VAR[a]="test1", VAR[b]="test2". Is it possible to use it for specific recipe, that is to say like this VAR_pn-[a]="test3" ? Feb 12 13:40:16 thanks for your help Feb 12 13:44:10 sylvain_: you can't use overrides with flags, no Feb 12 13:45:46 Ok. Thanks. Suppose i need to create variable like this VAR_a="", VAR_b="" and VAR_a_pn- I'm trying to understand the poky/oe-core workflow. I would've expected oe-core to be a submodule in the poky repository, but that's not the case. It seems the poky repo is a fork of oe-core, which then just cherry-picks patches? Feb 12 13:57:24 m314, I would expect the same thing :) When I build projects I use submodules to build the project to help manage the layers Feb 12 13:57:36 for some reason poky doesn't do that Feb 12 13:59:53 Yeah, the thing is I'm trying to figure out the "proper" way to roll your own distro. Having meta-poky as a submodule seems a bit messy, so I thought the reference distribution would be the place to look for how to do it :-) Feb 12 14:03:40 Hi all, an open question: Do you think which is your favorite builder inside Yocto meson/ninja or Makefile/make ? and maybe why you prefer one or the other as solution ? Thx all :) Feb 12 14:05:29 m314, there are several approaches, I've ended up using submodules Feb 12 14:18:41 i'm getting weird results operating on the bluez5 recipe with devtool Feb 12 14:19:02 1) devtool modify -x bluez5 src/bluez5 Feb 12 14:19:44 2) i modify a file in src/bluez5/oe-local-files/ Feb 12 14:21:12 ... (hang on, checking something) Feb 12 14:26:39 3) i devtool build bluez Feb 12 14:28:04 here's the weird part: 4) i cd to src/bluez5 and do a "git status". i see a whole bunch of other files modified which i didn't modify, e.g., "Makefile.in, aclocal.m4, install-sh, ...", and i do NOT see the file I modified in the git status output Feb 12 14:29:13 further, i built the image which utilizes bluez and the file i modified (/etc/init.d/variscite-bt) is not modified Feb 12 14:29:37 so what's happening here? Feb 12 14:31:37 kergoth: are you here yet or pacific time? Feb 12 14:44:48 backup, new (simpler) question: Feb 12 14:45:44 if two different layers have a .bbappend file for the same recipe, will they both be applied? if so, in what order? Feb 12 14:49:38 yates, it's called overlayering: sorry not to be of more help, don't know much about it Feb 12 14:49:39 anyone? Feb 12 14:50:24 ok thanks sk_tandt Feb 12 14:53:56 Btw, as for the modified files (Makefile.in): they seem to be due to a folder run of autotools Feb 12 14:54:12 perhaps an autoreconf or akin to it Feb 12 14:54:40 sk_tandt: yes, i finally realized that Feb 12 14:56:11 oe provides that (bluez5) recipe, but meta-variscite-fslc provides a .bbappend to it, including the use of some different run-time scripts. i want to modify one of the run-time scripts the mega-variscite-fslc .bbappend provides Feb 12 14:57:31 s/mega/meta/ Feb 12 14:58:14 Yup, figured you were referring to the BSPs : P Feb 12 14:58:55 What would you like to change? Feb 12 14:59:36 the variscite-bt script Feb 12 14:59:53 and the corresponding variscite-bt.conf configuration file Feb 12 15:00:38 in meta-variscite-fslc/recipes-connectivity/bluez5 Feb 12 15:00:59 Now, I may be way off as I am *very* new to the yocto usage, but Feb 12 15:02:16 I'd create a layer of my own, clone the bluiez5 recipe from the Variscite layer, change the config from files/variscite-bt and any other file, then include my forked recipe Feb 12 15:02:48 (And layer, ofc) Feb 12 15:03:45 i think that is discouraged by oe Feb 12 15:03:59 overriding recipes is what i think it's called/ Feb 12 15:04:01 ? Feb 12 15:04:09 Oh, I'd use another name Feb 12 15:04:13 Sorry, forgot to mention Feb 12 15:05:56 the solution may actually be very simple: define a bluez5_%.bbappend with a FILESEXTRAPATHS_prepend and the corresponding files in my own custom layer Feb 12 15:11:26 what i don't understand is the order in which these two .bbappends will be applied. Feb 12 15:13:13 yates: If there is any order, its probably based on layer priority Feb 12 15:21:17 JPEW: that is what i was thinking. so we should be good. trying it now anyway to see what happens Feb 12 15:39:45 where can i find the source for the xinput_calibrator utility that appears to come with sato matchbox? is matchbox a recipe i can devtool modify? Feb 12 15:41:02 behold the magic of google! https://www.freedesktop.org/wiki/Software/xinput_calibrator/ Feb 12 15:41:27 or, the recipe has a SRC_URI link in Feb 12 15:42:18 can also search the layer index for calibrator, or your localfilesystem Feb 12 15:44:01 thanks kindly Feb 12 16:04:40 sjolley: can't get zoom to work again :( Feb 12 16:38:43 I'd like to run a different init.d script in my development image than my release image. Is the best way to do this just to create different recipes for my init script and select the specific one in my image recipes? Feb 12 16:43:35 probably best, yes Feb 12 16:43:43 or install them both with different names and just selectively *enable* them, i suppose Feb 12 16:44:44 One of the things I want in my dev image allows some remote debugging, so I think I'd rather not have it on the system at all in the production image. Thanks for the ipunt. Feb 12 16:45:11 Would there be any way to do this with the -dev or -dbg suffix, or similar? Feb 12 17:05:52 if you're wanting a -dev to overwrite a file that's in another package, not going to happen Feb 12 17:06:02 both the binary package manager and bitbakew il get annoyed at conflicts Feb 12 17:18:09 i'm at my wit's end with this Feb 12 17:18:40 here is my bluez5 .bbappend: https://paste.fedoraproject.org/paste/SQnC6nS5cRuEyosRZE0PGQ Feb 12 17:19:16 which i was hoping would override the meta-variscite-fslc/recipes-connectivity/bluez5 bbappend Feb 12 17:19:51 it appears that my variscite-bt is getting installed, but my variscite-bt.conf is not. Feb 12 17:19:59 i'm at a loss to see why Feb 12 17:29:10 oh. Feb 12 17:29:15 yates, figured out? Feb 12 17:29:36 looks like priorities are wrong Feb 12 17:29:52 bitbake-layers show-appends is a nice tool.. Feb 12 17:33:58 in BBFILE_PRIORITY_layer1 = "n1" and BBFILE_PRIORITY_layer2 = "n2", layer1 will have priorty over layer 2 if n1 > n2? right? Feb 12 17:34:02 or is it n1 < n2> Feb 12 17:34:07 or is it n1 < n2? Feb 12 17:37:30 RTFM Randy! Feb 12 17:37:39 yates: yes sir! Feb 12 17:39:14 does BBFILE_PRIORITY apply to .bbappend files as well as .bb files? Feb 12 17:42:33 it's not working. i see my layer priority to 1000, which is higher than all others (from bitbake-layers shoiw-layers) Feb 12 17:44:47 but bitbake-layers show-appends shows my .bbappend lower than meta-variscite-fslc Feb 12 17:45:18 https://paste.fedoraproject.org/paste/aTrOnwpDLMfP6U0w-5c~LQ Feb 12 17:45:55 https://paste.fedoraproject.org/paste/N0Wl-4kb3sm5SdbS3WqLBA Feb 12 17:46:12 is there a know bug about this? Feb 12 17:48:20 known Feb 12 17:48:26 kergoth, rburton? Feb 12 17:49:35 my layer is meta-ebtron Feb 12 17:55:43 ? Feb 12 18:07:24 RP: any input you can give on my problem? Feb 12 18:07:51 LetoThe2nd: or you? you have been very kind and helpful in the past Feb 12 18:12:50 this is hanging me up badly. Feb 12 18:13:03 how about this: SHOULD this work? Feb 12 18:19:06 yates: Offhand I'd have said that should work. I'd put some bb.warn() into cooker.py and see what its really doing Feb 12 18:19:37 yates: filter the output with "if 'xxx' in name to only show details about the recipe you care about Feb 12 18:23:11 RP: thank you. i'm trying to google-up "bb.warn/cooker.py" because i have no idea what you mean. Feb 12 18:25:07 yates: cooker.py is part of bitbake, the code in there handles bb file priorities in layers. I was suggesting you could gain more insight into what is going on with some print statements (I use bb.warn("foo") ) Feb 12 18:25:48 ok Feb 12 19:42:01 will the "install" command (commonly used in a do_install()) overwrite a destinaion file if it already exists? Feb 12 19:43:02 nm. man is my friend Feb 12 19:43:18 i thought that was a yocto utility Feb 12 19:44:02 nope, standard gnu Feb 12 19:53:01 i suspect my project state is hosed and that's why i'm not getting my file inserted properly Feb 12 19:53:31 how do i force a total rebuild of everything for my image? Feb 12 19:53:58 e.g., should i rm -fR build/tmp ? Feb 12 19:54:17 i wipe tmp at least 20 times a day. doesn't hurt anything. most of it is pulled back from sstate again anyway Feb 12 19:54:26 no need to even hesitate Feb 12 19:55:37 can i trust that sstate is consistent? Feb 12 19:56:41 did you mean build/sstate-cache? Feb 12 19:57:06 yes, and yes Feb 12 19:57:34 alright.. Feb 12 19:57:40 sstate failure mode is generally a rebuild from scratch. that is, it' much, much, much more likely to rebuild from scratch than to incorrectly use existing sstate Feb 12 19:58:24 hmm. Feb 12 19:58:31 pretty rare to need to rebuild from scratch manually, though it is occasionally useful, and is easy to do. bitbake -C fetch somerecipe will build it from scratch if it's not already been built, bitbake --no-setscene will build *everything* from scrach after wiping tmp Feb 12 20:00:53 ok Feb 12 20:06:01 Hmm, isn't capabilities set on files in do_install() preserved to when the package is installed ? Feb 12 20:14:34 kroon: no extended attributes survive any packaging format (in any distribution) Feb 12 20:15:22 rburton: aha Feb 12 20:37:50 n Feb 12 22:05:31 my kernel module recipe is bombing in the do_compile. it looks like it's not grabbing the source file from the proper place in my repository (but the SRCURI looks right). where (in the tmp/ dir) do i look for where it is unpackted to? Feb 12 22:08:41 is it tmp/MACH/pn/pv-pr/module/ ? Feb 12 22:11:39 yates: see oe-core/meta/conf/bitbake.conf Feb 12 22:17:19 yates: well, tmp/work/machine/pn/pv-pr/ Feb 12 22:17:25 that's WORKDIR Feb 12 22:17:37 files are copied into there, tarballs unpacked Feb 12 23:19:33 New news from stackoverflow: SPI slave-mode on imx8qmlpddr4arm2 board using yocto Feb 13 00:19:44 New news from stackoverflow: How to export DBUS_SESSION_BUS_ADDRESS Feb 13 01:19:56 New news from stackoverflow: how to add recipe for yocto systemd service **** ENDING LOGGING AT Wed Feb 13 02:59:56 2019