**** BEGIN LOGGING AT Mon Mar 15 03:00:15 2021 Mar 15 07:41:57 good morning Mar 15 07:49:58 hi mckoan Mar 15 07:55:09 somthinf is unclear for me when I change something in my own recipe bb and make core-image.base bitbake doesn't run the modified recipe Mar 15 07:55:22 soemthing sorry Mar 15 08:20:24 yo dudX Mar 15 08:43:26 o/ Mar 15 08:45:24 jdrol: we need more info, is it a bb or a bbappend that you have and modify? what does "make core-image.base bitbake" mean to you? what's the exact command you are running? Mar 15 08:46:49 I have set a bb alone in a layer the layer is shown by bitbake-layers show-layers Mar 15 08:47:39 in the bb layer I put a SRC_URI+="file://exemplecprogram.tgz" Mar 15 08:48:35 but sometime it complain that it exemplecprogram.tgz" does not appear to contain CMakeLists.txt Mar 15 08:49:02 because on an old release I tested with makefile and he has a makekile Mar 15 08:49:33 I don't know how to clear everywhere the old recipe Mar 15 08:50:15 I do bitbake -c clear bb but it answer ERROR: Nothing PROVIDES 'exemplecprogram' Mar 15 08:50:43 jdrol: bitbake-layers show-recipes exemplecprogram Mar 15 08:51:05 if it returns nothing, your recipe is wrongly named/in the wrong path/in a layer that is not included Mar 15 08:52:02 qschulz, I don't see it with show-recipes Mar 15 08:52:36 but the old recipe in in DL and in work Mar 15 08:52:43 smurray: not expected, it should be able to handle that:( Mar 15 08:54:20 jdrol: what's the path to your exemplecprogram? what's the content of your BBLAYERS variable in build/conf/local.conf? Mar 15 08:54:46 and what's the content of BBFILES variable in your-layer/conf/layer.conf? Mar 15 08:56:03 ERROR: Nothing PROVIDES 'exemplecprogram' Mar 15 08:56:10 oups Mar 15 08:56:19 BBFILES += "${LAYERDIR}/recipes-*/*/*.bb \ Mar 15 08:56:19 ${LAYERDIR}/recipes-*/*/*.bbappend" Mar 15 08:58:07 in meta-cexample i have recipes-exemplecprogram/ Mar 15 08:58:40 in recipes-exemplecprogram/ I have exemplecprogram Mar 15 08:59:07 in exemplecprogram I have exemplecprogram_1.0.bb Mar 15 09:01:09 or the content of the recipe is whacked, e.g. PACKAGES overwritten to something unexpected? Mar 15 09:02:55 do you want to get a pastebin of bb content ? Mar 15 09:05:24 well, "want" is probably saying a bit much. :P Mar 15 09:06:19 LetoThe2nd, yes I know it is not a gift :) Mar 15 09:06:41 jdrol: triple check that bitbake-layers show-layers actually shows you meta-cexample in the list. Because it seems to me that BBFILES and your paths are correct Mar 15 09:06:56 I changed the recipe to inherit makefile to cmake Mar 15 09:08:41 meta-cexample /home/gilles/Developpements/yocto/meta-cexample 6 Mar 15 09:16:15 Hi everyone - I am writing a recipe that defines multiple packages, many of which install systemd services. However, after inheriting the systemd bbclass and defining the SYSTEMD_SERVICE variable for each of those packages, the service is not enabled in the package. Any idea what might cause that ? Mar 15 09:20:19 Yumasi: just guessing: this is not meant for separate sets of systemd services: https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes/systemd.bbclass#n9 Mar 15 09:20:47 https://controlc.com/f59735e9 Mar 15 09:21:05 for the content of layer Mar 15 09:22:25 jdrol: please send the content of your conf/layer.conf and of your recipe in a pastebin too Mar 15 09:26:05 https://controlc.com/c442b416 Mar 15 09:30:25 LetoThe2nd: since SYSTEMD_AUTO_ENABLE is set as a default value (which I have not changed; just checked with 'bitbake -e') shouldn't it propagate to the packages I define in my recipe ? I did try explicitly setting it to "enable" for each of my packages, but the services were still not enabled. Mar 15 09:30:39 qschulz, this one is full sorry https://controlc.com/94771913 Mar 15 09:32:04 jdrol: nothing weird going on for me, what's the content of your examplecprogram? Mar 15 09:32:47 * LetoThe2nd Yumasi: i've never worked in that situation so i can't really explain. its just my gut feeling that says this might not play nice with multiple packages form a recipe. hence, maybe consider splitting it. Mar 15 09:35:18 qschulz, https://controlc.com/efdbf0d5 Mar 15 09:37:45 jdrol: whacked. why are you setting PN, PV, P manually? this is probably causing your pain. Mar 15 09:38:14 jdrol: DEPENDS+=cmake is also outright wrong - unless you want cmake to be packaged and installed, too. Mar 15 09:39:03 manually setting RDEPENDS to ""... sorry, but how did that all get into a recipe? Mar 15 09:46:51 I used the yoctoproject.net generator Mar 15 09:47:23 *sigh* Mar 15 09:47:41 so you're the perfect example why i explicitly said i'm undecided about it. Mar 15 09:49:28 you shouldn't need more than this: https://controlc.com/93212a18 Mar 15 09:53:57 LetoThe2nd: I looked a bit for recipes using SYSTEMD_SERVICE in the meta-openembedded layer and I found a few that defines multiple packages with systemd services in each one, so I'd say it looks like it should work fine in my case (an exemple of one such package is the owfs one) Mar 15 09:55:57 ok thank you ... Mar 15 10:01:08 Yumasi: heh then you obviously know way more about it than I do. Mar 15 10:01:09 LetoThe2nd, who is in charge of yoctoproject.net generator ? Mar 15 10:01:53 jdrol: see the left sideboard of that site. Mar 15 10:02:35 huh ten line recipe with so many mistakes in Mar 15 10:03:11 does procps package override the busybox commands? Mar 15 10:03:25 LetoThe2nd: remove the .tgz at the end of S and make sure the tgz has only one directory in it which is named exemplecprogram Mar 15 10:03:32 jdrol: ^^^ Mar 15 10:04:23 thanks Mar 15 10:06:33 jdrol: otherwise, can only recommend starting with https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj Mar 15 10:08:02 now I got the message : Getting checksum for exemplecprogram SRC_URI entry exemplecprogram.tgz: file not found except in DL_DIR Mar 15 10:08:47 the pasth is relative from where ? Mar 15 10:09:31 LetoThe2nd: well, turns out it was a good old pebcak issue *facepalm*. Sorry for the inconvenience ! Mar 15 10:11:51 jdrol: pretty sure it's tackled in tutortials I sent you, at least in the bbappend one Mar 15 10:12:22 jdrol: recipes-exemplecprogram/examplecprogram and then in a subdir, called files, examplecprogram or examplecprogram-1.0 Mar 15 10:17:33 I know the file: I have to fille the file withe the tgz file ? Mar 15 10:18:11 Yumasi: no problem! can you please tell what the problem actually was, so we can check for it next time? Mar 15 10:21:47 I wanted to test the download from a local file and git Mar 15 10:25:05 jdrol: please watch the tutorials I send you, you can go 2x speed on it if you want, it really explains the basic concepts and recipes in Yocto. it's a good way to get started quite fast Mar 15 10:25:40 ok Mar 15 10:25:47 and i guess they also include some beer! Mar 15 10:37:35 hi , guys how can i remove getty@tty Mar 15 10:38:01 when i do a do_install_append() to systemd it tell me can remove the file because it dosen't exists Mar 15 10:39:05 can't Mar 15 10:41:00 LetoThe2nd: I was checking the package content for the symlink in /etc/systemd/system/multi-user.wants. It occured to me this symlink is actually created when the package is installed, either in an image or by a package manager. Mar 15 10:43:59 Yumasi: ah. yeah systemd.bbclass injects things into image creation :) Mar 15 10:45:35 LetoThe2nd: Exactly, with its post-inst task :) Mar 15 10:45:47 Yumasi: :) Mar 15 10:52:24 medaliyou, I think that's a systemd Mar 15 11:02:03 you override the /recipes-core/systemd/systemd-serialgetty.bb with .bbappend - then add SYSTEMD_AUTO_ENABLE_${PN} = "disable" Mar 15 11:02:59 mine is not even found due to some other setting Mar 15 11:07:05 hm it's there lol. guess I'll do the same thing Mar 15 11:33:40 hi, is there a way to disable a bbclass? For example I want to ban out of tree kernel modules so maybe I could enforce this by disabling module.bbclass? Mar 15 11:36:08 mcfrisk: you could override this class to empty it, see BBPATH for it Mar 15 11:36:12 but... why? Mar 15 11:36:14 :) Mar 15 11:36:49 LetoThe2nd, are you the man from livecoding ? Mar 15 11:39:23 jdrol: i have repeatedly been called "our man on twitch", yes :) Mar 15 11:41:10 congratulation I am watching your second video after having watching the first one.. it is clear and you are undestandable Mar 15 11:41:21 watched Mar 15 11:41:35 I like the libanswer exemple Mar 15 11:46:16 qschulz: to enforce that no out-of-tree kernel modules appear on my systems through BSP and other vendors. can't review everything anymore... Mar 15 11:47:05 out-of-tree means crappy quality, security holes, and depending on details, also issues with secure boot and signing Mar 15 11:48:34 mcfrisk: i would probably rather force-extend it (ugly hack with layer ordering, you know it.) so it errs out when yomebody tries to use it. Mar 15 11:49:38 mcfrisk: out-of-tree does not necessarily mean worse quality than vendor in-tree kernel driver :p Mar 15 11:50:41 LetoThe2nd: exactly what I suggested :) override module.bbclass and put a python __anonymous () { bb.fatal("No OoT kernel driver allowed anymore") } Mar 15 11:51:52 qschulz: oh yes it does, I've seen and I can't unsee... thanks for the tip! Mar 15 11:52:06 qschulz: you said "empty" it ;-) Mar 15 11:53:14 LetoThe2nd: jhreqwbfghuebrguygvbeurgvyurgv anyway it wouldn't build anymore :p Mar 15 11:54:04 qschulz: hehe. yeah sure. it depends a bit on where and how one wants the error to pop up. Mar 15 11:55:39 and how meaningful/helpful the error message should be :) Mar 15 11:56:02 I guess BBMASK doesn't work with bbclasses Mar 15 11:59:43 mcfrisk: not advertised at least no Mar 15 12:00:26 mcfrisk: We have done the bbclass overriding in other cases previously, too, so that works and is probably the way to go. Mar 15 12:03:20 hello, is there any work around building the esdk in dunfell with a linux-dummy kernel ? I'v only seen this https://www.mail-archive.com/yocto@lists.yoctoproject.org/msg03554.html Mar 15 12:07:58 aleblanc: maybe https://lists.yoctoproject.org/g/yocto/message/51560 FF Mar 15 12:09:37 LetoThe2nd I'll have a look. Is this something you have tried yourself ? Mar 15 12:13:50 aleblanc: nope. Mar 15 12:51:21 does somebody use ntpd? My device is not getting time. What i have done: i added ntp into the image. Modified /etc/ntp.conf -> added "server 0.de.pool.ntp.org". restarted ntpd. Still not getting time Mar 15 13:15:13 chris_ber: are you sure you have network connectivity? Mar 15 13:15:29 can you poing 0.de.pool.ntp.org from your target device? Mar 15 13:15:31 ping Mar 15 13:16:13 Just wanted to thank the OE/Yocto teams. I'm still learning, but I am feeling more and more comfortable using Yocto, and I really like it a lot! Mar 15 13:17:23 throw money! Mar 15 13:17:25 ;-) Mar 15 13:18:30 yates: yes i have (wasn't sure before). I updated the ntp.conf and received date. thx Mar 15 13:19:02 chris_ber: sure, but does anybody really know what time it is? https://www.youtube.com/watch?v=9FzCWLOHUes Mar 15 13:19:03 but i have to write a .bbappend for ntp to include public ntp servers Mar 15 13:19:37 cool Mar 15 13:20:03 yates: no sound at work ^^ Mar 15 13:20:18 now that's a serious problem! Mar 15 13:20:29 feel the same Mar 15 13:21:36 LetoThe2nd: is there a way to make a donation to Yocto/OE? Haven't found it. Though I'm learning in my free time for my own information, it's not like I get a revenue from that. So it's costing me time, but not bringing me anything in return. I feel like companies leveraging Yocto should make substantial contributions... Mar 15 13:22:41 here's a thumbs up to our Russian friends who did this song perhaps better than Chicago themselves! https://www.youtube.com/watch?v=FB-nXQc6LMU Mar 15 13:23:39 jonesv[m]: hehe nah, it was more like a joke i stole from a concert. stick around, spread the word and if you already know something that another laerning person asks, help them. thats all you need to give back. Mar 15 13:24:00 BTW freelancers using an open source project should donate a bit (or contribute code) to that project if they leverage it to get a revenue. I don't know if that's common practice 🤔. I do maintain 1-2 open source projects myself, and I can say that nobody benefiting from them contributes anything. I mainly get feature requests 😆 Mar 15 13:24:22 jonesv[m]: you can donate to openembedded though. they will happily accept! Mar 15 13:24:53 But there are surely people working actively on Yocto. Are they not being paid for that? Mar 15 13:25:29 jonesv[m]: OE is a community run organization. Donations are through Software in the Public Interest (a lot like debian IIRC). You can donate through the "Donate" button on https://www.openembedded.org/wiki/Main_Page Mar 15 13:26:20 Yocto is part of the Linux Foundation. A company can become a memeber and join the Yocto project (I think there are fees that go to fund the project through that) Mar 15 13:26:47 jonesv[m]: https://www.yoctoproject.org/ecosystem/members/ Mar 15 13:27:58 Thanks JPEW , that makes it clearer :) Mar 15 13:45:35 JPEW: are you aware of any pitfalls when building inside a docker container inside a lxc one? any hoops you had to jump through with pyrex? we're seeing strange build "halts" in our usual kas-docker environment, but pyrex seems to happily go on forever, so i'm wondering what the problem might be.. and if you've maybe already hit and solved it :) Mar 15 13:54:59 * mcfrisk lols at I_SWEAR_TO_MIGRATE_TO_PYTHON3 *snort* Mar 15 13:59:41 mcfrisk: where is that? :) Mar 15 14:00:37 Is there a trick that I should be aware of when using `bootloader --ptable gpt` (instead of mbr)? When I change that, my device just does not boot anymore and I don't get any serial output. Could it be that I need to change `part /boot --source bootimg-partition` if I use gpt instead of mbr? Mar 15 14:01:27 RP: <20210312145903.1829036-1-Martin.Jansa@gmail.com> Mar 15 14:30:27 LetoThe2nd: Is there a screen recording tool you can recommend? Mar 15 14:32:15 manuel1985: i usually just push the record button on OBS, as thats what I already have anyways. Mar 15 14:37:02 https://twitter.com/fede2_cr/status/1371168008786300928 hrhrhr Mar 15 14:45:03 LetoThe2nd, thanks! Mar 15 14:45:08 Hi Everyone, I am getting this error when i try to build my Yocto core-image-minimal Mar 15 14:45:19 https://www.irccloud.com/pastebin/XirnmaAm/ Mar 15 14:46:10 dorinda: usually a couple of lines up there is a full path to the log file of the failing task. you can look at that, or post its contents. Mar 15 14:48:26 the log was actually really long, but i will post it Mar 15 14:50:26 https://www.irccloud.com/pastebin/T11DbRcB/ Mar 15 14:51:09 oops sorry this Mar 15 14:51:37 https://www.irccloud.com/pastebin/6TCKfEYK/ Mar 15 14:54:18 these are the last lines that looked relevant to me, since the log was too long. Mar 15 14:55:29 I also noticed there's a new release of cmake 3.20.0 Mar 15 14:55:32 dorinda: that certainly looks strange. what state of poky and host distro is that? Mar 15 14:56:22 LetoThe2nd: Ya, there are a lot of quirks when building in docker. What's the halt? Mar 15 14:57:18 my distro is Ubuntu 18.04 Mar 15 14:57:40 and my poky build configuration is Mar 15 14:57:50 https://www.irccloud.com/pastebin/FpPD3swT/ Mar 15 14:58:12 dorinda: so basically latest master? Mar 15 14:58:41 Yes Mar 15 14:58:47 JPEW: i think the only real hint we're seeing is Summary: There were 2 ERROR messages shown, returning a non-zero exit code. Mar 15 14:58:47 Exception ignored when trying to write to the signal wakeup fd: BlockingIOError: [Errno 11] Resource temporarily unavailable Mar 15 14:58:47 2021-03-12 11:32:36 - ERROR - Command returned non-zero exit status 1 Mar 15 14:59:14 dorinda: hmm. Mar 15 14:59:58 dorinda: you could look into the autobuilder for a comparable setup and see if its works there (presumably) https://autobuilder.yoctoproject.org/typhoon/#/ Mar 15 15:00:39 As a headsup, I just created hardknott branches for core and updated layer.conf. This will cause the usual ripple effect through the layers Mar 15 15:01:24 RP: you get a branch, and you get a branch... everybody gets a branch!!! Mar 15 15:03:00 LetoThe2nd: Alright, will try that now thanks. Mar 15 15:03:42 JPEW: other than that, it just "stops". whatever tasks were in progress, the time counters just go up indefinetely Mar 15 15:09:46 LetoThe2nd: Hmm, not sure. We do run pyrex-in-docker (not LXC though), but it's done DooD style Mar 15 15:10:45 LetoThe2nd: TBH, it's been a while since I've done too much pyrex work (it's been working alright). I know it was really tricky to get working, and we had a lot of crashes, hangs and other things, but I don't remember them specifically anymore Mar 15 15:10:59 I don't think we ever saw the BlockingIOError specifically Mar 15 15:11:23 DooD style? Mar 15 15:12:22 LetoThe2nd: Docker-out-of-docker: http://tdongsi.github.io/blog/2017/04/23/docker-out-of-docker/ Mar 15 15:12:52 Basically, we bind mount the host docker socket into the first docker container, then pyrex uses the host docker Mar 15 15:12:58 through that socket Mar 15 15:13:11 I don't recommend it :) Mar 15 15:13:18 hmmmmm. Mar 15 15:14:02 JPEW: well the situation is that pyrex seems to work nice on lxc, i just successfully built world. now i wonder what its doing differently. Mar 15 15:14:58 LetoThe2nd: Ya, you're running docker inside of LXC? Mar 15 15:15:04 JPEW: yup Mar 15 15:17:22 LetoThe2nd: Well, you could have pyrex print out the docker command it's using... add a print before the os.execvp on line ~803 Mar 15 15:22:19 JPEW: thanks, will do. sorry, i'm a complete python idiot so i have to ask the most trivial things. Mar 15 15:50:30 LetoThe2nd: no problem :) Mar 15 15:58:31 * zeddii hits rm -rf on his build Mar 15 16:59:11 hi Mar 15 17:01:12 I'm updating the meson recipe and am wondering about the execution time replacements within `nativesdk-meson`. Mar 15 17:06:45 while it _seems_ to make sense that the cross file template is part of `nativesdk`, the replacements (i.e. `TARGET_SYS`) translate to the wrong architecture (i.e. host architecture instead of the target architecture). So the template contains `x86_64...` instead of `arm-...`. Mar 15 17:09:02 as the nativesdk variant is the only one used/customized during the SDK installation process, I assume it's supposed to be used to crosscompile for the target architecture, isn't it? Mar 15 17:09:57 zeddii: I merged perf and removal from the exclusion list, thanks! Mar 15 17:23:01 nice! Mar 15 17:30:52 i'm trying to rewrite some code that uses the sysfs gpio interface to use the new libgpiod interface - anybody have any clue how to convert numbers from one to the other? under the old interface i'm looking at gpio 44 but now i have a series of gpiochips with ios numbered from 0-31 - is old 44 new gpiochip1 13? or maybe 12? Mar 15 17:33:05 rburton: does https://autobuilder.yocto.io/pub/non-release/20210315-10/testresults/buildperf-ubuntu1604/perf-ubuntu1604_master_20210315121706_cca5433baf.html make any sense to you? Mar 15 17:34:07 halstead: was there any change to ubuntu1604 during maint on Friday? Mar 15 17:34:23 halstead: it looks like its taking 5 mins longer on builds for reasons which I can't see Mar 15 17:36:48 force a run of the last fast build to see if that repeats? Mar 15 17:39:26 @jord Mar 15 17:40:31 RP, Nothing major. Let me grab a list. Mar 15 17:41:00 @jordan: just to safe you some debugging time: keep in mind that gpios that are managed via libgpiod will return to their initial state when you close the corresponding file. Mar 15 17:41:57 t_unix: thanks! Mar 15 17:42:09 i've figured out i can look at /sys/kernel/debug/pinctrl/30330000.iomuxc/gpio-ranges to find the mapping Mar 15 17:44:13 RP, no change at all on perf-ubuntu1604. Just a reboot. For ubuntu1604-ty-1 looks like minor updates to git, python2, and libglib2. Mar 15 17:46:43 i'm confused about how to organize my layers, specifically a bsp layer. Mar 15 17:47:07 looking at http://git.yoctoproject.org/cgit/cgit.cgi/meta-arm/tree/ i see the meta-arm-bsp snuggled underneath Mar 15 17:48:12 but i thought that the bsp layer go under a specific board layer, like meta-raspberrypi Mar 15 17:49:32 help? Mar 15 17:58:30 so in one case it's organized by processor type (arm) and the other by board type (raspberrypi). so which should it be? Mar 15 17:58:41 halstead: it would have been on the perf worker. Very odd :/ Mar 15 17:59:00 rburton: I've queued a rerun Mar 15 18:05:27 yates. there's no real guidance there. It's actually more about who maintains the layers, the BSP and what structure works best for them. As long as the separate between distro/board/etc is clean, no one really cares. Mar 15 18:08:42 zeddii: thanks for clarifying Mar 15 19:23:20 * zeddii bets wherever rburton is right now. We should hide is keyboard. Mar 15 19:23:29 s/is/his/ Mar 15 19:25:47 What do you guys think of using the word "Yocto" standalone? Is this accepted yet, or is it still annoying? Mar 15 19:26:46 * derRichard smells a linux vs. gnu/linux discussion Mar 15 19:45:19 matthewcroughan_: Our branding people would be keen to point out it is Yocto Project however lots of people do abbreviate it Mar 15 19:45:41 zeddii: what happened with ross? Mar 15 19:45:51 checking your poky mailing list :D Mar 15 19:46:15 s/checking/check/ Mar 15 19:46:20 man, I can't type today Mar 15 19:47:03 ah, yocto list? Mar 15 19:47:16 yah. that. Mar 15 20:13:36 does "bitbake meta-toolchain" build a specific recipe? is there bitbake option to show the top level being buitl or other information besides -g? Mar 15 20:14:44 yates: see meta/recipes-core/meta/meta-toolchain.bb Mar 15 20:14:53 also is there a way to graphically view the task-depends.dot file? Mar 15 20:18:38 I'm trying to understand what the following means: WARNING: sdps-experimental+gitAUTOINC+4ccf80e21d-r0 do_package: Manifest /tmp/sstate-control/manifest-x86_64_x86_64-nativesdk-device-tree-lops.packagedata not found in versal_generic versal-prime versal microblazeel allarch x86_64_x86_64-nativesdk (variant '')? Mar 15 20:19:03 the package device-tree-lops is NOT a nativesdk -- and it doesn't produce packages, intentionally.. (I have do_package it it disabled) Mar 15 20:19:45 * LetoThe2nd resists to type snarky comment about understanding WARNINGs Mar 15 20:20:03 yates: depends on your definition of "graphically view" Mar 15 20:20:19 well I ASSUME the warning has a purpose, thus I want to understand what it means Mar 15 20:20:29 LetoThe2nd: i mean other than using a text editor Mar 15 20:21:15 fray: I believe there's a discussion ongoing between RP and JaMa about what looks like that issue on oe-core Mar 15 20:21:26 yates: hum well you could use a text pager. or you could hexdump it. or you could pipe the file to /dev/fb0 Mar 15 20:21:35 this is gatesgarth BTW Mar 15 20:21:48 the last one is probably the most "graphic" approach, though. Mar 15 20:22:20 fray: I merged a fix for images in master just this morning, I'm likely thinking the change should be reverted in gatesgarth and dunfell though Mar 15 20:22:39 fray: the nativesdk thing is a red herring, its just the last thing it looked for Mar 15 20:23:03 yates: you can generate a jpeg or whatever of the graph from the dot, how useful that is is debatable Mar 15 20:23:51 sakoman: can I fast track a revert of that in? Mar 15 20:24:31 smurray: fb0! fb0! Mar 15 20:24:34 ok, so the warning isn't really what it's saying, but still a warning about 'something'? Mar 15 20:24:45 this is gatesgarth from around January or so.. I'm not yet 'caught up' Mar 15 20:24:51 LetoThe2nd: do you enjoy being a jerk-off? Mar 15 20:25:22 yates: It can viewed, in pieces... Creative use of gvpr and tred (from graphwiz) can make it workable... Mar 15 20:25:25 fray: I believe it does list all the things it looked at in the message, the nativesdk is just the last one Mar 15 20:25:32 Basically I have a recipe (that produces a package) that depends on something that only deploys.. it needs to use the deployed content for part of the build.. Mar 15 20:25:34 yates: I believe graphviz has a program named "dot" that can convert it to many formats Mar 15 20:25:38 yates: yup I do! Mar 15 20:25:48 JPEW: right, that is what i suspected. Mar 15 20:26:07 so this is coming from the _user_.. Mar 15 20:26:21 since it's do_deploy only, do I need to specify the DEPENDS some other way? Mar 15 20:26:35 yates: but seriously. the pain point about dot / xdot is that on your average box they outright explode in out of memory, because the dependency graphs are way too big for them. Mar 15 20:27:11 yates: hence, the easiest way is in fact a text pager/editor. the second easiest *might* be oe-pkgdata-browser. Mar 15 20:27:25 fray: hmm, do these two recipes have packagedata tasks? Mar 15 20:27:56 yates: You can also use `bitbake -g -u taskexp ` to get a GUI to explore the task dependencies. Mar 15 20:27:59 fray: if there is just a deploy task, the first recipe probably doesn't? Mar 15 20:28:18 yates: Also true. I never look at the graphical output.... just grep around the dot file and figure out the DAG mentally (I usually only care about a very small subset anyway Mar 15 20:28:20 fray: I'm curious if it was deltask'd or not Mar 15 20:28:31 Basically recipe created a special device tree and deploys it.. recipe B needs to look at the device tree to figure out what components to process.. Mar 15 20:28:36 yes, there are deltasks.. sec Mar 15 20:29:07 do_fetch[noexec] = "1" Mar 15 20:29:07 do_unpack[noexec] = "1" Mar 15 20:29:07 do_patch[noexec] = "1" Mar 15 20:29:07 do_configure[noexec] = "1" Mar 15 20:29:07 do_install[noexec] = "1" Mar 15 20:29:07 deltask do_populate_lic Mar 15 20:29:08 deltask do_populate_sysroot Mar 15 20:29:08 do_package[noexec] = "1" Mar 15 20:29:09 deltask do_package_qa Mar 15 20:29:09 do_packagedata[noexec] = "1" Mar 15 20:29:10 deltask do_package_write_ipk Mar 15 20:29:10 deltask do_package_write_deb Mar 15 20:29:11 deltask do_package_write_rpm Mar 15 20:29:16 thats the stupidness I'm trying Mar 15 20:29:44 (basically there is no reason for this thing to ever 'package' anything', and it kept trying to package until I added all of that stuff.. (fetch/unpack/patch was for other reasons) Mar 15 20:33:03 fray: try changing that to a deltask do_packagedata Mar 15 20:33:55 change it all, or just add the deltask Mar 15 20:34:32 RP: delltask... Mar 15 20:35:24 fray: change do_packagedata[noexec] = "1" for deltask do_packagedata Mar 15 20:35:39 ok.. I think that fixed it.. but I've got a bunch of stuff to build to really see if it did.. Mar 15 20:36:38 ya, that looks like it worked Mar 15 20:39:21 fray: basically the messages means "I can't find a manifest yet this task exists. What happend?" Mar 15 20:39:32 fray: if the task doesn't exist, it can't complain about it Mar 15 20:40:25 I see.. that makes more sense Mar 15 21:16:17 zeddii: HAHAHA Mar 15 21:20:58 rburton: PUT YOUR KEYBOARD DOWN Mar 15 21:24:02 BUT I AM THE POWER OF IOTG Mar 15 21:26:59 LOL Mar 15 21:27:03 how does it feel ? Mar 15 21:27:07 the promotion and all ? Mar 15 21:34:54 being a double agent? it's great. double PAY BABY Mar 15 21:37:19 RP: fwiw i have a fix for ovmf, will be posting shortly. hoping to get an upstreamable version before posting. Mar 15 21:37:45 rburton: ah, cool! Was it a pain to fix? Mar 15 21:38:13 rburton: its not often I look at a build system and run away :) Mar 15 21:38:14 half the time was figuring out where to put a --zero option Mar 15 21:38:19 oh god its horrible! Mar 15 21:38:51 "build system" isn't being fair :/ Mar 15 21:58:54 rburton, now that autoconf update is done, can you take care of automake as well? Mar 15 22:11:43 kanavin: did you look at the reproducible pyc issues? Mar 15 22:11:59 RP: context please? Mar 15 22:12:11 RP: I did fix the ones I saw, but that was weeks ago Mar 15 22:12:32 kanavin: when we enabled reproducible builds someone looked at the pyc issues and fixed them somehow. I can't remember who it was Mar 15 22:12:51 kanavin: the context is that meson is one of two failures outside of go left and its a pyc issue Mar 15 22:14:13 I'm also looking for anyone with ruby knowledge! :) Mar 15 22:14:16 RP: right, the issue in python3 was usage of frozensets, which I fixed by deleting the offending pyc (there were only two of them) Mar 15 22:14:25 RP: and it was me :) I can look up the specific commit Mar 15 22:14:41 RP: this was reported upstream, but no progress there Mar 15 22:14:49 kanavin: I remember seeing that recently. I wonder if this meson issue is a frozenset problem Mar 15 22:15:41 RP: I can take a look, but not right now! Mar 15 22:15:42 kanavin: can we tell if https://autobuilder.yocto.io/pub/repro-fail/2021-2-25-rp/meson/diff/ was from a frozenset? Mar 15 22:16:03 kanavin: fair enough :) Mar 15 22:16:43 kanavin: I just thought I'd ask, see if we can figure out these remaining couple of issues :) frozenset is a good lead Mar 15 22:22:41 RP: if language not in {'c', 'cpp', 'fortran'}: Mar 15 22:22:45 it's a frozenset Mar 15 22:23:11 https://github.com/mesonbuild/meson/blob/master/mesonbuild/dependencies/mpi.py#L35 Mar 15 22:28:05 kanavin: thanks. Guess that is the same fix as the other then (or we patch to a list) Mar 15 22:28:28 RP: yup Mar 15 22:29:11 kanavin: thanks, I'd have taken an age to figure out the frozenset issue :) Mar 15 22:29:57 RP: I think I cobbled together a .pyc disassembler from bits and pieces on the internet, but can't for the life of me remember on which of the AB machines I left it! :) Mar 15 22:30:20 kanavin: I was just looking at https://github.com/rocky/python-decompile3 :) Mar 15 22:31:08 looks like this ruby issue could be the last one Mar 15 22:32:33 hello bitbake load in cache my recipe even if I changed the bb file when i execute bitbake recipe ... how can I force to don't read in cache ? Mar 15 22:33:02 RP: it would be nice to beat everyone else to the 100% reproducibility claim Mar 15 22:33:35 kanavin: I wondered about proposing removing go from core to help that :) Mar 15 22:34:03 RP: if nothing in core actually needs go, I can't say no Mar 15 22:34:25 kanavin: you could remove the entire of core recursively that way :) Mar 15 22:34:42 RP: you know that's my eventual goal ;) Mar 15 22:35:03 :D Mar 15 22:35:07 but really, rust is far more needed in core Mar 15 22:35:18 kanavin: I know, and that is being worked on Mar 15 22:35:25 go can be in meta-oe for now, but not rust Mar 15 22:35:57 RP: yeah, I've followed Randy's work, it's getting there nicely Mar 15 22:36:24 kanavin: it was close for 3.3 but just not quite there Mar 15 22:36:41 go has the problem of being a 'google thing', which I can imagine some people have a problem with Mar 15 22:37:16 kanavin: we deal with each case as needed, they're all a little different Mar 15 22:37:58 RP: I mean, go as implementation language for something open source - e.g. in the linux kernel Mar 15 22:38:06 or core libraries Mar 15 22:38:10 kanavin: oh, yes, definitely **** ENDING LOGGING AT Tue Mar 16 03:03:12 2021