**** BEGIN LOGGING AT Mon Oct 28 02:59:58 2019
Oct 28 07:03:57 <yocti> New news from stackoverflow: How to add support for screen timeout while running a qt5.9 app on linuxfb platform? <https://stackoverflow.com/questions/58337249/how-to-add-support-for-screen-timeout-while-running-a-qt5-9-app-on-linuxfb-platf>
Oct 28 08:04:44 <armpit> RP, have we seen "devtool.DevtoolAddTests.test_devtool_add_git_local: FAILED"
Oct 28 08:06:44 <Crofton> Failed!
Oct 28 08:06:56 <Crofton> Shouldn't you be watching a keynote?
Oct 28 08:13:03 <LetoThe2nd> already dropped the first box of giveaways at the yocto booth so be sure to grab some
Oct 28 08:15:32 <Crofton> Thanks
Oct 28 08:25:31 <armpit> getting top Euro for them too
Oct 28 08:35:09 <mcfrisk> I need a lot of yocto/oe stickers. Managed to update a project to 3.0 zeus at the same time as official release announcement came out :)
Oct 28 08:49:05 <LetoThe2nd> mcfrisk: you can have glitter smiley stickers!
Oct 28 08:51:11 <LetoThe2nd> gkh has a beard!
Oct 28 08:51:55 * florian too ;)
Oct 28 08:52:05 <florian> scnr
Oct 28 08:53:34 <LetoThe2nd> florian: yeah me too but i've had for ~20years, so thats little surprise
Oct 28 08:55:06 <florian> LetoThe2nd: Same for me... nevermind :)
Oct 28 08:55:14 <LetoThe2nd> florian: you around?
Oct 28 08:56:38 <florian> LetoThe2nd: Nope... I would like to but I currently have to care about three kids and three ongoing customer projects.
Oct 28 08:57:23 <LetoThe2nd> florian: aw. so no glitter sticker for you!
Oct 28 08:57:43 <florian> eeks!
Oct 28 09:00:43 * florian gets another coffee... I'm not used to get up with the kids at 6am
Oct 28 09:12:24 <pharaon2502> hi, does aynone know, for bitbake -c populate_sdk, how to specify in the image/config if i want to package the compiler for target?
Oct 28 09:14:21 <kroon> pharaon2502, see TOOLCHAIN_TARGET_TASK I think
Oct 28 09:34:24 <yocti> New news from stackoverflow: Should I set src_uri? <https://stackoverflow.com/questions/58588135/should-i-set-src-uri>
Oct 28 09:59:35 <Foxdie> Good morning all - I'm struggling with a Yocto build with Bitbake - can anyone advise how I can globally enable building of dev packages in my image? (Or if this isn't the correct channel to ask, please advise me where I should ask?)
Oct 28 10:32:15 <alessioigor> Foxdie: (EXTRA_)IMAGE_FEATURES += "dev-pkgs" in the image recipe.
Oct 28 10:32:28 <alessioigor> good morning to all!
Oct 28 10:32:43 <Foxdie> It's that simple?? I wish that was on all the Google hits I trawled :D
Oct 28 10:32:48 <Foxdie> Also good morning :)
Oct 28 10:33:35 <Foxdie> I may have to re-evaluate my life if I couldn't find that on Google lol
Oct 28 10:35:54 <neverpanic> Foxdie: The dev packages are always built, they're just not in the image by default (but I guess you want them in the image?)
Oct 28 10:36:17 <Foxdie> Exactly - specifically pkgconfig stuff for OpenCV crap :)
Oct 28 10:36:40 <Foxdie> I'm gonna try now with alessioigor's suggestion
Oct 28 10:37:07 <neverpanic> You sure you don't want to use an SDK?
Oct 28 10:37:15 <wertigon> Hey, bit of a conundrum here; bitbake testimage -c populate_sdk compiles but at the last possible moment returns with
Oct 28 10:38:29 <wertigon> Problem: package perl-module-encoding-5.24.4-r0.armv7at2hf-neon requires perl-module-config, but none of the providers can be installed
Oct 28 10:39:19 <wertigon> And tells me to not install either target-sdk-provides-dummy or packagegroup-test-devtools, both seems like I should have...?
Oct 28 10:40:53 <Foxdie> neverpanic; Admittedly no I'm not sure, what benefits would that give me? I'm trying to build a production-ready image that includes dev libs for OpenCV but there are too many to list, so I was just gonna blanket enable dev pkgs as a test build and see how much larger it made my image
Oct 28 10:47:37 <erbo> Foxdie: Using an image + SDK would nicely separate the image from the tools needed to build for that image. The SDK comes in form of an installer that installs the toolchain needed to build for the image, a sysroot containing all the dev-pkgs etc.
Oct 28 10:48:08 <erbo> Otherwise your production image would need to contain compilers etc
Oct 28 10:54:56 <Foxdie> The developers I'm building the image for have already said they want the compilers in the image, "yay"
Oct 28 10:55:15 <Foxdie> I'll do some more digging though, thanks chaps :)
Oct 28 10:57:40 <neverpanic> Foxdie: you can just install opencv-dev into your image, the -dev packages should depend on all the other -dev packages they require.
Oct 28 10:58:58 <Foxdie> o.O without having to manually unset FILES_${PN}-dev ?
Oct 28 11:12:05 <neverpanic> No, don't modify FILES_${PN}-dev. Just use IMAGE_INSTALL += "${PN}-dev" where ${PN} = packagename of the package you want
Oct 28 11:23:08 <LetoThe2nd> would anybody please walk up to the ibm booth and ask for those cool glitter stickers? ;-)
Oct 28 11:24:40 <kanavin> I am in the ELC-e as well :)
Oct 28 11:25:26 <kanavin> if anyone wants to meet up, use their official app/serbivr (brella)
Oct 28 11:25:43 <LetoThe2nd> kanavin: i don't remember giving you one yet. so, at one of the ibm demo stands, there is a card of the deposited. if you're in for some fun, ask ibm for a glitter sticker!
Oct 28 11:25:55 <litb> hello folks
Oct 28 11:26:11 <kanavin> a card of the deposited?
Oct 28 11:27:25 <LetoThe2nd> kanavin: a card of glitter stickers. because embedded linux needs more glamour! and whats more glamour than glitter?!?
Oct 28 11:27:45 <LetoThe2nd> kanavin: also, be sure to get some weihenstephaner swag at the yocto booth :)
Oct 28 11:28:17 <kanavin> I'll try to get there on my way to lunch :)
Oct 28 11:28:37 <LetoThe2nd> lunch plans already, anybody?
Oct 28 11:28:58 <kanavin> yes - booked a table at 33 cite for 13:00
Oct 28 11:29:07 <Foxdie> neverpanic; Thanks bud, I'll try that :)
Oct 28 11:29:12 <kanavin> the only place around that is michelin listed :)
Oct 28 11:30:06 <LetoThe2nd> kanavin: whoa
Oct 28 11:30:22 <kanavin> it's not super pricey!
Oct 28 11:30:42 <kanavin> and if I am in Lyon, quality food should be had :)
Oct 28 11:31:43 <LetoThe2nd> good plan
Oct 28 11:32:05 <LetoThe2nd> should head there too some time
Oct 28 11:32:50 <rburton> damn you all
Oct 28 11:33:02 <rburton> i might have to drop from irc before i explode with jealousy
Oct 28 11:33:13 <kanavin> rburton, you have a michelin place in Truro too
Oct 28 11:33:28 <kanavin> I did look it up :) as I still plan to get there at some point
Oct 28 11:34:08 <LetoThe2nd> rburton: pics or it didn't happen
Oct 28 11:34:13 <LetoThe2nd> rburton: (the explosion!)
Oct 28 11:34:48 <Foxdie> Oh my :)
Oct 28 11:35:00 <Foxdie> Jealousy-induced spontaneous combustion
Oct 28 11:36:14 <kanavin> rburton, https://www.viamichelin.com/web/Restaurant/Truro-TR1_3BZ-Tabb_s-34dy02j
Oct 28 11:36:21 <kanavin> do you know it? :)
Oct 28 11:37:16 <rburton> i do but wasn't aware it was in there!
Oct 28 11:37:25 <rburton> from the outside it looks a bit dismal ;)
Oct 28 11:38:04 <kanavin> "Tucked away in the backstreets you’ll find this unassuming former pub which comprises a series of homely cream and lilac rooms. The owner works alone in the kitchen, cooking refined, classically based dishes with masculine flavours. Sauces are a strength and the deep-fried courgettes are a must. "
Oct 28 11:38:08 <Crofton> LetoThe2nd, I told tracey to get a glitter sticker at ibm
Oct 28 11:38:14 <kanavin> unassuming, should learn that word ;)
Oct 28 11:38:28 <wertigon> Hmmm... Ok, so bitbake myimage -c output_sdk says nativesdk-autoconf requires nativesdk-perl
Oct 28 11:38:42 <wertigon> I can build image just fine though.
Oct 28 11:38:49 <rburton> kanavin: https://www.google.com/maps/@50.2625955,-5.057198,3a,75y,293.1h,85.16t/data=!3m6!1e1!3m4!1sal0eAcqXApDv4yX4PdPIgA!2e0!7i13312!8i6656
Oct 28 11:38:56 <LetoThe2nd> Crofton: awesome!
Oct 28 11:38:58 <wertigon> How can I fix this?
Oct 28 11:39:27 <rburton> wertigon: do you mean -c populate_sdk
Oct 28 11:39:40 <wertigon> rburton: Yeah
Oct 28 11:40:01 <wertigon> Sorry, typing in commands from my build machine which is next to me
Oct 28 11:40:02 <rburton> wertigon: recipes-core/meta/nativesdk-buildtools-perl-dummy.bb is meant to provide nativesdk-perl and friends so the question is why isn't that being recognised
Oct 28 11:40:35 <kanavin> rburton, the signs do imply 'fancy' though
Oct 28 11:41:56 <rburton> guess so
Oct 28 11:42:02 <wertigon> rburton: This is not exactly a stock build, it's based on a TI SDK and then we modify that a bit further as well
Oct 28 11:42:18 <rburton> i'd be asking TI what they broke :)
Oct 28 11:42:21 <wertigon> I have no clue what I'm doing right now :P
Oct 28 11:42:34 <wertigon> Yeah, but the TI works
Oct 28 11:42:41 <wertigon> But not our extensions to it
Oct 28 11:44:52 <LetoThe2nd> sounds like you should ask yourself what you broke </SCNR>
Oct 28 11:46:38 <rburton> wertigon: try bitbake nativesdk-buildtools-perl-dummy and see if that produces any errors
Oct 28 11:47:23 <wertigon> None
Oct 28 11:47:32 <rburton> (we use host perl in SDKs, that package just has lots of provides)
Oct 28 11:47:53 <wertigon> Hmm...
Oct 28 11:48:06 <wertigon> Could it be that I'm missing perl packages? Ubuntu 18.04
Oct 28 11:48:15 <rburton> no
Oct 28 11:48:16 <wertigon> In host that is
Oct 28 11:49:52 <LetoThe2nd> kanavin: so you're already there, or what?
Oct 28 11:53:42 <wertigon> Hmm, can I simply add the package to the SDK output only?
Oct 28 11:54:17 <wertigon> And how do I do that?
Oct 28 12:00:31 <Crofton> https://lists.yoctoproject.org/pipermail/yocto-security/2019-October/thread.html
Oct 28 12:00:35 <Crofton> hey this works!
Oct 28 13:02:56 <dallas> Attempting to build a package (chrony) which requires libcap in order to drop root priviledges.
Oct 28 13:03:20 <dallas> libcap is built by default and added to rootfs.
Oct 28 13:04:22 <dallas> But chrony configure apparently doesn't grok that, because it disables the drop_priviledges feature.
Oct 28 13:04:58 <dallas> Not sure how to get around this or even how to search for a solution.
Oct 28 13:05:04 <dallas> Any ideas?
Oct 28 13:05:29 <yann> I have a package (gstreamer-plugins-base) which compiles properly, but fails to translate its FILERDEPENDS (on libGES2.so) into proper RDEPENDS (on my heavily modified sunxi-mali) - any hint where to look ?  pkgdata seems correct at first glance but I probably overlooked something...
Oct 28 13:06:17 <yann> dallas: did you add libcap to the package's DEPENDS ?  maybe a PACKAGECONFIG item deals with that already ?
Oct 28 13:09:04 <dallas> I didn't because I found that it was being built by the image target already. Lemme go try that - Thx
Oct 28 13:10:46 <Foxdie> Well, adding "dev-pkgs" to my extras definitely worked.. OS image jumped from 332 MB compressed to 485 :D  (but that was to be expected)
Oct 28 13:10:54 <yann> dallas: recipes are built with custom sysroot each nowadays, you need to have DEPENDS setup so as to pull everything you require
Oct 28 13:25:07 <dallas> Thanx yann. I was in the weeds entirely. Turns out the recipe disables the root drop feature by default.
Oct 28 13:25:57 <yann> dallas: you're welcome :)
Oct 28 13:26:21 <dallas> Should have looked at it to begin with but it didn't occur to me that it would do that. Comments suggest it was done
Oct 28 13:27:08 <dallas> because additional changes to the conf file would be needed ... well yeah???
Oct 28 13:28:35 <dallas> Anyway, thx for getting straightening me out.
Oct 28 13:28:51 <LetoThe2nd> but isn't chrony in oe anyways?
Oct 28 13:29:24 <LetoThe2nd> http://layers.openembedded.org/layerindex/recipe/51402/
Oct 28 13:29:40 <Crofton> LetoThe2nd, can you have the YP twitch account link to the youtube archives?
Oct 28 13:30:04 <Crofton> We talked about how great you are in the AB meeting. And are you giving a talk?
Oct 28 13:30:33 <LetoThe2nd> Crofton: no, i am a regular attendee this time :)
Oct 28 13:31:00 <Crofton> I looked at the twitch account and couldn't find the archives :)
Oct 28 13:31:11 <LetoThe2nd> Crofton: however i am willing to do ad-hoc speaking/ranting/headbanging/drinking on request :)
Oct 28 13:31:16 <Crofton> ok, someone mentioned you were iregualr
Oct 28 13:31:26 <LetoThe2nd> Crofton: I can link to it, sure.
Oct 28 13:31:31 <LetoThe2nd> Crofton: give me a few
Oct 28 13:31:34 <Crofton> from twitch
Oct 28 13:31:40 <LetoThe2nd> yeah got that
Oct 28 13:31:49 <Crofton> I just looked and then thought oh search youtube
Oct 28 13:35:51 <LetoThe2nd> Crofton: I liked directly to the archive from the Schedule block and moved it to a more prominent position. if andreea makes an additional banner for it, we can use that. ok?
Oct 28 13:36:12 <Crofton> We can tal with her later
Oct 28 13:37:03 <LetoThe2nd> if you look now, a link to the yt playlist should be about the first thing that you see when you look beneath the video player, right?
Oct 28 13:37:42 <Crofton> much better
Oct 28 13:38:33 <LetoThe2nd> great
Oct 28 13:38:44 <Crofton> this internet stuff is hard
Oct 28 13:39:11 <LetoThe2nd> oh yeah
Oct 28 13:48:40 <dallas> @LetoThe2nd: Yeah chrony is in meta-networking, but it explicitly disables the root drop feature.
Oct 28 13:49:19 <dallas> Normally one might not care. However, if you are attempting to clock chrony from gpsd
Oct 28 13:49:56 <dallas> Using the SOCK interface and gpsd *does* drop root (to user nobody for instance) then you have a problem
Oct 28 13:50:55 <dallas> Since chronyd runs as root it creates a socket owned by root and when gpsd subsequently runs and attempts to write to that socket ...
Oct 28 13:51:13 <dallas> oops ... not priviledges to do that.
Oct 28 13:51:29 <LetoThe2nd> dallas: ah ok. well then, what i have to point out is that there has been quite some churn on libpcap lately, so before digging into that direction it is highly advisable to read the mailing list
Oct 28 13:52:03 <dallas> K.  I'll definitely do that.
Oct 28 13:52:59 <dallas> I have a target that is *never* connected to a network where it can reach an ntp server
Oct 28 13:53:23 <dallas> It can only get time from an RTC *or* GPS.
Oct 28 13:53:40 <inf> Hey. I'm runnning warrior but I need to pull in some newer version of u-boot (http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/meta/recipes-bsp/u-boot/u-boot_2019.10.bb?id=fdc1ccbb969b2fe80f0b32dce818ee237451c816) - what would be a proper way of doing this? pulling in the changes into my own layer?
Oct 28 13:53:59 <dallas> So I want chrony to manage this situation automajikly.
Oct 28 13:54:50 <dallas> However, I could get chronyd to sync with gpsd. Could see the socket created and gpsd connecting to it.
Oct 28 13:55:04 <LetoThe2nd> inf: yep, this is a classic candidate for a recipe in your BSP layer
Oct 28 13:55:50 <dallas> Rog. Gotta go work that out now.
Oct 28 14:05:11 <yocti> New news from stackoverflow: How to set i.p for usb0 before imx board boots? <https://stackoverflow.com/questions/58592145/how-to-set-i-p-for-usb0-before-imx-board-boots>
Oct 28 14:34:18 <alessioigor> rburton: I finally able to reproduce a glib (doc) build fail with "/lib/ld.so.1: No such file or directory" on powerspe using Poky. I'm going to open a bug which information/file do you sugget that I should provide?
Oct 28 14:54:35 <rburton> alessioigor: powerpc host or target?
Oct 28 14:55:17 <alessioigor> rburton: target. It works on powerpc it fails on powerpcspe only.
Oct 28 14:55:37 <rburton> alessioigor: is ld.so.1 in a different path in the sysroot?
Oct 28 15:04:12 <alessioigor> rburton: I'm not sure which sysroot I should look into. :-(
Oct 28 15:04:54 <rburton> alessioigor: the recipe-sysroot in the glib-2.0 workdir, if thats the recipe that breaks
Oct 28 15:08:41 <alessioigor> rburton: recipe-sysroot is completely empty on ppc7400, Instead on ppce500v2 ld.so.1 is in /usr/lib (and not in /lib). FIY I have usrmerge enabled.
Oct 28 15:09:11 <rburton> its only empty if the recipe hasn't been built or you have rm_work enabled
Oct 28 15:09:37 <rburton> bitbake glib-2.0 -c addto_recipe_sysroot should do the trick to force population
Oct 28 15:13:15 <alessioigor> rburton: I made prepare_recipe_sysroot
Oct 28 15:13:37 <alessioigor> rburton: The file is in the same place.
Oct 28 15:13:58 <alessioigor> find tmp/work/*/glib-2.0/1_2.62.1-r0/recipe-sysroot -iname ld.so.1
Oct 28 15:14:04 <alessioigor> tmp/work/ppc7400-poky-linux/glib-2.0/1_2.62.1-r0/recipe-sysroot/usr/lib/ld.so.1
Oct 28 15:14:06 <rburton> sounds like a good starting point, can you dump all of that into a bug
Oct 28 15:14:11 <rburton> what machines, etc.
Oct 28 15:14:18 <alessioigor> rburton: Ok.
Oct 28 15:14:18 <rburton> bonus points for replicating without usrmerge
Oct 28 15:14:37 <alessioigor> I start immediately.
Oct 28 15:59:46 <JPEW> litb: I just pushed a few meta-mingw patches you might find interesting
Oct 28 16:00:06 <JPEW> litb: Nothing terribly exciting
Oct 28 16:03:15 <litb> JPEW, ah dbus
Oct 28 16:03:23 <litb> no, that's not you
Oct 28 16:05:37 <lpapp> hi, where does the kernel $PN come from with kernel name and version?
Oct 28 16:05:44 <lpapp> is it from inheriting kernel?
Oct 28 16:05:51 <litb> oh no, it is you ! according to your mail address :)
Oct 28 16:05:53 <lpapp> Because I do not have any special $PN setting
Oct 28 16:06:10 <litb> JPEW, nice, Qt can also make use of dbus on mingw
Oct 28 16:07:24 <JPEW> litb: I was in a hurry when I chose that email address... but I think I'm stuck with it for time being :)
Oct 28 16:07:59 <rburton> lpapp: how is the kernel PN different?
Oct 28 16:08:11 <rburton> its linux-yocto or linux-intel here, depending on machine
Oct 28 16:11:30 <rburton> JPEW: the dbus FILES thing says we need an EXEEXT in bitbake.conf and encourage poeple to use it
Oct 28 16:11:56 <JPEW> rburton: agreed
Oct 28 16:12:58 <JPEW> rburton: All three patches are on master-next; I'd like to get them in an AB build before pushing to master
Oct 28 16:13:16 <rburton> i can fire just the mingw bit now
Oct 28 16:13:20 <JPEW> Thanks
Oct 28 16:14:43 <JPEW> Although, I suspect none of those recipes actually build on the AB, so maybe it doesn't matter?
Oct 28 16:14:49 <rburton> JPEW: https://autobuilder.yoctoproject.org/typhoon/#/builders/89/builds/1176
Oct 28 16:14:50 <rburton> ha
Oct 28 16:14:58 <rburton> maybe we need to expand the test matrix then
Oct 28 16:15:13 <rburton> explicitly adding some further targets would be sufficient
Oct 28 16:15:28 <rburton> it just does a core-image-minimal sdk, we can throw more recipes into the mix
Oct 28 16:15:45 <JPEW> What's the best way to add more recipes?
Oct 28 16:16:46 <rburton> well, either change yocto-autobuilder-helper directly, or make a packagegroup in meta-mingw which y-a-h then builds
Oct 28 16:17:11 <rburton> adding a packagegroup to represent recipes that are known to build fine and are supported would be sensible
Oct 28 16:17:27 <rburton> then y-a-b can build that and a minimal sdk
Oct 28 16:17:56 <rburton> actually an image might be neater
Oct 28 16:18:23 <rburton> with bonus recipes added to toolchain_host_targets or whatever the variable is
Oct 28 16:18:43 <rburton> then y-a-h just builds the sdk for that image
Oct 28 16:18:51 <JPEW> Ok, I'll do that
Oct 28 16:19:42 <rburton> cool
Oct 28 16:28:20 <litb> JPEW, why does the SDK need dbus?
Oct 28 16:29:57 <LetoThe2nd> btw, anybody seing libcap-ng build failures on master?
Oct 28 16:30:27 <LetoThe2nd> playing around a bit with building distro-less for the bbb, it gives me a sad face
Oct 28 16:30:50 <tgamblin> LetoThe2nd: yes, tlwoerner submitted a patch to fix I think
Oct 28 16:31:03 <LetoThe2nd> tgamblin: ah yes
Oct 28 16:38:15 <LetoThe2nd> tgamblin: yeah adding the LDFLAGS append seems to fix it.
Oct 28 16:43:17 <JPEW> litb: Our application build converts the XML file to code, and we cross compile from Windows
Oct 28 16:44:31 <litb> ah, I see
Oct 28 16:46:06 <JPEW> litb: Perhaps there is a better way to excise the dbus dependency from the code generation tools, but dbus supports MinGW anyway AFAIK
Oct 28 16:52:25 <alessioigor> rburton: Re: build glib on powerpcspe. It builds if I remove usrmerge from DISTRO_FEATURES.
Oct 28 16:52:34 <rburton> alessioigor: good to know
Oct 28 16:52:42 <rburton> hardcoded path somewhere
Oct 28 16:52:47 <rburton> include that fact inthe bug report
Oct 28 16:52:57 <rburton> and attach a full compile log
Oct 28 16:54:34 <lpapp> rburton: it contains also the version
Oct 28 16:54:55 <rburton> lpapp: sounds like something your kernel recipe is doing.  bitbake -e virtual/kernel will tell you
Oct 28 16:55:19 <lpapp> rburton: but the yocto is the same
Oct 28 16:55:23 <lpapp> that is what I adopted
Oct 28 16:57:29 <rburton> lpapp: # pre-expansion value:
Oct 28 16:57:29 <rburton> #   "${@bb.parse.vars_from_file(d.getVar('FILE', False),d)[0] or 'defaultpkgname'}"
Oct 28 16:57:30 <rburton> PN="linux-intel"
Oct 28 16:57:53 <rburton> (ditto for linux-yocto)
Oct 28 16:58:34 <rburton> so if PN really contains a version then thats something your kernel is doing. maybe the recipe has the version in the filename PN part
Oct 28 16:59:18 <rcrudo> I'm creating a recipe for cpp_redis and currently it looks like this: https://pastebin.com/raw/xEcpsFMt
Oct 28 16:59:52 <rcrudo> the build is failing because the submodule is not being cloned although I'm using gitsm://
Oct 28 17:00:13 <rcrudo> what am I missing?
Oct 28 17:03:44 <lpapp> rburton: I am confused
Oct 28 17:03:46 <lpapp> kernel-image-4.4.26-yocto-standard_4.4.26+git0+3030330b06_ca6a08bd7f-r0_beaglebone.ip
Oct 28 17:04:04 <lpapp> what I meant is the generated package name contains both PN and PV
Oct 28 17:04:10 <lpapp> maybe I should not have said PN
Oct 28 17:04:32 <lpapp> What I want to see is: kernel-image_4.4.26+git0+3030330b06_ca6a08bd7f-r0_beaglebone.ipk
Oct 28 17:04:50 <lpapp> or kernel-image-yocto-standard_4.4.26+git0+3030330b06_ca6a08bd7f-r0_beaglebone.ipk, whatever
Oct 28 17:05:04 <lpapp> minus version either way, so opkg replaces the kernel properly on upgrade rather than accumulating them
Oct 28 17:13:25 <xthunderheartx> Question related to PACKAGECONFIG. I understand it should be a comma separated quad which something like this:
Oct 28 17:14:02 <xthunderheartx> PACKAGECONFIG[foo] = "-with,without,depend,rdepend"
Oct 28 17:14:59 <xthunderheartx> But in the chrony recipe I see this: PACKAGECONFIG[privdrop] = ",--disable-privdrop,libcap"
Oct 28 17:15:31 <xthunderheartx> Since there are only three elements, how does that work?
Oct 28 17:16:24 <xthunderheartx> I need chrony not to be compiled with "--disable-privdrop"
Oct 28 17:17:15 <rburton> lpapp: all package filenames contain PN and PV
Oct 28 17:17:27 <rburton> the kernel ones are funky because it embeds *more* versions
Oct 28 17:18:09 <xthunderheartx> Which by default it *is* compiled with --disable-privdrop
Oct 28 17:18:50 <rburton> xthunderheartx: add privdrop to PACKAGECONFIG
Oct 28 17:19:07 <rburton> the recipe should have --enable-privdrop or something in the first item for clarity
Oct 28 17:19:20 <rburton> instead of passing ''
Oct 28 17:21:46 <xthunderheartx> The configure script looks like it would yakk on "--enable-privdrop".  It is enabled by default and must be explicitly *disabled* with "--disable-privdrop"
Oct 28 17:22:25 <xthunderheartx> So if an element in the quad is left off it just get null ''?
Oct 28 17:23:30 <rburton> yeah the item is there, its just ''
Oct 28 17:23:33 <rburton> see the leading comma
Oct 28 17:23:49 <rburton> between " and , is the "pass this when the option is enabled string"
Oct 28 17:27:01 <xthunderheartx> Got it.  And that must me that the pattern: "with,without,depends" just means RDEPENDS is '' (empty)
Oct 28 17:27:23 <xthunderheartx> *mean
Oct 28 17:28:25 <xthunderheartx> Example: PACKAGECONFIG[libcap] = "--with-libcap,--without-libcap,libcap"
Oct 28 17:38:15 <rburton> xthunderheartx: correct
Oct 28 17:38:24 <rburton> rdepends is often empty
Oct 28 18:04:23 <lpapp> rburton: no, other packages are name only
Oct 28 18:04:34 <lpapp> ergo if you update busybox, you do not get two busiboxes like with the kernel
Oct 28 18:04:43 <lpapp> what I want is I do not want to end up with multiple kernels
Oct 28 18:04:54 <lpapp> I would like the kernel to behave exactly like any other packages
Oct 28 18:04:58 <lpapp> upgrade means replace
Oct 28 18:05:01 <lpapp> not accumulate
Oct 28 18:05:12 <lpapp> in our use case, anywayu
Oct 28 18:05:14 <lpapp> anyway*
Oct 28 18:26:51 <frosteyes> Hi folks. I know that overrides can be used like "DEPENDS_machinename" in a bbappend file, but is it possible to have a expression for machine name. E.g. all machines starting with "p1-.*"
Oct 28 18:27:07 <frosteyes> Alternative have a bbappend for specific machines.
Oct 28 18:28:24 <tgamblin> ls
Oct 28 18:28:30 <tgamblin> oops!
Oct 28 19:02:29 <rburton> lpapp: right thats something your kernel recipe is doing
Oct 28 19:29:07 <rburton> lpapp: fwiw setting KERNEL_PACKAGE_NAME in the recipe will override the default PN
Oct 28 19:30:06 <fury> is there a way to tell the package names and versions that a particular image depends on? e.g. for a gemfile.lock it's fairly easy to tell what an application was built with. bitbake -s appears to be the closest thing to what I'm looking for, only it doesn't seem to filter by a particular image, it's showing all recipes even if those recipes wouldn't be used
Oct 28 19:31:10 <fury> i can use bitbake -g imagename to generate the graphviz for a dependency graph but it only includes package name, not version
Oct 28 19:31:31 <rburton> fury: build the image, then look at the manifest in deploy
Oct 28 19:32:30 <rburton> that tells you what is in the image
Oct 28 19:32:55 <rburton> *does not include* things that were needed on the build host but didn't get into the image
Oct 28 19:33:02 <rburton> not sure if thats important or not, you didn't say
Oct 28 19:33:34 <fury> i think for this purpose just the dependencies for the image would be sufficient
Oct 28 19:33:41 <fury> looking for the manifest now
Oct 28 19:35:23 <fury> i am attempting to make a suggestion to GitLab to add support to their "dependency scanning" feature to include support for scanning yocto / openembedded / bitbake projects, so looking for whatever the equivalent would be of the gem file or package.json or python requirements.txt
Oct 28 19:38:49 <fury> that manifest looks pretty good as an example, i'll go with that. thanks! :)
Oct 28 19:39:07 <fury> i assume there's no way to obtain such a manifest before building?
Oct 28 20:05:23 <RP> JPEW: There were some reproduc fails in the last build. Hash equiv was enabled
Oct 28 20:07:06 <JPEW> RP: Ok. Were my patches in place to capture them?
Oct 28 20:09:06 * JPEW needs more than 16 GB RAM :(
Oct 28 20:09:57 <rburton> fury: the manifest is exactly what goes into the image so you need the image to know that
Oct 28 20:10:54 <fury> makes sense. thanks :D
Oct 28 20:50:07 <RP> JPEW: yes. Not looked to see what they were but patches in place
Oct 28 20:50:29 <JPEW> Ok. Do you have a link to the failed AB? I'll ssh in and take a look
Oct 28 20:52:47 <RP> JPEW: https://autobuilder.yoctoproject.org/typhoon/#/builders/83/builds/449 - all four selftests failed, at least two I looked  at had reproducible failures.
Oct 28 20:52:59 <JPEW> RP: Thanks
Oct 28 20:53:20 <RP> JPEW: sorry, iirc connection here is laggy :/
Oct 28 20:53:38 <JPEW> RP: No problem
Oct 28 20:53:42 <LetoThe2nd> RP: at the speaker thingy? :P
Oct 28 20:54:15 <RP> LetoThe2nd: no
Oct 28 20:59:01 <LetoThe2nd> RP: crappy hotel wifi?
Oct 28 20:59:31 <RP> LetoThe2nd: yes, and remote system running the irc ui
Oct 28 21:00:16 <LetoThe2nd> RP: same concept here (remote tmux session) but it seems to be fine
Oct 28 21:02:02 <RP> LetoThe2nd: this is sometimes one, sometimes laggy, I suspect the wifi
Oct 28 21:02:12 <RP> er, sometimes ok
Oct 28 21:04:08 <RP> JPEW: looks like the repro save code failed  :/
Oct 28 21:04:32 <RP> perhaps a test for the test failure case? :)
Oct 28 22:18:39 <mischief> is there a way to depend on a package and not the corresponding -dev package?
Oct 28 22:20:04 <LetoThe2nd> mischief: have the package provide something, like virtual/mything, and then depend on that. should do the trick, if i understood it right.
Oct 28 22:22:45 <rburton> mischief: runtime depends, just depend on the package.  buildtime, that is meaningless.
Oct 28 22:22:48 <rburton> what do you *want to do*
Oct 28 22:23:24 <mischief> i am trying to work around a bug in the build system of this old and broken u-boot :)
Oct 28 22:24:15 <rburton> build time depends are recipe-scoped and the pn/-dev split is meaningless there. you depend on a recipe, and you get its sysroot.
Oct 28 22:24:42 <mischief> aha. okay.
Oct 28 22:55:30 <rreignier> Hi! I am new to Yocto customization. On x86-64, I would like to move the kernel (bzImage) to the '/' rootfs in order to have several root partitions (updates with RAUC) and let the bootloader (grub) on the /boot partition. What is the recommended way of changing the kernel destination? (I have tried to set "KERNEL_IMAGEDEST" to "" (empty string) but the QA checks outputs some errors).
Oct 28 22:57:00 <rreignier> And the recommended way of replacing the grub.cfg? Because wic does not seem to take it into account when generating the bootimg-efi.
Oct 28 22:58:01 <rreignier> For now, I have tried this recipe: https://github.com/romainreignier/meta-rauc-qemux86/blob/030664873129cfc1f38a2e97d7884115c4b0e31e/recipes-bsp/rauc-qemu-grubconf/rauc-qemu-grubconf_0.1.bb
Oct 28 22:58:08 <rreignier> Thank you
Oct 28 23:45:12 <mischief> qualcomm sucks
Oct 28 23:45:16 <mischief> :(
**** ENDING LOGGING AT Tue Oct 29 03:02:49 2019