Friday, 2018-09-07

*** lexano <lexano!> has quit IRC00:04
*** dc13ff <dc13ff!uid190567@gateway/web/> has quit IRC00:07
*** lexano <lexano!> has joined #yocto00:18
rettichschnidikergoth: Yes, it does apply e.g. the patches appended to SRC_URI from withing the .bbappend file. But when I execute "devtool edit-recipe swupdate" it always opens up "just" the .bb file and does not e.g. merge them. But I guess I should just simply edit the .bbappend file directly then.00:19
khemRP: we can pass CFLAGS="-D__WORDSIZE=64" or CFLAGS="-D__WORDSIZE=32" depending upon bitness00:27
*** rubdos <rubdos!> has quit IRC00:31
rburtonkhem: got a better fix00:33
khemrburton: cool __WORDSIZE is glibc-specific I love that phrase00:43
khemrburton: btw. you also need a clang specific fix00:43
khemare you interested00:43
rburtonnot right now as its nearly 2am ;)00:43
rburtoni've a fork of sysprof so throw the patch to me and i can file a merge request for it though :)00:44
khemrburton: patch should be in your inbox00:56
*** rubdos <rubdos!> has joined #yocto00:57
*** rburton <rburton!> has quit IRC01:02
*** dc13ff <dc13ff!uid190567@gateway/web/> has joined #yocto01:17
*** linuxjacques <linuxjacques!~jacques@nslu2-linux/jacques> has quit IRC01:47
* armpit how does khem know what is in rburton's inbox01:53
*** grahamgelding <grahamgelding!~grahamg@> has joined #yocto01:58
yoctiNew news from stackoverflow: Command to extract Linux source code in Yocto <>02:00
*** tgraydon <tgraydon!~textual@> has quit IRC02:16
khemarmpit: ehem shhh02:22
*** arielmr <arielmr!> has quit IRC02:29
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto02:49
kergothrettichschnidi: if you want devtool to update or alter a bbappend, use finish or update-recipe with appropriate arguments02:56
yoctiNew news from stackoverflow: How to modify already existing linux driver in Yocto <>03:00
*** dc13ff <dc13ff!uid190567@gateway/web/> has quit IRC03:25
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC03:46
*** mario-goulart <mario-goulart!> has quit IRC03:48
*** peniwize <peniwize!~peniwize@> has joined #yocto04:50
peniwizeHas any work been done in Yocto to try to ensure that binaries built in a previous build exactly match binaries built in a subsequent build?  This is with the exact same recipes and configuration and all intermediate and cached content deleted between builds to ensure that packages are rebuilt from scratch?  I've found that ctime must be used to update sym links, etc, in the file system to ensure consistency.  Further investigation shows differences in bi04:56
peniwizenaries, e.g. /bin/busybox.nosuid, so I assume their coming from __TIME__, __DATE__, etc, during compilation.  Does anyone know if there's a way to force those values in gcc/g++ or a way for force system time queries to always return the same value during the Yocto/BitBake build process?04:56
*** comptroller <comptroller!> has quit IRC04:59
peniwizePerhaps it's too late to ask questions.  I'll ask again tomorrow.04:59
*** comptroller <comptroller!> has joined #yocto05:15
*** gtristan <gtristan!~tristanva@> has joined #yocto05:22
*** demonimin <demonimin!~demonimin@unaffiliated/demonimin> has quit IRC05:23
*** morphis <morphis!> has joined #yocto05:24
*** georgem_home <georgem_home!uid210681@gateway/web/> has quit IRC05:37
*** mario-goulart <mario-goulart!> has joined #yocto05:42
*** pohly <pohly!> has joined #yocto06:04
*** lusus <lusus!~lusus@> has joined #yocto06:51
*** Kitsok <Kitsok!~kitsok@2a02:6b8:0:506:d681:d7ff:fe6d:d499> has joined #yocto06:52
*** fl0v0 <fl0v0!> has joined #yocto06:52
KitsokHello all! Guys, I need help. I'm getting "image too large" for the fit image (kernel + dtb + initramfs). Quesion is - how to extract initramfs image to find out what's there?06:53
*** rob_w <rob_w!~bob@unaffiliated/rob-w/x-1112029> has joined #yocto06:53
*** T_UNIX <T_UNIX!uid218288@gateway/web/> has joined #yocto06:58
*** karlyeurl <karlyeurl!~Karlssel@> has joined #yocto07:11
*** grahamgelding <grahamgelding!~grahamg@> has quit IRC07:18
*** nayfe <nayfe!uid259604@gateway/web/> has joined #yocto07:23
nayfeMorning all,  i'm trying to get a /data separate partition with some folders in it with WIC, do I need to create a different image recipe that creates /data/x/y/z folders and add it with rootfs-dir or can I use wic to get the current rootfs /data subdirectory to be copied in that partition?07:28
*** joaocfernandes <joaocfernandes!~Joao@> has joined #yocto07:40
*** varjag <varjag!> has joined #yocto07:42
*** BCMM <BCMM!~BCMM@unaffiliated/bcmm> has joined #yocto07:59
*** morphis <morphis!> has quit IRC08:20
*** morphis <morphis!> has joined #yocto08:21
lucaceresoliI'm upgrading my setup from rocko to sumo. Smooth with minimal trivial changes, but almost at the end I got this error:08:30
lucaceresolipackage packagegroup-my-apps-1.0-r0.my_machine  requires my-app, but none of the providers can be installed08:31
lucaceresoliSolution 1:08:31
lucaceresoli- do not ask to install a package providing packagegroup-my-apps08:31
lucaceresoliI fail to understand. 'bitbake my-app' and 'bitbake packagegroup-my-apps' do build successfully and I have no changes to all of those recipes08:32
lucaceresolioh, the above error happens when doing 'bitbake my-image'08:33 does IMAGE_INSTALL += "packagegroup-my-apps ..."08:33
lucaceresoliand does RDEPENDS_${PN} = "my-app"08:34
*** Bunio_FH <Bunio_FH!> has joined #yocto08:34
karlyeurlHello. I'm trying to get two versions of PostgreSQL based on this recipe: in the same image. I get this error: ERROR: The recipe postgresql9 is trying to install files into a shared area when those files already exist. [snip lots of files matched in manifest-qemux86-64-postgresql.populate_sysroot] Please verify which recipe should provide the above files. [Really08:34
karlyeurllong chunk of text, with this relevant part: It could also be your build is including two different conflicting versions of things (e.g. bluez 4 and bluez 5) and the correct solution for that would be to resolve the conflict.08:34
karlyeurlI have no clue how to do that.08:34
karlyeurl(The recipe is long, looks full of dirty hacks and I have little understanding of what it's actually doing)08:35
*** Bunio_FH <Bunio_FH!> has quit IRC08:38
*** Bunio_FH <Bunio_FH!> has joined #yocto08:40
*** Kitsok <Kitsok!~kitsok@2a02:6b8:0:506:d681:d7ff:fe6d:d499> has quit IRC08:47
*** jkliemann <jkliemann!~jk@2a03:4000:1c:73::ac17:1000> has joined #yocto09:00
jkliemannhi, I created a custom recipe that fetches oracle java unpacks it and puts the directories into the rootfs, however it fails with a missing dependency to I have no idea where this comes from, I didn't state any dependencies in my recipe at all09:01
*** awe00 <awe00!~awe00@unaffiliated/awe00> has joined #yocto09:21
nayfeIt scans so libs to find dépendancies automatically09:24
jkliemannthanks, i already found out, as it seems there is no good way to disable this? I read some discussions about this yet I still just want to put the files in place, without any dependency checking, is there a way to do this?09:32
nayfeYou can use readelf to patch dépendancies09:39
jkliemannyeah I read that, too. I hoped i wont need to outsmart the build system, but it seems I have to...09:42
*** kanavin <kanavin!~kanavin@2a02:8109:b5c0:1a5d:a1cc:a2ec:7b6b:68e1> has joined #yocto09:42
nayfeYou can maybe also use Java profiles?09:45
*** rburton <rburton!> has joined #yocto09:45
nayfei'm trying to get a /data separate partition with some folders in it with WIC, do I need to create a different image recipe that creates /data/x/y/z folders and add it with rootfs-dir or can I use wic to get the current rootfs /data subdirectory to be copied in that partition?09:47
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto10:10
xtronif a recipe A RDEPEND on recipe B, does it has the same effect as IMAGE_INSTALL += "B"10:19
*** gtristan <gtristan!~tristanva@> has quit IRC10:24
*** bluelightning_ <bluelightning_!~paul@pdpc/supporter/professional/bluelightning> has joined #yocto10:25
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC10:25
*** bluelightning <bluelightning!~paul@pdpc/supporter/professional/bluelightning> has quit IRC10:28
*** bluelightning_ is now known as bluelightning10:28
*** diego_r <diego_r!> has joined #yocto10:31
rburtonxtron: xtron if A RDEPENDS on B then installing A will  install B10:38
*** jofr <jofr!~jofr@> has left #yocto10:39
*** jofr <jofr!~jofr@> has joined #yocto10:40
*** jofr <jofr!~jofr@> has left #yocto10:40
*** jofr <jofr!~jofr@> has joined #yocto10:41
*** tgoodwin <tgoodwin!> has joined #yocto10:47
xtronrburton: right10:53
rburtonRP: so, the machines running oe-buildperf need to move.  currently four old desktops and a xeon running a range of host distros.  it looks like there's been little value in having OS coverage: do you object to consolidating the machines down to a single xeon that can run the same tests but a lot faster?10:55
rburtonfaster runs mean more data points, which is good10:55
nayferburton: hi, if you have time, do you have any idea on how to split a rootfs in several partitions? Or do I need a single image recipe for each rootfs partitions?11:01
rburtonthe latter11:02
yoctiNew news from stackoverflow: Yocto Xfce4 Session-Manager not starting <> || Overwriting Yocto Classes through meta-layer <>11:02
nayfeThx for quick answer, do you think It can be useful to add this kind of feature in wic?11:03
RPrburton: original idea was to run on the kind of hardware end users have11:04
RPrburton: but no objection really, its going to need to change11:04
rburtonRP: end users have xeons right11:04
RPrburton: some :)11:04
rburtonRP: currently markus is looking at them when he gets a moment, so one was off for a month and they're all using old distros now11:05
RPrburton: right, its not maintainable :(11:05
rburtonRP: there's a 'spare' xeon in JF we can contribute though that tracy never got around to using11:05
rburtoni guess then we can ask the AB if anyone else wants to donate.  an arm64 box would be interesting for sure11:06
RPrburton: right, we should have some in LF racks now too11:06
rburtonyeah, if we can migrate to LF hardware eventually that's great.11:06
rburtoni just want to stop markus running an illicit cluster right now ;)11:07
RPrburton: don't we have two machines at LF doing this already?11:07
rburtonRP: huh, didn't think so11:07
RPrburton: where does come from? Markus or LF?11:08
RPrburton: ok, I thought at least some was LF machines :/11:11
RPrburton: we need to talk to halstead11:11
rburtonwell, markus has five he can see11:11
* rburton looks at the machines on the list11:11
rburtonyeah that looks like his list11:12
rburtoni shall ping halstead11:12
kanavinrburton: cheers for fixing sysprof11:12
RPrburton: yes, looks like they probably are the ones markus has11:12
rburtonhe has four desktops (f25, ubuntu, suse, centos) and a xeon11:13
rburtonwhich matches the posts11:13
RPrburton: yes, agreed. I know there were some meant to be setup by michael, lets talk to him11:14
RPgood news is I figured out the populate_sdk warning about linux-libc-headers11:15
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto11:16
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has quit IRC11:29
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has joined #yocto11:45
*** kanavin <kanavin!~kanavin@2a02:8109:b5c0:1a5d:a1cc:a2ec:7b6b:68e1> has quit IRC11:50
tgoodwinIs there a method in yocto similar to C to guard against repeated inclusions?11:57
tgoodwinOr probably more accurately, is there a method in bitbake for such a thing?  I'm trying to get around having all these warnings about multiple inclusions but the only "real" way is to define the variables globally.11:57
mcfrisktgoodwin: move things from .inc to .bbclasses12:05
tgoodwinmcfrisk: I'll look into if that clears things up.  As I chip at this, I think I need to re-think how to get this going.  The context of the issue is like a "work-shared" directory for the kernel (but poorly done, since I'm ending up with these multi-inclusions).  The multi-inclusions aren't breaking anything; it's just that nuisance warning.12:17
tgoodwinI've tried "include ${@bb.utils...}" to conditionally include the file based on whether or not the variables within it are defined.  Naturally that doesn't work as a guard.12:19
tgoodwinWhere this is all coming from is concurrently building an initrd image and the runtime image (via a wic plugin).  The wic imager will output an updated fstab and other files that the intird image needs, so it's a coordination of the tasks and storage locations.12:23
*** BCMM <BCMM!~BCMM@unaffiliated/bcmm> has quit IRC12:32
tgoodwinmcfrisk: thanks, that was ultimately a cleaner way to go.12:47
*** jofr <jofr!~jofr@> has left #yocto12:49
*** jofr <jofr!~jofr@> has joined #yocto12:53
*** jofr <jofr!~jofr@> has left #yocto12:53
*** BCMM <BCMM!~BCMM@unaffiliated/bcmm> has joined #yocto13:01
*** jofr <jofr!~jofr@> has joined #yocto13:02
* RP doesn't understand why fails intermittently, possible distro specific :/13:03
*** kanavin <kanavin!~kanavin@2a02:8109:b5c0:1a5d:a1cc:a2ec:7b6b:68e1> has joined #yocto13:04
zeddiidoes anyone know if there's a way to point the layer index at something besides the master branch for a layer ?13:06
RPzeddii: bluelightning but he's likely asleep13:07
bluelightningoddly I'm not13:08
bluelightningzeddii: what do you need?13:08
zeddiiI have a layer I registered a while ago, and was noting that my recipes weren't being listed13:08
zeddiiand then I realized, the development isn't on master, so that's why.13:08
zeddiithis is a crime of my own creation.13:08
bluelightningok, which layer is this?13:09
zeddiiso if it isn't possible, that's cool. I'll just eventually have to stop being a putz and do some sort of branch renaming.13:09
bluelightningwhat should it be?13:09
zeddiithere's some open container recipes in that branch that I'd like others to see if they search, since that would be a trigger to move them to meta-virt, etc, or figure out a path.13:09
zeddiiright now, they are hidden.13:10
RPzeddii: any idea on the timing of the warnings fixes? Just trying to get a feel for when builds should happen13:10
zeddiiRP: I'll send you what I have in the next 30 mins.13:10
zeddiiit should get most, if not all of them.13:10
RPzeddii: cool, thanks. I have a raft of other fixes so together... :)13:11
RPfixing universe wasn't as bad as I was expecting13:11
zeddiiheh. we take what we can get :D13:11
*** rob_w <rob_w!~bob@unaffiliated/rob-w/x-1112029> has quit IRC13:12
bluelightningzeddii: ok, fixed - recipes should show up with the next update13:12
bluelightning(within 3 hours)13:12
zeddiiawesome. many thanks.13:13
bluelightning(FYI, how it works internally is there is an "actual branch" field for situations like this, so I just set the master layerbranch for meta-cube to have master-oci as its actual branch)13:14
bluelightningit's only settable from the admin interface though13:14
zeddiiahah. I was poking around in the interface, but didn't see where I could change it myself.13:16
zeddiimakes sense that it isn't something people can just bounce around.13:16
zeddiiI admit that it isn't typical to not want master. I shall eventually fix that, but it will be a PITA so I'm avoiding it.13:17
*** JPEW <JPEW!cc4da337@gateway/web/freenode/ip.> has quit IRC13:25
RPrburton, kanavin: /var/tmp/rpm-tmp.Ya4Mf0: line 14: gconftool-2: command not found13:27
RPrburton, kanavin: failure from the api documentation build13:27
RPwonder if a dependency is being trampled somewhere13:28
RPI think build success/failure depends if its installed on the host :/13:28
yoctiNew news from stackoverflow: automake native:oe_runmake failed in yocto krogoth <>13:32
rburtonRP: what recipe?13:35
*** marka <marka!~masselst@> has joined #yocto13:35
RPrburton: ERROR: core-image-sato-1.0-r0 do_populate_sdk: Postinstall scriptlets of ['settings-daemon', 'matchbox-session-sato'] have failed. If the intention is to defer them to first boot,then please place them into pkg_postinst_ontarget_${PN} ().Deferring to first boot via 'exit 1' is no longer supported.13:35
RPrburton: both the same gconftool-2 missing error. DOesn't reproduce locally though13:36
rburtonpresumably you have gconftool locally13:36
RPrburton: does reproduce on centos7. I don't have gconftool-2 locally :/13:36
*** JPEW <JPEW!cc4da337@gateway/web/freenode/ip.> has joined #yocto13:37
RPrburton: theory - may happen when building from sstate. testing that13:38
JPEWRP: I think I *finally* tracked down those bitbake server timeout errors with my persist_data changes. It looks like using WAL mode with sqlite makes it more sensitive to I/O delay, which means on a system with load it will more easily surpass the 30 second timeout13:39
RPJPEW: that would fit with my experiences with sqlite13:40
kanavinRP: all mentions of gconftool in oe-core are accompanied by PACKAGE_WRITE_DEPS += "gconf-native", so it should be working?13:40
RPkanavin: sure it should. But it isn't ;-)13:41
*** bluelightning <bluelightning!~paul@pdpc/supporter/professional/bluelightning> has quit IRC13:41
RPkanavin: I'm trying to figure out the pattern13:41
JPEWI'm not sure what to do about that. I suppose we could drop the WAL mode patch and see if any issues crop up.... I was using it because I was seeing database corruption issues, but maybe they are caused by something else13:42
RPJPEW: I have to wonder if there is a better way to do what we're doing than sqlite :/13:42
RPor perhaps one sqlite connection and multiple xmlrpc calls to the server?13:43
kergothsqlite seems to really suck with lots of parallel queries. ran into a lot of issues with that when first doing the parallel parsing stuff13:44
JPEWRP: Ya, I did some research there as well. We are effectively using sqlite as an IPC mechanism (shared data cache), which at first glance sounds odd...13:44
JPEWkergoth: Ya, that was part of why WAL mode can be nice... writers don't block readers anymore.13:45
* kergoth nods13:45
RPJPEW: I've used sqlite in a few situations and ended up having to use something else in nearly all of them :(13:46
RPJPEW: Originally I guess it was a seemingly cheap way to sort some IPC issues13:46
JPEWRP: What did you switch to? When I was doing research, I didn't really didn't see many other turn-key solutions (for Python at least).13:47
RPkanavin: log says Installed into sysroot: [..., 'gconf-native' but it clearly didn't13:48
JPEWReally, I think we just need some Python dicttionany in SHM (or a RPC server that does the same)13:48
RPJPEW: did something else entirely in one case, mysql in another13:48
RPJPEW: using xmlrpc to do the rpc and sqlite as the database wth a single thing accessing it may be easier13:49
JPEWRP: Ya, that might be the best option. I still want to look into why a "DELETE FROM table;" when the table is empty is taking 30 seconds for a bit before I start on that.13:52
RPJPEW: that does sound bad13:53
RPkanavin: WARNING: Manifest /home/pokybuild/yocto-worker/nightly-qa-extras/build/build/tmp/sstate-control/manifest-x86_64-gconf-native.populate_sysroot not found in x86_64 (variant '')?13:53
RPthat sounds bad13:53
*** dmoseley <dmoseley!48b86a4a@gateway/web/cgi-irc/> has joined #yocto13:53
*** morphis <morphis!> has quit IRC14:01
yoctiNew news from stackoverflow: Tomcat startup time <>14:02
*** florian_kc is now known as florian14:02
*** fdanis_away is now known as fdanis14:06
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC14:20
*** jkliemann <jkliemann!~jk@2a03:4000:1c:73::ac17:1000> has quit IRC14:21
*** fl0v0 <fl0v0!> has quit IRC14:23
*** fdanis is now known as fdanis_away14:23
*** stephano <stephano!~stephano@> has joined #yocto14:26
*** varjag <varjag!> has quit IRC14:54
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto15:01
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has quit IRC15:02
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has joined #yocto15:03
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has quit IRC15:10
*** jacques <jacques!~jacques@nslu2-linux/jacques> has joined #yocto15:20
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has joined #yocto15:23
*** ntl <ntl!> has quit IRC15:25
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has joined #yocto15:31
*** florian <florian!~florian_k@Maemo/community/contributor/florian> has quit IRC15:44
*** lusus <lusus!~lusus@> has quit IRC15:49
RPDEBUG: SState: Looked for but didn't find file /srv/autobuilder/
*** sa2ajj <sa2ajj!> has quit IRC15:56
RPbingo. Reproducer is "bitbake gconf-native -c cleansstate; bitbake core-image-sato -c populate_sdk"15:56
RPI kind of dread to look into why that may be :/15:56
*** sa2ajj <sa2ajj!> has joined #yocto15:58
*** ntl <ntl!> has joined #yocto16:00
armpitthen close your eyes16:09
*** jdel <jdel!~jdel@> has quit IRC16:19
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC16:22
*** joaocfernandes <joaocfernandes!~Joao@> has quit IRC16:37
RPits a genuine bug in runqueue...16:40
*** flihp <flihp!~flihp@> has joined #yocto16:41
*** diego_r <diego_r!> has quit IRC16:42
RPrburton: I think I have all the fixes for warnings/errors and patch tweaks into -next17:00
RPhalstead: I'm assuming maintenance will be happening so could you start -next once done please?17:00
* RP is hoping -next will be then able to merge, then we just have the allarch stuff to sort before we can build 2.617:01
halsteadRP: can we run -next now and I'll update after?17:02
RPhalstead: yes, sure17:03
RPhalstead: fired17:05
RPhalstead: I think you rebooted them fairly recently anyway17:07
halsteadYes. There is nothing super important updatewise.17:08
*** awe00 <awe00!~awe00@unaffiliated/awe00> has quit IRC17:10
*** stephano <stephano!~stephano@> has quit IRC17:11
*** zeddii <zeddii!~bruce@> has quit IRC17:12
*** arielmr <arielmr!> has joined #yocto17:17
*** T_UNIX <T_UNIX!uid218288@gateway/web/> has quit IRC17:18
*** kanavin <kanavin!~kanavin@2a02:8109:b5c0:1a5d:a1cc:a2ec:7b6b:68e1> has quit IRC17:38
jae1Hi, I have an eventhandler set to run on event.BuildStarted, And I want to change a flag between multiple runs of the function.17:41
jae1I print the flag in the event handler and set it using a default value and a task override (ex. FLAG_task-populate-sdk-ext), but Im not seeing the changed print when in the populate_sdk_ext class.17:41
jae1Any ideas? anything about how eventhandlers handle variables that I am missing?17:42
kergothtask overrides are only set when that actual task is running17:57
kergothit's *not* based on what you happened to include on the bitbake commandline17:57
jae1yes! thats why when I do a bitbake -c populate_sdk_ext, I expect the flag value to be the default, and only be changed when in populate_sdk_ext task, but it is not switching over as i expect17:59
kergothno, that's not the behavior18:00
kergothoh, i see18:00
kergothit also depends on when it's expanded. can you pastebin?18:01
jae1sorry, what does pastebin mean18:04
kergothit's a means of posting a file to share without spamming the irc channel with a big paste18:04
kergothcan copy and paste into it, then give us the link18:04
kergothi.e. let us see what you're doing, or at least part of it18:04
*** ntl <ntl!> has quit IRC18:05
jae1I see ok18:07
jae1something like this
kergothoh, i see, you're confused about what the events do and when they run18:10
kergothBuildStarted runs once per bitbake command, before any tasks are run18:10
kergothif you want to run something every time a task starts, use TaskStarted18:10
jae1ok, taking a step back, I am trying to do something almost identical to what uninative tarball class is doing, and there, they use a buildstarted and configparsed events18:12
kergothagain, buildstarted runs once, before tasks run, and the task override only applies when the tasks run, long after buildstarted18:12
jae1Ah I see ok18:13
peniwizeI've built a few of the stock images that come with Yocto and it seems that they all use busybox.  Is it possible to build an image that doesn't use busybox?18:17
kergothpeniwize: you could create a custom image that doesn't pull in any of our packagegroups to do that18:18
kergothmore common is to create an image based on our existing packagegroups, with busybox, but just install all the real tools on top of it18:18
kergothi.e. if findutils is installed, its find binaries take precedence over busybox's18:18
kergothit uses the debian/ubuntu update-alternatives mechanism18:19
peniwizeAh, I see.  I wasn't sure if busybox was critical or not.18:20
*** ntl <ntl!> has joined #yocto18:22
*** varjag <varjag!> has joined #yocto18:23
kergothshouldn't be, no, just most images rely on packagegroup-core-boot or packagegroup-base which pull it into the baseline18:24
*** geissonator <geissonator!~geissonat@> has joined #yocto18:33
*** ntl <ntl!> has quit IRC18:35
*** ntl <ntl!> has joined #yocto18:52
*** rcw <rcw!~rcw@> has joined #yocto18:53
*** jacques <jacques!~jacques@nslu2-linux/jacques> has quit IRC18:55
*** jacques <jacques!~jacques@nslu2-linux/jacques> has joined #yocto18:55
*** jacques <jacques!~jacques@nslu2-linux/jacques> has joined #yocto19:02
*** vmeson <vmeson!> has quit IRC19:14
peniwizeI’m working on a device for a highly regulated industry that requires that the software binaries I deliver exactly match the software source code I deliver, when the software is re-built by a third party.  This is how the industry verifies that the source code on record was used to generate the binaries.  Is there anything in BitBake or Yocto to handle this kind of thing?19:15
peniwizeThis is often a complex problem to solve due to file system metadata, time stamps in files, etc.  I’ve looked at some differences in binaries between two Yocto builds of core-image-minimal and it appears that __TIME__ and __DATE__ are being compiled into some.  There are also other differences that I have yet to identify.19:15
peniwizeHas anyone ever dealt with this issue before (or anything similar)?  Does anyone know how to force the values of gcc’s __TIME__ and __DATE__ preprocessor macros or perhaps how to force the system to always return the same time value from ‘time_t time(time_t *)’ for the compiler process?19:15
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has quit IRC19:16
rcwI think this may be what you are looking for:
*** ntl <ntl!> has quit IRC19:20
jae1kergoth: sorry, going back to the  event.BuildStarted, if this is the case, then populate_sdk_ext class must be triggering something that calls that BuildStarted event again.19:20
jae1And for the second time that it runs, I want a Flag to be changed19:22
jae1thats why I tried using a task overrides to change the flag but doesnt seem to work19:23
*** vmeson <vmeson!> has joined #yocto19:23
*** Bunio_FH <Bunio_FH!> has quit IRC19:24
peniwizercw: Thanks.19:24
kergothmost likely it just calls bitbake again. that wouldn't result in the task override being set if that's the case19:25
*** jacques is now known as linuxjacques19:31
*** jdel <jdel!~jdel@> has joined #yocto19:40
jdeli see that package.bbclass uses objdump/strip19:41
jdelthough it doesn't appear to have a direct dependency on binutils19:41
jdeli'm working with a class that at-build-time figures out of it needs to compile or use pre-built binaries for a given recipe19:42
jdelpart of the class's requirement is that it doesn't need to do_prepare_recipe_sysroot if using prebuilt binaries19:43
jdelbut i think that packaging depends on the binutils in the recipe-sysroot dir19:43
jdelit looks like I can mostly turn off that behavior using variues INHIBIT_ variables19:43
jdeldoes that make sense?19:44
jdelis there some way to have binutils available without otherwise populating the sysroot?19:44
khemboost 1.68 broke kea hmmm19:45
*** stephano <stephano!~stephano@> has joined #yocto19:46
*** linuxjacques <linuxjacques!~jacques@nslu2-linux/jacques> has quit IRC19:48
*** jacques <jacques!~jacques@nslu2-linux/jacques> has joined #yocto20:05
*** BCMM <BCMM!~BCMM@unaffiliated/bcmm> has quit IRC20:11
*** marka <marka!~masselst@> has quit IRC20:17
*** lexano <lexano!> has quit IRC20:18
RPzeddii_home: warnings looking much better thanks :)20:23
RPmost of the remaining ones have bugzilla entries20:23
RP(non kernel)20:23
jae1kergoth: Ok then if task override wont work, Is there any other way to know that its in the second run of that event?20:26
*** WillMiles <WillMiles!> has joined #yocto20:28
*** tgoodwin <tgoodwin!> has quit IRC20:30
*** ant_home <ant_home!> has joined #yocto20:54
*** rcw <rcw!~rcw@> has quit IRC21:11
RPzeddii_home: beaglebone-yocto and qemuarm are both showing a warning on 4.1421:16
RPzeddii_home: still much improved though :)21:17
*** geissonator <geissonator!~geissonat@> has quit IRC21:17
RPzeddii_home: and mpc8315e-rdb 4.1421:18
*** WillMiles <WillMiles!> has quit IRC21:18
*** ant_home <ant_home!> has quit IRC21:19
*** pohly <pohly!> has quit IRC21:21
*** ntl <ntl!> has joined #yocto21:24
kergothRP: does that runqueue bug exist in 1.38 too, or just master? will it be cherry picked?21:25
RPkergoth: I suspect it needs backporting as that code hasn't changed in a while21:26
RPkergoth: I'll likely cherry-pick it but since it took me three goes to make things build, it needs a little settling time21:28
RPkergoth: so many different ways through that code :/21:28
RP(I tested on existing tmpdirs initially)21:28
* kergoth nods, fair enough, it's complex21:29
RPwould make an intersting fuzz test - delete sstate artefact(s), check everything still builds, repeat21:30
kergothpick a random set of artifacts to delete? That would be interesting21:30
RPkergoth: yes21:30
*** georgem_home <georgem_home!uid210681@gateway/web/> has joined #yocto21:41
*** rburton <rburton!> has quit IRC21:43
*** varjag <varjag!> has quit IRC21:45
*** awe00 <awe00!~awe00@unaffiliated/awe00> has joined #yocto21:52
*** varjag <varjag!> has joined #yocto21:56
*** Saur <Saur!> has quit IRC22:01
*** Saur <Saur!> has joined #yocto22:02
*** ntl <ntl!> has quit IRC22:14
*** varjag <varjag!> has quit IRC22:18
*** mattsm <mattsm!~mattsm@> has quit IRC22:26
*** tdonahue <tdonahue!~tdonahue@2600:8805:d500:fb00:1aba:515a:ee6f:cd17> has joined #yocto22:26
*** tdonahue <tdonahue!~tdonahue@2600:8805:d500:fb00:1aba:515a:ee6f:cd17> has quit IRC22:29
*** mattsm <mattsm!~mattsm@> has joined #yocto22:29
*** tdonahue <tdonahue!~tdonahue@2600:8805:d500:fb00:1aba:515a:ee6f:cd17> has joined #yocto22:30
*** arielmr <arielmr!> has quit IRC22:55
jdelRDEPENDS must get automatically computed for things that otherwise have DEPENDS properly set23:19
jdeldoes anyone know where what calculation happens?23:19
kergothsearch for shlibs in package.bbclass23:21
*** agust <agust!> has quit IRC23:24
*** arielmr <arielmr!> has joined #yocto23:52
*** stephano <stephano!~stephano@> has quit IRC23:59

Generated by 2.11.0 by Marius Gedminas - find it at!