**** BEGIN LOGGING AT Mon May 04 02:59:57 2020
May 04 07:16:49 <mckoan> good morning
May 04 08:40:24 <RP> kanavin_home: I ran more patches through the autobuilder. The selftests look like kernel and load issues, not sure what is causing the other testsdkext issues
May 04 09:15:44 <kanavin_home> RP: thanks, if there are specific links I could look at, let me know
May 04 09:46:15 <RP> kanavin_home: https://autobuilder.yoctoproject.org/typhoon/#/builders/83/builds/920 - the qemuarm/qemux86*/qemumips ones
May 04 10:04:12 <yocti> New news from stackoverflow: need help in using bitbake INCOMPATIBLE_LICENSE flag <https://stackoverflow.com/questions/34047903/need-help-in-using-bitbake-incompatible-license-flag>
May 04 10:40:06 <kanavin_home> RP: could those failures be caused by git version update?
May 04 10:40:20 <kanavin_home> bb.data_smart.ExpansionError: Failure expanding variable do_compile[file-checksums], expression was ${@srctree_hash_files(d)} which triggered exception CalledProcessError: Command '['git', 'add', '-A', '.']' returned non-zero exit status 1.
May 04 11:30:29 <LetoThe2nd> what would happen if one has multiple, long-running memory resident bitbakes on one host, for several build directories? is this problematic?
May 04 12:34:38 <yocti> New news from stackoverflow: Configuring and Using Plymouth in Yocto-based Linux Distribution <https://stackoverflow.com/questions/50437496/configuring-and-using-plymouth-in-yocto-based-linux-distribution>
May 04 12:55:46 <rburton> LetoThe2nd: should be fine, the sockets and so on are all in the build directory
May 04 12:56:37 <LetoThe2nd> rburton: ok, thanks
May 04 13:16:37 <FrazerClews> hi, im trying to submit patches to bitbake, but getting this error `Series does not apply on top of target branch` even thought the target branch is correct, i keep getting the error with 2 different patches. is this a known issue and can i continure to submite patches, or should i wait?
May 04 13:19:59 <kroon> FrazerClews, u sure youre using the correct mailing list ?
May 04 13:20:08 <kroon> FrazerClews, bitbake is bitbake-devel@...
May 04 13:20:34 <FrazerClews> kroon, thanks, didnt seem right at first
May 04 13:34:52 <yocti> New news from stackoverflow: Multiple rootfs through .wks file fails in yocto <https://stackoverflow.com/questions/61191659/multiple-rootfs-through-wks-file-fails-in-yocto>
May 04 13:47:27 <qschulz> rburton: I don't understand your mail wrt CC presence in a makefile
May 04 13:48:08 <rburton> qschulz: make -p /dev/null shows that the default environment for make has CC = cc
May 04 13:48:20 <rburton> so really, a makefile doesn't need to set CC
May 04 13:48:43 <LetoThe2nd> if you're writing makefiles, you're doing it wong anyways (TM)
May 04 13:48:46 <rburton> well, yeah
May 04 13:49:20 <rburton> but either use the default rules, or replicate them properly. its everyone doing replicate-but-badly that causes the problems with make :)
May 04 13:49:32 <qschulz> rburton: oh! I'm learning new things :) what's cc?
May 04 13:49:53 <rburton> the traditional unix name for the c compiler
May 04 13:50:14 <rburton> e.g. on my mac /usr/bin/cc -> clang
May 04 13:50:21 <rburton> typical linux it points at gcc
May 04 13:51:01 <LetoThe2nd> hrhr. https://img-9gag-fun.9cache.com/photo/arpnqKB_700bwp.webp
May 04 13:51:04 <qschulz> isn't that somehow risky? I mean I'd expect different flags for different compilers for example?
May 04 13:51:55 <LetoThe2nd> its like a historical convention that you can get a compiler by invoking cc.
May 04 13:53:22 <qschulz> LetoThe2nd: that i understand.. it's just, don't we want to say the default cross compiler for this piece of SW is gcc for example? to basically explicit that you should compile it with gcc or else you might wander on some unchecked paths?
May 04 13:53:42 <rburton> qschulz: in the specific example of OE, CC CFLAGS etc are all exported for you already
May 04 13:53:44 <qschulz> rburton: but thanks, very interesting, I see many things define
May 04 13:54:03 <qschulz> rburton: indeed, I was more talking manual compilation or with scripts
May 04 13:54:10 <LetoThe2nd> qschulz: why and where would we want to set set besides in OE? and there its already done.
May 04 13:54:12 <qschulz> defined*
May 04 13:54:46 <qschulz> LetoThe2nd: Not everybody is using Yocto to build a SW :) I sometimes build tools for my desktop from some github repo
May 04 13:55:16 <LetoThe2nd> qschulz: if a build makes assumptions about its environments, it should honor the environment and stop/break/yell/explode if its need are not fulfilled, instead of hardcoding overrides that give everybody a headache. my $.02
May 04 13:56:48 <LetoThe2nd> qschulz: for example, hardcoding arm-none-eabi-gcc in the Makefile. hooray.
May 04 13:56:54 <qschulz> LetoThe2nd: I was talking about weakly setting CC to gcc, not "hardcoding" it :) It's like a "suggestion". But anyway, that's more philosophical than anything :)
May 04 13:57:48 <LetoThe2nd> qschulz: people who actually know about weakly setting and such don't need the advice anyways. its one of the "if you have to ask, then don't do it" cases.
May 04 13:59:13 <qschulz> LetoThe2nd: I didn't know about `make -p /dev/null` :) I remember all my university writing those rules and variables with most already defined. I hate that I missed that :)
May 04 13:59:47 <LetoThe2nd> qschulz: makefiles were already uncool when i went to uni....
May 04 14:00:17 <rburton> kids these days eh
May 04 14:01:06 <qschulz> LetoThe2nd: I never saw anything else than Makefiles at uni... But I guess it was because my major was embedded systems? everyone knows, when it works, you don't touch it
May 04 14:01:30 <LetoThe2nd> rburton: cue https://youtu.be/7iNbnineUCI
May 04 14:01:59 <qschulz> autoconf, Cmake, never heard of it. Barely heard about Buildroot back then. Anyway :) Thanks for the knowledge sharing :)
May 04 14:02:46 <LetoThe2nd> autotools and buildroot were already around 10yrs before you went to uni. i must know because i used them in my masters thesis and i'm 10 yrs older :)
May 04 14:03:33 <qschulz> LetoThe2nd: I know. Sometimes I wonder what I've been taught at uni /me shrugs
May 04 14:03:39 <LetoThe2nd> :)
May 04 14:04:23 <LetoThe2nd> anyways, good reason to crank up some offspring.
May 04 14:41:22 <RobertBerger> https://github.com/siemens/kas
May 04 14:41:36 <RobertBerger> sorry wrong chat ;)
May 04 14:43:39 <qschulz> RobertBerger: not a bad recommendation though :)
May 04 14:44:55 <kanavin_home> RP: I confirm, the testsdkext failures are due to git upgrade
May 04 15:04:27 <RP> kanavin_home: thanks, I'd not gotten to looking at that yet!
May 04 15:05:09 <yocti> New news from stackoverflow: Problem with IDE finding wrong python version during build system prep (install) <https://stackoverflow.com/questions/61595312/problem-with-ide-finding-wrong-python-version-during-build-system-prep-install>
May 04 15:10:46 * paulg still thinks PATA hard disk when seeing "IDE"
May 04 15:11:49 <RP> paulg: there is probably a generation who don't know what PATA is now :/
May 04 15:12:32 <paulg> sad, but true.
May 04 15:12:56 * qschulz cough cough
May 04 15:13:46 <kanavin_home> question to the audience: who remembers how the predecessor to IDE was called?
May 04 15:13:55 <kanavin_home> my *very* first hard drive was that.
May 04 15:14:42 <kanavin_home> (it had separate cables for data and control)
May 04 15:15:15 <RP> kanavin_home: I knew it as MFM, I have drives like that somewhere
May 04 15:15:31 <kanavin_home> RP: yep, or ST506
May 04 15:21:04 <qschulz> the oldest I felt was when I talked with a colleague 2 years younger who didn't know 5 1/2 floppy disks existed (never seen a 3" as well but at least knew about it)
May 04 15:23:02 <RP> qschulz: you mean 3.5" right :)
May 04 15:23:32 <RP> and 5.25" :)
May 04 15:23:54 <qschulz> RP: but... but... I wanted to feel old too :((
May 04 15:24:28 <RP> qschulz: it happens all too quickly, I'd enjoy the other side while you can!
May 04 15:28:18 <sstiller> Can I run psplash with thud and systemd? The systemd service is masked in psplash_git.bb
May 04 15:28:45 <RP> sstiller: dunfell was the first release with psplash systemd support so you'd have to backport that
May 04 15:32:07 <sstiller> RP: Thanks!
May 04 15:46:54 <RP> rburton, Saur: Since you're the listed maintainers for meta-gplv2, could you figure out how we should fix the glib dependency issue we have in master-next please? We have a few options such as disabling ptests, limiting its dependencies etc., just not sure which is best
May 04 15:53:37 <kanavin_home> the two options proposed by me were: drop ptests from DISTRO_FEATURES, or turn license restrictions into per-image ones (so they kick in only in do_rootfs)
May 04 15:53:52 <kanavin_home> my preference is for the latter
May 04 15:59:48 <RP> kanavin_home: that is a huge change in behaviour for the layer and tests
May 04 16:02:49 <kanavin_home> RP: yes, but if new glib requires a gpl3 item to be able to run tests, I'm not sure what else we can do. Patch it out from glib? No thanks :)
May 04 16:04:45 <rburton> meta-gpl has a bbappend to turn off ptest?
May 04 16:06:27 <kanavin_home> as far as I know, no.
May 04 16:06:58 <kanavin_home> but those ptests aren't exercised in any way, and probably won't work due to older versions of various items.
May 04 16:07:00 <rburton> i mean, how about
May 04 16:07:12 <rburton> what tests in particular?
May 04 16:07:55 <kanavin_home> rburton: what I mean is that the AB has a specific configuration to test the layer, and ptests aren't in it in any form
May 04 16:08:25 <rburton> i admit i haven't looked at the failure so have no idea what the problem is
May 04 16:08:27 <kanavin_home> so adding DISTRO_FEATURES_remove = "ptest" to the layer config is an option, just not my preferred one
May 04 16:08:52 <rburton> thats a very heavy hammer
May 04 16:08:56 <kanavin_home> rburton: new glib tests require a gpl3 item (python3-dbusmock)
May 04 16:09:08 <rburton> i suspected you'd say that
May 04 16:09:08 <kanavin_home> so it's in RDEPENDS-ptest, which breaks meta-gplv2
May 04 16:09:11 <rburton> require is a strong word
May 04 16:09:14 <rburton> one sec
May 04 16:09:55 <rburton> https://gitlab.gnome.org/GNOME/glib/-/commit/2709d5e1bb6d0cef692ba37a9dd3d250e1cf8c51
May 04 16:09:58 <rburton> would that help?
May 04 16:12:14 <kanavin_home> sure, but I'd want to keep the test enabled when meta-gplv2 is not
May 04 16:12:45 <rburton> that patch allows another option: don't recommend on dbusmock
May 04 16:13:07 <kanavin_home> don't recommend?
May 04 16:13:13 <rburton> let the tests skip
May 04 16:13:31 <rburton> the portal one won't run anyway
May 04 16:14:32 <kanavin_home> oh, would RRECOMMENDS avoid the issue? (vs RDEPENDS)
May 04 16:15:05 <kanavin_home> with meta-gplv2 presumably would skip the recipe, and not  install it when making an image
May 04 16:17:53 <rburton> not sure of the semantics for recommends and licenses
May 04 16:18:03 <rburton> i'd argue that recommends is a good idea anyway for stuff like that
May 04 16:18:09 <kanavin_home> running a quick checl
May 04 16:18:11 <kanavin_home> check
May 04 16:33:49 <kanavin_home> RRECOMMENDS did not work, but RSUGGESTS seems to do the trick
May 04 16:34:08 <kanavin_home> RP: I guess we can change the offending dependency to that?
May 04 16:41:40 <nacknick> Hi. I'm trying to compile shared object in arm. `ARM_INSTRUCTION_SET = "arm"` has not affect for some reason. Any idea why?
May 04 16:48:57 <nacknick> I added it in *local.conf*
May 04 17:04:01 <RP> kanavin_home: RSUGGESTS won't build it and may not even pull it in :(
May 04 17:05:55 <nacknick> Someone please? It's urgent
May 04 17:11:14 <nacknick> I'm trying to compile shared object in arm. `ARM_INSTRUCTION_SET = "arm"` has not affect for some reason. Any idea why?
May 04 17:11:30 <nacknick> In *local.conf*
May 04 17:11:59 <JPEW> nacknick: Make sure it's taking effect: bitbake -e YOURRECIPE
May 04 17:13:14 <nacknick> JPEW it's affecting the bina
May 04 17:13:56 <nacknick> Affecting the executable, but not the shared objects
May 04 17:14:08 <nacknick> Executables
May 04 17:14:10 <kanavin_home> rburton: so what do we do? just drop the dependency and backport a patch that skips the tests? I'm not comfortable with that, it's side-stepping the issue.
May 04 17:14:26 <kanavin_home> meaning, it's a coincidence that we can do that
May 04 17:14:26 <rburton> its easy though, and isn't horrible
May 04 17:14:39 <kanavin_home> what if it was a hard dependency?
May 04 17:14:55 <rburton> then meta-gpl can have a glib.bbappend that turns off ptest forcibly
May 04 17:15:08 <rburton> for that recipe
May 04 17:15:14 <kanavin_home> maybe that's the best option then?
May 04 17:15:22 <RP> kanavin_home: I'm thinking meta-gplv2 just has to force ptest off
May 04 17:15:28 <JPEW> nacknick: Is the so in the same recipe as the executable?
May 04 17:15:48 <kanavin_home> RP: so you're ok with DISTRO_FEATURES_remove = "ptest" in that layer then?
May 04 17:16:01 <kanavin_home> or just for glib?
May 04 17:16:02 <rburton> RP: i don't like the layer turning off ptest for the entire distro
May 04 17:16:03 <RP> kanavin_home: for that recipe
May 04 17:16:12 <kanavin_home> right
May 04 17:16:25 <nacknick> JPEW how should I find .so recipes?
May 04 17:16:46 <kanavin_home> rburton: could you make a patch then? :)
May 04 17:17:29 <rburton> kanavin_home: give me a few days, haven't done a build for ~weeks
May 04 17:17:45 <kanavin_home> right, maybe I can cook up smth quicker
May 04 17:21:43 <RP> kanavin_home: the challenge is keeping that layer YP compatible too :/
May 04 17:24:20 <RP> kanavin_home: wondering if we should create a .inc inside meta-gplv2 which users would use to enable the layer. The autobuilder could then just include that rather than its current settings?
May 04 17:24:34 <RP> kanavin_home: more complex as an approach but easier to fix other things in future?
May 04 17:24:45 <kanavin_home> RP: I placed PTEST_ENABLED_pn-glib-2.0 = "0" into layer.conf, which seems to work
May 04 17:27:41 <RP> kanavin_home: that breaks YP Compat as inclusion of the layer changes behaviour :/
May 04 17:31:03 <kanavin_home> RP: will that be reported by yocto-check-layer?
May 04 17:31:30 <RP> kanavin_home: yes
May 04 17:32:03 <kanavin_home> bah, I should really get my employer to join the project, and then vote the layer out :)
May 04 17:32:43 <RP> :D
May 04 17:33:22 <RP> kanavin_home: I can sort the include thing later on
May 04 17:35:02 <kanavin_home> RP: somehow I am getting test_signature failures even when testing master meta-gplv2 with that script
May 04 17:36:09 <kanavin_home> RP: so what to do then?
May 04 17:45:33 <RP> kanavin_home: something must be wrong with the setup as it passes on the AB
May 04 17:46:14 <kanavin_home> RP: maybe, I have a ton of settings in local.conf
May 04 17:49:03 <RP> zeddii: FWIW its the later two kernel updates that cause the reproducibile failure as current master-next passes
May 04 17:49:45 <RP> zeddii: so 5.4.32 onwards
May 04 17:57:02 <zeddii> cool. that helps. ran out of space on my builder, so had to do a rm on tmp.
May 04 17:57:09 <zeddii> just starting builds locally now.
May 04 18:50:53 <sailanj> \O/
May 04 18:51:05 <sailanj> =#
May 04 18:51:39 <sailanj> Hi anybody her???
May 04 18:51:56 <LetoThe2nd> no.
May 04 18:51:58 <LetoThe2nd> never.
May 04 18:52:13 <sailanj> Yes, I see that
May 04 18:52:25 <LetoThe2nd> we just play virtual hide and seek. pretty neat, don't you agree? :P
May 04 18:53:01 <sailanj> Nice, let me play too please
May 04 18:53:33 <sailanj> I agree
May 04 18:54:04 <LetoThe2nd> all is well, then.
May 04 18:54:50 <sailanj> Please Leto while you are hear let me ask you a question
May 04 18:55:40 <LetoThe2nd> unless you have a personal question, just ask the channel. if somebody knows, you will get an answer.
May 04 18:57:50 <derRichard> hey! i ran into the problem this patch tries to fix: https://www.yoctoproject.org/pipermail/yocto/2018-January/039584.html
May 04 18:57:56 <derRichard> (on zeus)
May 04 18:58:09 <derRichard> looks like the patch did not make it upstream :(
May 04 18:58:15 <sailanj> Thanks sure not personal :) . Could you let me where can i find good documentation or training for how to customized embedded Linux for a certain target board?
May 04 18:58:33 <sailanj> of course using Yocto
May 04 18:58:36 <LetoThe2nd> sailanj: https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj
May 04 19:00:39 <LetoThe2nd> sailanj: other than that, it depends quite a bit on your knowledge and if you have coin to spend, basically.
May 04 19:02:48 <sailanj> Really! thank you Leto looks like the videos titles are so interested and close to what am looking for. I can spend if it's good.
May 04 19:03:23 <LetoThe2nd> enjoy, then :)
May 04 19:04:59 <lpapp> hi, upgrading init-ifupdown via opkg means rerunning init/networking, yeah?
May 04 19:08:08 <paulg> kanavin_home, you discussion earlier about MFM/RLL reminded me of writing this gem...
May 04 19:08:12 <paulg> http://ibiblio.org/pub/linux/kernel/patches/scsi/adaptec-40XX.lsm
May 04 19:08:34 <erbo> lpapp: I would expect that yes, but safest is probably to extract the ipk and look at the pre/post install scripts
May 04 19:09:19 <kanavin_home> paulg: you knew linux before it was famous :)
May 04 19:09:29 <lpapp> thanks, so I can see it working from ssh running opkg manually
May 04 19:09:49 <lpapp> but when I run it through my C program with fork and execvp, it seems that the interface stays down
May 04 19:09:58 <lpapp> it does not come up until I manually bring it up by starting
May 04 19:10:04 <lpapp> how is this possible?
May 04 19:10:23 <lpapp> somehow, the environmeny different from the C program as opposed to the manual ssh run?
May 04 19:10:57 <LetoThe2nd> paulg: its the era we roughly call "lathe based data processing" here :)
May 04 19:31:06 <derRichard> hmm, when i want a FILES_${PN} to trigger only for a specific machine, isn't FILES_${PN}_mymachine += ... the way to go?
May 04 20:03:45 <Moh3N> HI , when I want to build my image by yocto, I got an error that said unable to fetch URL....
May 04 20:05:21 <Moh3N> the URL is for gstreamer, how I can disable it ?? I  wont to have error! I dont need gstreamer
May 04 20:06:08 <yocti> New news from stackoverflow: Yocto poky build fails on function do_rootfs <https://stackoverflow.com/questions/61600231/yocto-poky-build-fails-on-function-do-rootfs>
May 04 20:10:32 <Moh3N> my error is : fetcher failure:...<URL>
May 04 20:14:36 <Moh3N> in which recipe I can disable downloading gstreamer package ?? (my Error URL)
May 04 20:16:46 <alejandrohs> Moh3N: you need to run bitbake -g <mage> to check what is pulling gstreamer as a dependency
May 04 20:17:07 <alejandrohs> Moh3N: but its likely that its easier to fix the URL where gstreamer is coming from
May 04 20:17:28 <Moh3N> the url is correct, thanks :)
May 04 20:23:33 <lpapp> erbo: it looks like the postinst should start it
May 04 20:24:31 <lpapp> erbo: or at least, it has a line like this, update-rc.d $OPT networking start 01 2 3 4 5 . stop 80 0 6 1 .
May 04 20:40:37 <Moh3N> alejandrohs excuse me, I have another question, I have an script that I want to runs at linux startup, How I can do this in yocto for my Image ??
May 04 20:49:59 <alejandrohs> Moh3N: there are many ways to achieve that depending on what you want specifically, which init system you are using, etc, but you can look at the update-rc class as a starting point
May 04 20:50:50 <alejandrohs> Moh3N: it seems that rng-tools has a simple working example, it might help to look at that one too
May 04 20:51:03 <Moh3N> thanks a lot
May 04 21:02:15 <paulg> kanavin_home, the original linux hard disk driver "hd.c" lived out its final years as a ST-506 MFM/RLL driver - which amusingly enough got a brief 2nd wind with CF-IDE crap
May 04 21:02:35 <paulg> line 17,18 of  https://github.com/spotify/linux/blob/master/drivers/block/hd.c
May 04 21:03:17 <paulg> back when IDE drives were "new and shiny".
May 04 21:04:48 <paulg> and here is another retro computing flashback -- the PC-XT linux disk driver...
May 04 21:04:52 <paulg> https://lore.kernel.org/patchwork/patch/349710/
May 04 21:05:58 <paulg> Good times.
May 05 02:22:54 <OutBackDingo> curious, what would be the deficiencies in using Yocto on X86_64 as base OS for a server farm if any ?
May 05 02:23:11 <OutBackDingo> as opposed to say CentOS or Ubuntu?
May 05 02:55:53 <paulg> OutBackDingo, I think you'd need to refine your requirements and use case and what level of work you can do in-house vs. contracting out before anyone can give a meaningful answer to that.
**** ENDING LOGGING AT Tue May 05 02:59:57 2020