**** BEGIN LOGGING AT Wed Jun 10 02:59:57 2020
Jun 10 08:40:10 <clementp[m]> Hi all, do you know a nice platform for getting logs of IoT devices available in yocto ?
Jun 10 08:40:44 <Letothe2nd> define logs, define nice, define iot :P
Jun 10 08:42:20 <mckoan> Letothe2nd: LOL
Jun 10 08:42:56 <Letothe2nd> hey at least i think we can agree on....erm wait.
Jun 10 08:43:04 <Letothe2nd> clementp[m]: please define "platform".
Jun 10 08:43:22 <Letothe2nd> i think the "yocto" part at least is clear.
Jun 10 08:45:13 <mckoan> Letothe2nd: more and more frequently nowadays I think I'd need the help of a 'crystal ball'(TM)
Jun 10 08:45:48 <mckoan> to understand requests for quotation and tech specs :-D
Jun 10 08:46:11 <clementp[m]> Letothe2nd:  okay i have 50devices on the field and want to check if everything is fine or if they reboot what happened.
Jun 10 08:46:33 <clementp[m]> They have internet connection and I don't save anything in the flash right now
Jun 10 08:46:41 <Letothe2nd> clementp[m]: so we are talking about the system logs of the underlying linux, right?
Jun 10 08:46:45 <clementp[m]> Yes
Jun 10 08:47:25 <Letothe2nd> sysdig, maybe
Jun 10 08:47:35 <Letothe2nd> should be in meta-oe
Jun 10 08:54:24 <clementp[m]> thanks i will check
Jun 10 08:55:04 <Dracos-Carazza> maybe zabbix?
Jun 10 08:55:51 <Dracos-Carazza> eventhough this might be overkill :D
Jun 10 09:06:56 <Letothe2nd> as usual, it "depends".
Jun 10 09:07:09 <Letothe2nd> one could also just pipe rsyslog out, AFAIK
Jun 10 09:08:10 <Dracos-Carazza> yeah thats true... I used zabbix during my time at the institute in order to monitor our development VMs
Jun 10 09:08:39 <Letothe2nd> nothing wrong with that.
Jun 10 09:10:11 <Dracos-Carazza> well, "it depends" is really the correct term to describe such a solution :D
Jun 10 09:11:38 <mckoan> Dracos-Carazza: looks interesting, thank you for sharing
Jun 10 09:19:01 <kanavin_home> RP: what's the current schedule for running AUH? I couldn't figure this out from the autobuilder.
Jun 10 09:19:15 <RP> kanavin_home: its manually triggered
Jun 10 09:19:36 <RP> kanavin_home: We can schedule it if we wanted
Jun 10 09:20:01 <kanavin_home> RP: right, I guess  it's not too much trouble to press the button once a month
Jun 10 09:20:50 <RP> kanavin_home: we can make it work whichever way we want...
Jun 10 09:21:18 <kanavin_home> RP: maybe once a month on the 15th would be good, same as before
Jun 10 09:25:26 <RP> kanavin_home: http://git.yoctoproject.org/cgit.cgi/yocto-autobuilder2/commit/?id=6375d089c23ed152c9f647a384c34f9ae875c9dc - not live yet but will get added
Jun 10 09:40:57 <chris_ber> I try to create a cmake shared library and got the message: "ERROR: libprom-1.0-r0 do_package_qa: QA Issue: -dev package contains non-symlink .so: libprom-dev path '........./usr/lib/libprom.so' [dev-elf]"
Jun 10 09:41:01 <chris_ber> any idea?
Jun 10 09:43:08 <Letothe2nd> chris_ber: sounds like versioning is not correct
Jun 10 09:43:44 <Letothe2nd> chris_ber: i think you need to specify the library version so all the naming and symlinking works properly, then it gets packaged as intended.
Jun 10 09:45:13 <qschulz> chris_ber: libprom.so.1.0 is what your library should be named. Then you can have a symlink libprom.so.1 and libprom.so, which will make it to ${PN}-dev package
Jun 10 09:48:17 <chris_ber> ok
Jun 10 09:51:43 <Letothe2nd> the key point is that the naming works automatically once you properly use cmakes library versioning
Jun 10 09:55:56 <chris_ber> you talking about the missing version info in the CMakeLists.txt?
Jun 10 09:57:50 <Letothe2nd> chris_ber: i haven't looked at your cmakelists. jsut saying that if you do things correctly, then they get packaged correctly.
Jun 10 10:01:21 <chris_ber> i added in CMakeLists.txt: project(prom VERSION 1.0.0) and got still the saem Problem -> https://bitbin.it/Nj015pMb/
Jun 10 10:08:17 <Letothe2nd> chris_ber: can't find it right now, but i guess you need to make sure the SOVERSION and all is set
Jun 10 10:08:57 <chris_ber> ok
Jun 10 10:11:09 <rburton> who is arranging the devday at ELC-NA?
Jun 10 10:11:13 <rburton> i bet Letothe2nd  knows
Jun 10 10:11:29 <Letothe2nd> rburton: i do?
Jun 10 10:11:43 <rburton> Seemed like the sort of thing you'd know ;)
Jun 10 10:12:24 <Letothe2nd> rburton: i now whom i would get in touch with if i had something to discuss concerning that, no idea if that also the organizing person :)
Jun 10 10:12:28 <Letothe2nd> *know
Jun 10 10:12:37 <rburton> who is that?
Jun 10 10:13:09 <Letothe2nd> rburton: david reyna
Jun 10 10:13:21 <rburton> I shall hunt him down later, thanks
Jun 10 10:13:26 <Letothe2nd> have fun.
Jun 10 10:29:41 <chris_ber> Letothe2nd: when i execute cmake itself folowed by make the shared library (incl symlink xx.so-> xx.so.1.0.0) are created. So maybe i missing something in the recipe? any idea?
Jun 10 11:39:09 <kroon> RP, regarding the features_check/warn-if-unused thing, could you elaborate on the parsing hit ? I did some before/after measurements but I can't detect any noticeable differences
Jun 10 11:44:27 <Letothe2nd> chris_ber: then are you maybe installing to non-canonical places? but no real idea, sorry.
Jun 10 11:46:46 <chris_ber> ok
Jun 10 11:53:30 <RP> kroon: out of interest why couldn't you use BBINCLUDED ?
Jun 10 11:56:48 <kroon> RP, as I see it, I can only warn of those variables are unset *and* features_check is inherited directly in the .bb file. If its in an .inc it could be false positive (which there are in OE-Core). I don't think that information can be extracted from BBINCLUDED
Jun 10 11:58:43 <kroon> RP, for instance there is a "inherit features_check" in xorg-lib-common.inc, but not all xorg bb recipes that include it use the feature check functionality
Jun 10 12:01:54 <RP> kroon: the xorg issue looks like its aviodable by checking for None vs "" in the set value?
Jun 10 12:03:27 <RP> kroon: I understand why you've built it the way you have but I fear its trying to be a little bit too clever :/
Jun 10 12:06:15 <kroon> RP, xorg was a bad example
Jun 10 12:06:33 <kroon> RP, I think
Jun 10 12:06:44 <kroon> RP, libva-initial is another one
Jun 10 12:07:24 <kroon> RP, it includes libva.inc, but doesnt set any of the variables
Jun 10 12:08:13 <RP> kroon: shouldn't that just be cleaned up?
Jun 10 12:08:29 <kroon> RP, how ?
Jun 10 12:08:43 <kroon> RP, move the inherit from the inc ?
Jun 10 12:08:43 <RP> kroon: dropped from libva.inc ?
Jun 10 12:09:10 <kroon> RP, id say that depends on how recipes include the .inc file
Jun 10 12:09:20 <kroon> RP, *how many recipes include*
Jun 10 12:09:32 <RP> kroon: ok, but looking at the way those recipes use the inc, it can be dropped
Jun 10 12:09:59 <kroon> RP, then there is glib and glibc
Jun 10 12:11:24 <RP> kroon: glib doesn't use it at a quick glance?
Jun 10 12:11:33 <kroon> RP, well, glib should just be fixed, right
Jun 10 12:12:07 <RP> can't spot any glibc usage immediately either
Jun 10 12:12:11 <kroon> right
Jun 10 12:12:19 <RP> so we just fix those?
Jun 10 12:14:35 <kroon> RP, ya, and bluez5
Jun 10 13:25:58 <ka6sox> Good Morning, If you want to use cve-check.bbclass what is the best practice for adding cve-update-db in Zeus?
Jun 10 13:34:45 <rburton> the class is already in zeus isn't it?
Jun 10 13:39:35 <ka6sox> yes, the class is there but to use it there is a cve-update-db recipe. what I'm trying to understand is if the parser knows that needs to be run first or you need a separate build to update the db.
Jun 10 13:40:48 <rburton> the class knows to run the tool
Jun 10 13:40:50 <neverpanic> mcfrisk: ^?
Jun 10 13:46:18 <chris_ber> hwen i have to versions of my recipe: app_0.0.1.bb and app_0.0.2.bb how can i instruct bitbake to install one specific version?
Jun 10 13:47:18 <Letothe2nd> chris_ber: PREFERRED_VERSION, see ref-manual
Jun 10 13:47:55 <chris_ber> o
Jun 10 13:49:52 <chris_ber> ok
Jun 10 13:52:11 <qschulz> chris_ber: in your distro conf file or machine conf file, elsewhere it has no effect
Jun 10 13:54:05 <Letothe2nd> qschulz: hey you just ruined the fun!
Jun 10 13:59:38 <qschulz> Letothe2nd: sorry, I foresaw the next question :)
Jun 10 14:03:53 <Letothe2nd> qschulz: i did too, yet i remained purposefully silent
Jun 10 14:21:54 <armpit> I am adding dunfell to patchwork
Jun 10 15:33:45 <chris_ber> Does somebody experienced the same? changes in CMakeLists.txt (e.g. changing SOVERSION) are unrecognized. When i update the content of the recipe it will be
Jun 10 15:35:19 <Letothe2nd> chris_ber: is it still a combined metadata-source thing?
Jun 10 15:36:09 <chris_ber> you don't wanna hear this: yes
Jun 10 15:36:20 <Letothe2nd> then you screwed yourself.
Jun 10 15:36:27 <Letothe2nd> even if you don't wanna hear it
Jun 10 15:36:41 <qschulz> chris_ber: you probably have used devtool before right and forgot to reset?
Jun 10 15:36:51 <qschulz> or forgot you reset and still use devtool workspace
Jun 10 15:37:04 <Letothe2nd> reason: bitbake looks at the recipe, sees no change, decides not to act.
Jun 10 15:37:24 <chris_ber> qschulz: no, i am not using devtool
Jun 10 15:37:32 <Letothe2nd> if you had followed given advice (separate repo, or at least EXTERNALSRC) it would have worked.
Jun 10 15:37:38 <qschulz> chris_ber: what do you have in SRC_URI?
Jun 10 15:37:44 <Letothe2nd> blame your wasted time at your unwillingness to listen, then.
Jun 10 15:37:46 <chris_ber> glob
Jun 10 15:37:53 <Letothe2nd> and with that, i call it a day.
Jun 10 15:38:18 <qschulz> chris_ber: there is no glob support in SRC_URI
Jun 10 15:38:23 <qschulz> we told you yesterday
Jun 10 15:38:31 <qschulz> files or directories only
Jun 10 15:39:24 <qschulz> the issue is that Yocto pulls the files correctly the first time (not even guaranteed?) but the files aren't watched over
Jun 10 15:39:30 <qschulz> when you're using a glob
Jun 10 15:39:48 <chris_ber> tomorrow i restart over and do it your way. i am sick of getting results and got stuck in another place.
Jun 10 15:42:28 <qschulz> chris_ber: whatever suits you best. You need to give directories and files to SRC_URI only, explicitely, not with glob or regexp. If you just explicit them, you can keep your (bad practice) current way of doing it
Jun 10 16:09:10 <lukma> Maybe a strange question - I'm using podman/docker to build Yocto with some older OSes
Jun 10 16:09:10 <lukma> The problem is that I need to build it as root (as this avoids a lot of issues with UID/GID mismatches on namespaces and mount directories)
Jun 10 16:09:10 <lukma> I do know that just running bitbake as root is forbidden and can be fixed with using:
Jun 10 16:09:10 <lukma> meta/conf/sanity.conf (#INHERIT += "sanity")
Jun 10 16:09:10 <lukma> But is there any plan to make it optional (as use case in containers)?
Jun 10 16:12:03 <neverpanic> lukma: Running builds as root hides operations that are not properly run under pseudo, Yocto's fakeroot implementation.
Jun 10 16:12:46 <neverpanic> When running builds as user, such issues will be caught by a sanity check.
Jun 10 16:12:59 <paulbarker> lukma: Have you looked into the CROPS containers? They're designed to fix exactly that issue
Jun 10 16:13:17 <paulbarker> A startup script in the container ensures uids/gids match
Jun 10 16:13:21 <smurray> or JPEW's pyrex tooling
Jun 10 16:14:48 <JPEW> lukma: Pyrex is designed to preserve the proper permissions of your user when running bitbake in a container
Jun 10 16:20:45 <JPEW> lukma: https://github.com/garmin/pyrex
Jun 10 18:20:03 <rburton> lukma: also some upstream packages *refuse* to build if root, if you want more reasons to run builds as non-root user
Jun 10 18:45:01 * Letothe2nd gets beer, starts late shift.
Jun 10 18:46:21 * qschulz should really leave the office now :)
Jun 10 18:46:49 <Letothe2nd> qschulz: alone there? best time to get $stuff done!
Jun 10 18:47:18 <qschulz> 9pm before a holiday? sure thing I'm alone at the office
Jun 10 18:47:32 <Letothe2nd> best time to get $STUFF done.
Jun 10 18:53:59 <Letothe2nd> happy dance due to random click: git blame annotations in vs code. /me likes!
Jun 10 19:02:27 <khem> smurray: or yoe distro also uses docker container :)
Jun 10 19:05:03 <Letothe2nd> kas also supports docker
Jun 10 19:05:29 <Letothe2nd> i think its harder to find proper project not relying on a docker build than those which do these times.
Jun 10 19:15:58 <khem> yeah building master I prefer to use archlinux, but for releases docker is so much better, I can still use archlinux of today to build morty
Jun 10 19:16:26 <khem> its that pill which gets sweeter with time :)
Jun 10 19:16:39 <smurray> heh
Jun 10 19:17:26 <smurray> I do my own tinkering with master on Fedora if it's not broken, otherwise a debian stable container
Jun 10 19:18:06 <khem> broken ? fedora or OE or both
Jun 10 19:22:55 <smurray> khem: e.g. the occasional bleeding edge brokenness like gcc 10 in F32
Jun 10 19:23:02 <khem> ah yes
Jun 10 19:23:37 <khem> yeah someone has to fix them and unfortunately its some of us at that edge
Jun 10 19:24:43 <khem> in my experiments I see systemd images have 30MB extra ram usage compared to sysvinit
Jun 10 19:25:01 <khem> regardless of musl or glibc
Jun 10 19:33:13 <smurray> khem: runtime, you mean?
Jun 10 19:41:52 <khem> yes
Jun 10 19:42:37 <khem> so certainly not a lot of you have modern machines, but still a lot if you have 32MB flash and 128MB DRAM parts
Jun 10 19:45:04 <smurray> khem: yeah.  Is it spread around, or does e.g. journald seem to be larger than expected?
Jun 10 19:50:23 <khem> its spread
Jun 10 19:50:49 <khem> its fresh isntall so journald is not housing much
Jun 10 19:55:54 <smurray> gotcha
Jun 10 21:45:40 <paulg> v5.2.x SRBDS released, for all those who monitor CVEs.    https://lists.yoctoproject.org/g/linux-yocto/message/8744
Jun 10 21:47:57 <paulg> https://www.vusec.net/projects/crosstalk/    <------   CVE-2020-0543
Jun 10 21:49:53 <paulg> the v5.4.x and v5.7.x can be found in the "normal" integrations of linux-stable that zeddii does.
Jun 10 23:21:50 <armpit> paulg, thanks
Jun 10 23:27:32 <paulg> armpit, as always I'm glad to see it used.   There have even been side benefits, like finding backport issues in v5.4
Jun 10 23:27:43 <paulg> https://www.spinics.net/lists/stable/msg392076.html
**** ENDING LOGGING AT Thu Jun 11 02:59:59 2020