**** BEGIN LOGGING AT Thu Jun 02 02:59:58 2016
Jun 02 10:03:26 <kteza1> Are there any established ways to update yocto rootfs over the air?
Jun 02 10:05:17 <CTtpollard> kteza1:nfsroot or similar?
Jun 02 10:06:26 <rburton> kteza1: the easy option is to ship the package manager and use that
Jun 02 10:06:49 <kteza1> @CTtpollard Will check it out. Not aware. Just wanted to checkout all the options available and research
Jun 02 10:06:52 <rburton> ie have package-management in IMAGE_FEATURES so you have rpm/dpkg/opkg as appropriate, and then publish a feed
Jun 02 10:07:17 <rburton> there's more complex options - meta-swupd uses the Clear updater, and denx have a meta layer using a firmware/rootfs updater that i can't recall the name of
Jun 02 10:13:10 <kteza1> @rburton Thanks. Are there any container based solutions so that I can fallback easily?
Jun 02 10:25:35 <LetoThe2nd> rburton: denx' is meta-swupdate by sbabic
Jun 02 10:26:15 <rburton> yeah that :)
Jun 02 10:28:47 <LetoThe2nd> given the number of people asking for stuff like that lately, one might really ponder creating and selling such a solution.
Jun 02 10:43:45 <zedboard> hi, wanted to ask where i can find information to change the U-Boot SAVE_ENV_OFFSET define
Jun 02 10:44:09 <zedboard> possibly in the local.conf or in a separate layer...
Jun 02 11:29:34 <maelcum> hello. i'm running into the infamous "Can't use 'defined(@array)' (Maybe you should just omit the defined()?)" perl error... in the toradex kernel, but i'd probably hit it later in gtk as well.
Jun 02 11:30:15 <maelcum> is it possible to override distro perl with an older "perl-native" package, maybe?
Jun 02 13:41:55 <willnewton> By default in jethro it seems that the python3 recipe does not install .pyc files for source files in the standard library. Is there a way to enable that?
Jun 02 13:52:46 <rburton> i thought we had it installing pyc but not pyo
Jun 02 13:56:02 <willnewton> rburton, i find .pyc files in site-packages but not in /usr/lib/python3.4
Jun 02 13:56:18 <willnewton> not sure if i have something misconfigured somewhere
Jun 02 14:57:34 <kergoth> we should probably think about aligning bb.process more with subprocess, given the updates to the latter. use CalledProcessError, etc.
Jun 02 15:02:20 <rburton> kergoth: yes
Jun 02 15:02:50 <kergoth> when we did that module and other related modules, no such exception existed :)
Jun 02 15:02:54 <kergoth> no check_output() either
Jun 02 15:03:14 <rburton> i wonder if we can just deprecate the wrapper in bb
Jun 02 15:04:41 <kergoth> i think we still need the SIGPIPE handling hook. not sure, though
Jun 02 15:05:01 <kergoth> if so, we could do a *really* minimal wrapper. Popen() that just defaults the preexec function, check_output() & friends that use that Popen
Jun 02 15:05:12 <kergoth> that's a good idea though, need to check
Jun 02 15:05:34 <kergoth> though, bb.process did have some bits for logging to a file, not sure if we need that or not
Jun 02 15:06:28 <rburton> build.py uses that to log the output
Jun 02 15:06:34 <kergoth> ah, right
Jun 02 15:06:37 <rburton> thats the only user in bitbake+oe-core though
Jun 02 15:10:45 <kergoth> rburton: python3 isn't exhibiting the same sigpipe behavior with subprocess
Jun 02 15:10:58 <rburton> result
Jun 02 15:11:00 <kergoth> test case: subprocess.call('cat /dev/zero | head -c 10 | base64', shell=True) -> cat: write error: Broken pipe on python2, no errors on python3
Jun 02 15:11:15 <rburton> please file a bug so we can remember to prune it
Jun 02 15:11:18 <rburton> or a patch!
Jun 02 15:11:21 <kergoth> so afaict we should be able to just drop 90% of that
Jun 02 15:11:27 <kergoth> we have other wrappers laying around too i think
Jun 02 15:11:33 <kergoth> oe.path has a copy of check_output()
Jun 02 15:11:39 <rburton> yeah
Jun 02 15:24:11 <kergoth> ah, there it is
Jun 02 15:24:32 <kergoth> rburton: there's a new restore_signals argument to Popen, defaults to True, it sets all the signals python normally sets to SIG_IGN to SIG_DFL
Jun 02 15:24:39 <kergoth> was wondering how it was handled
Jun 02 15:38:38 <kergoth> huh, cool, python 3.5 adds subprocess.run() as a new high level api
Jun 02 15:39:06 <kergoth> class subprocess.CompletedProcess
Jun 02 15:39:06 <kergoth> The return value from run(), representing a process that has finished.
Jun 02 15:39:44 <kergoth> ooh, timeout stuff
Jun 02 15:40:25 <rburton> oooo
Jun 02 15:40:27 <kergoth> oh nice, there's a DEVNULL you can pass as stdout/stderr
Jun 02 15:40:31 <kergoth> i got sick of creating that manually
Jun 02 15:40:42 <kergoth> (reading the 3.5 docs on subprocess :)
Jun 02 15:41:48 <rburton> yeah i noticed last week they finally added a convenience thing for dev/null
Jun 02 15:41:50 <kergoth> The start_new_session parameter can take the place of a previously common use of preexec_fn to call os.setsid() in the child.
Jun 02 15:41:52 <kergoth> nice
Jun 02 15:44:56 <aurele> hi everyone
Jun 02 15:47:39 <kergoth> RP: thoughts on deprecating __anonymous? I saw a recent patch added one instead of just python ()
Jun 02 15:48:04 <RP> kergoth: it would seem to be a good idea
Jun 02 15:48:14 <RP> kergoth: certainly at least make it print a warning
Jun 02 15:48:45 <kergoth> yeah, i wouldn't want a hard deprecation yet, should follow typical deprecation process, warn, then next release error or something
Jun 02 15:48:49 * kergoth adds to todo
Jun 02 15:56:43 <aurele> does anyone knows if it is possible to open a devshell in current terminal? (I would like to script something in jenkins, but It would be easyier if I could run a devshell)
Jun 02 16:05:59 <willnewton> rburton, i suspect the problem is that generate-manifest-3.5.py doesn't include __pycache__ in its globs
Jun 02 16:07:44 <kergoth> that does sound about right
Jun 02 16:24:59 <kergoth> ugh, what the hell was i thinking when i wrote oe.patch, this module is *terrible*
Jun 02 16:32:57 <kergoth> this needs major refactoring
Jun 02 16:36:13 <kergoth> Hmm, I think prefuncs/postfuncs are being included in the checksums for any function, but they're only *run* for tasks, not all functions
Jun 02 17:36:39 <bottazzini> guys in do_configure phase, is it possible to have a "if not ${@bb.utils.contains('DISTRO_FEATURES...." ?
Jun 02 17:41:02 <kergoth> bottazzini: not isn't a valid command in shell.
Jun 02 17:41:09 <kergoth> use !
Jun 02 17:43:10 <bottazzini> kergoth, thanks :D
Jun 02 17:43:46 <bottazzini> kergoth, It is a little bit strange also we don't use [ ] for bb.utils contains
Jun 02 17:43:52 <bottazzini> or should I put those ?
Jun 02 17:44:05 <kergoth> ?
Jun 02 17:44:21 <bottazzini> if [ ${@bb.utils.contains('DISTRO_FEATURES... ] ...
Jun 02 17:44:42 <kergoth> that would be pointless, depending on what the contains call is returning for treu and false
Jun 02 17:44:48 <kergoth> not only pointless, but wrong in most cases
Jun 02 17:44:56 <bottazzini> i see
Jun 02 17:45:01 <kergoth> the common case is for the contians() call to return 'true' for success and 'false' for failure
Jun 02 17:45:08 <kergoth>  /bin/true and /bin/false are valid commands
Jun 02 17:45:14 <kergoth> we aren't testing the value, it's running the binary
Jun 02 17:45:25 <kergoth> the if statement only uses [] when running tests, not when running arbitrary commands
Jun 02 17:45:32 <kergoth> you could certainly reconstruct it to do so
Jun 02 17:45:43 <bottazzini> oh
Jun 02 17:45:47 <bottazzini> I understand
Jun 02 17:45:50 <kergoth> i.e. if [ "${@bb.utils.contains('FOO', 'bar', '1', '0'}" -eq 1 ] or something
Jun 02 17:45:50 <bottazzini> makes sense
Jun 02 17:45:57 <kergoth> but it wouldnt' add any real value
Jun 02 17:46:09 <bottazzini> yes indeed it is pointless
Jun 02 17:46:20 <kergoth> specfiically it'd offload comparison logic to the shell when we've already done that at parse time
Jun 02 17:46:31 <kergoth> well, expansion time
Jun 02 17:47:10 <bottazzini> i got it
Jun 02 17:47:12 <bottazzini> thanks :)
Jun 02 17:47:14 <kergoth> np
Jun 02 20:51:26 <TwisteR> Hi! I'm new to Yocto. Having some trouble to launch it on Wandboard Quad (quided by http://wiki.wandboard.org/index.php/Getting_started_with_Yocto_on_Wandboard). Here is the boot log: http://tfsoft.org.ua/~twister/work/wand.log
Jun 02 20:51:57 <TwisteR> looks like u-boot is unable to find the kernel
Jun 02 20:52:12 <TwisteR> guided*
Jun 02 20:53:37 <TwisteR> why does the default u-boot configuration fails in my case?
Jun 02 20:55:01 <TwisteR> btw, I had to use MACHINE=wandboard, as there were no separate configurations (I suppose they are merged together recently and the recipe is out of date a bit)
Jun 02 21:07:14 <TwisteR> Found another howto, which is differs from that one I've mentioned http://www.jumpnowtek.com/wandboard/Wandboard-Systems-with-Yocto.html
Jun 02 21:07:56 <TwisteR> last step this time is mk2parts.sh script
Jun 02 21:08:41 <TwisteR> and a set of other scripts for preparing an microSD card
Jun 02 21:09:50 <TwisteR> so, what is the difference between that approach and using generated SD card image?
**** ENDING LOGGING AT Fri Jun 03 02:59:58 2016