**** BEGIN LOGGING AT Tue Jan 29 02:59:58 2013 Jan 29 09:04:12 Crofton|work: thanks Jan 29 09:04:58 hi hrw Jan 29 10:05:22 mmh... why have images now all the -lic packages? Jan 29 10:06:45 JaMa: might this be caused by dc78ef91a2bf01efb8028c9afbe69e506e016265 which checks for 'd.getVar('LICENSE_CREATE_PACKAGE', True)' evaluating to 'True' for every string (including the default "0") Jan 29 10:06:49 ? Jan 29 10:55:30 good morning Jan 29 10:55:38 hi mckoan Jan 29 10:55:55 I did not know italy is another timezone *g* ;) Jan 29 10:56:05 but again universal chattime Jan 29 10:56:23 morning is after aweking Jan 29 10:56:43 woglinde: yes it's universal chattime Jan 29 10:57:11 awake hrms Jan 29 10:57:22 seems I am till sleepy Jan 29 10:57:31 damn rain today Jan 29 10:58:28 woglinde: hete it's worse, yesterday snow today fog Jan 29 10:59:02 and pollution over top limits :-( Jan 29 11:01:52 Hello everybody. I added a bb layer to compile firefox or chrome.. but both of them have compilation error.. any advise? Jan 29 11:14:45 ensc|w: hmm right, should be false by default Jan 29 11:30:26 hi jama Jan 29 12:07:59 woglinde: hi Jan 29 12:37:52 ensc|w: patch on ML Jan 29 12:52:01 do you know how much memory OE needs to do bitbake? Jan 29 12:52:13 ERROR: ExpansionError during parsing /home/ubuntu/jenkins-setup/meta-linaro/recipes-linaro/images/linaro-image-lamp.bb: Failure expanding variable toolchain_create_sdk_version: ExpansionError: Failure expanding variable METADATA_REVISION, expression was ${@base_detect_revision(d)} which triggered exception OSError: [Errno 12] Cannot allocate memory Jan 29 12:52:21 613MB was not enough Jan 29 12:52:34 check cache/bb-codeparser* Jan 29 12:52:42 if you have more then one then it eats a lot of MEM Jan 29 12:54:27 one Jan 29 12:56:48 hrw h, 4 gig was enough Jan 29 12:56:58 maybe lower bbthreads? Jan 29 12:57:12 lowered to 2 and restarted Jan 29 12:57:44 woglinde: do you know the state of systemd in oe-core? Jan 29 12:57:56 zecke should be workable now Jan 29 12:58:00 with eglibc Jan 29 12:58:07 did not try with uclibc Jan 29 12:58:10 again Jan 29 12:58:23 jama knows more Jan 29 12:58:28 he is king Jan 29 12:58:30 *g* Jan 29 12:58:55 :) Jan 29 12:59:03 hehe Jan 29 12:59:23 2 bbthreads also out of mem Jan 29 12:59:32 meta-systemd is still needed to provide .service files for many recipes Jan 29 12:59:42 is busybox-syslogd disabled and only journald enabled? Jan 29 13:00:03 but systemd recipe itself should work from oe-core (if you don't care about upgrade path) Jan 29 13:01:15 otherwise it is not that hard to switch back to sysvinit Jan 29 13:01:30 if you figured out all the variables Jan 29 13:01:55 and you have to rebuild some packages of course Jan 29 13:02:08 OE has an upgrade path? ;) Jan 29 13:02:09 hi folks Jan 29 13:02:18 hi stefan_schmidt yes Jan 29 13:02:21 it hase Jan 29 13:02:29 stefan_schmidt_w: moin Jan 29 13:02:30 if you are distribution Jan 29 13:02:41 dont forget that Jan 29 13:02:46 woglinde: i know, i know, just kidding :) Jan 29 13:02:48 it have Jan 29 13:02:52 bah Jan 29 13:02:52 hi hrw Jan 29 13:03:00 hrw: I will not be at FOSDEM Jan 29 13:03:07 stefan_schmidt_w: happens Jan 29 13:03:12 hrw: yeah Jan 29 13:04:45 well it had :) Jan 29 13:20:13 woglinde: I wanted to look into meta-systemd with OE-core again (i had to stop halfway through) but now that Rails is the new Php i need to do some upgrades of our shop. :} Jan 29 13:30:15 zecke: Nobody wants to do security audits. You know that already. :) Jan 29 13:31:27 zecke what has your shop do with oe? Jan 29 13:31:50 woglinde: they share his dev time :) Jan 29 13:32:30 stefan_schmidt_w: exactly Jan 29 13:34:28 buy more indians Jan 29 13:35:13 hi rsalveti Jan 29 13:35:22 woglinde: well, it would all be better if hipsters used typewriters instead of rails Jan 29 13:36:58 zecke: Without hipster there would be no rails -> everyone would use PHP -> you have the security problems there :) Jan 29 13:37:04 * stefan_schmidt_w is a bit chatty today Jan 29 13:37:12 woglinde: hey Jan 29 13:37:23 we hear you stefan_schmidt Jan 29 13:38:09 hm, rails feel like alien to OE (python based) Jan 29 13:38:20 I would expect djangonauts... Jan 29 13:38:23 ynezz? Jan 29 13:38:46 ynezz do you know a sane shopsystem opensource with django? Jan 29 13:38:58 sane != opensource :) Jan 29 13:39:05 nah Jan 29 13:39:07 heh, in a few years OE will rebuild itself in C Jan 29 13:39:14 ant? Jan 29 13:39:17 AI Jan 29 13:39:25 aehm Jan 29 13:39:31 HAL9000 Jan 29 13:40:45 stefan_schmidt_w: well, php in itself might not be that bad after all. :} Jan 29 13:44:52 yes its nearly as old as python Jan 29 13:46:53 zecke: agreed, got better over the years. Jan 29 13:48:53 * zecke is in bad mood.. just looked at systemd commits Jan 29 13:49:15 stefan_schmidt_w: http://cgit.freedesktop.org/systemd/systemd/commit/?id=f84190d8cad66f36eef1d851213711244ca297be, "numerous modernizations" Jan 29 13:49:28 well, another hipster generation product :p Jan 29 13:49:32 stefan_schmidt_w: http://cgit.freedesktop.org/systemd/systemd/commit/?id=69bd386c4c0618d686e1b5ac01c0e9658b1539b6 Jan 29 13:50:01 Calling the last one a type fix is funny Jan 29 13:50:05 typo one is nice Jan 29 13:50:09 :) Jan 29 13:50:40 ynezz I like it more than shell sysv foo Jan 29 13:50:50 That almost look like some of the efl commit message Jan 29 13:50:54 but of course it should working Jan 29 13:51:13 I think "oops" is the message with the highest score Jan 29 13:51:24 hopefully Red Hat pays them like the amateurs they are. :} Jan 29 13:51:31 hm, I wonder what's that _cleanup_free_ stuff Jan 29 13:51:43 zecke: They pay in USD. So, yes. :) Jan 29 13:51:51 seeing it for the first time Jan 29 13:51:53 zecke its more lack of developement process Jan 29 13:53:24 ah, those neat gcc attributes Jan 29 13:54:00 http://cgit.freedesktop.org/systemd/systemd/commit/?id=858c33bf6002bcdbea185ddedde8f8d660afc731 Jan 29 13:54:04 and the next commit to fix it Jan 29 13:54:56 zecke: Lennart has the same style raster has it seems. :) Jan 29 13:55:18 woglinde: I like openwrt a lot, their uci config system, ubus (micro bus) instead of dbus etc. Jan 29 13:55:25 ieehks assert Jan 29 13:55:26 okay Jan 29 13:56:04 ynezz convert it to oe Jan 29 13:56:16 I wanted, really Jan 29 13:56:24 but it's a lot of work :) Jan 29 13:56:33 stefan_schmidt_w: 8755586eba8b824851f11aa6c3f8f184ed69a30a is another of the fall outs.. Jan 29 13:56:44 stefan_schmidt_w: and this one has probably even raised a compiler warning... Jan 29 13:57:01 it is such an amateur project.. but he feels like bitching about eudev. Jan 29 13:57:02 ynezz: I hate openwrt for excatly that reason :) Jan 29 13:57:37 zecke: Warnings only help if you look at them. Heros can even ignore errors and still commit :) Jan 29 13:58:05 zecke: Stop right now looking through it. Better now before you start crying :) Jan 29 13:58:25 lack of ci and di Jan 29 13:58:34 but redhat is no company Jan 29 13:59:25 stefan_schmidt_w: well, I will use systemd on the BTS.. because right now we have a respawn shell script and this part (thanks to Pengutronix) is nicely solved Jan 29 13:59:54 zecke: so you want to see it working correctly. I figure. Jan 29 14:00:01 Still, better stop before crying Jan 29 14:01:08 stefan_schmidt_w: well, you can't put systemd in 2MB of flash, could you? :) Jan 29 14:01:20 not to mention dbus... Jan 29 14:01:52 stefan_schmidt_w: http://cgit.freedesktop.org/systemd/systemd/commit/?id=86ed7ec58b9b6a0907bbb3b8d07c930e52915de0 Jan 29 14:02:03 stefan_schmidt_w: this one is funny.. it looks so wrong. :} Jan 29 14:02:32 it's copy&paste :) Jan 29 14:04:32 zecke Pengutronix fixed your script? Jan 29 14:05:02 woglinde: no, they implemented watchdog handling for userspace processes Jan 29 14:05:20 ah hm Jan 29 14:05:25 stefan_schmidt_w: "Warnings only help if you look at them. Heros can even ignore errors and still commit :)" ROFL :) Jan 29 14:05:41 ynezz systemd do needs dbus Jan 29 14:05:49 ups do not Jan 29 14:06:19 woglinde: systemd needs lidbus (for the marshalling) but this is a sensible design choice Jan 29 14:06:32 stefan_schmidt_w: 6d3a739048deab5c10a0debdad788e19898144cc, another 'typo' fix. *sigh* Jan 29 14:07:54 done with browsing of systemd commits. :) Jan 29 14:08:14 any idea how simple tar call can cause "OSError: [Errno 7] Argument list too long"? http://paste.debian.net/230086/ Jan 29 14:08:38 jama underlying fs Jan 29 14:08:53 xfs is know for not handle path stuff longer 512 bytes Jan 29 14:09:04 ext4 Jan 29 14:09:13 and running it in shell works fine Jan 29 14:09:35 the same with tmpfs on other builder Jan 29 14:09:46 hm ah no its driffrent here Jan 29 14:10:27 maybee problem in popen Jan 29 14:11:13 can you run the tar line on shell Jan 29 14:11:19 and see whats happens? Jan 29 14:11:32 15:09:18 < JaMa> and running it in shell works fine Jan 29 14:11:32 15:09:40 < JaMa> the same with tmpfs on other builder Jan 29 14:11:50 ups Jan 29 14:11:55 it works correctly license-destdir is copied to dest Jan 29 14:12:02 then maybee its a problem with popen Jan 29 14:12:29 yes, lets blame python :) Jan 29 14:12:55 fwiw: 2.6.3 and 2.7.2 behave the same Jan 29 14:15:52 jama do you have the line in bitbake at hand? Jan 29 14:17:23 btw. Jan 29 14:17:25 http://docs.python.org/2/library/subprocess.html#frequently-used-arguments Jan 29 14:17:30 Sorry, had a request for pair debugging here :) Jan 29 14:17:58 ynezz: well, I make it easy for myself. I don't build full linux systems for 2MB. As easy as this for me. :) Jan 29 14:18:50 *g* Jan 29 14:22:49 woglinde: trying with debug output added to execute_child in python code Jan 29 14:23:08 hi woglinde Jan 29 14:23:55 ynezz, you want to build a rootfs+kernel for a target that has less than 2M? Jan 29 14:24:48 ynezz, I want to do the same thing.... Jan 29 14:25:04 but my target is x86 Jan 29 14:27:50 I don't use OE for such projects Jan 29 14:28:00 ynezz O.o Jan 29 14:28:11 we klibc Jan 29 14:28:14 + have Jan 29 14:28:22 and I should put dietlibc Jan 29 14:28:37 over from old oe Jan 29 14:28:55 I'm lazy bastard, so I like to reuse the stuff as much as I can Jan 29 14:29:26 ynezz, what do you use? openwrt? Jan 29 14:29:33 since for such lowend devices I find owrt ideal solution Jan 29 14:29:46 ok Jan 29 14:29:57 depends on what you need in the rootfs Jan 29 14:30:06 just something to boot Jan 29 14:30:10 ? Jan 29 14:30:18 like for instance something to replace ipxe Jan 29 14:30:59 woglinde: no big surprise in execute_child http://paste.debian.net/230092/ Jan 29 14:31:13 at 2M I think the rootfs is the last concern, the big issue is the kernel size Jan 29 14:31:36 denisATeukrea: lzma/xz + klibc will do the trick Jan 29 14:31:51 we are under 1MiB Jan 29 14:32:09 yes but how much drivers do you have? Jan 29 14:32:23 all block devices but no net Jan 29 14:32:42 I need net(ath5k, pc-card, evdev and i915) Jan 29 14:33:03 s/(/and/ Jan 29 14:33:11 ah x86 Jan 29 14:33:13 thats tricky Jan 29 14:33:15 yes Jan 29 14:33:17 indeed Jan 29 14:33:27 + I cannot use the whole 2M Jan 29 14:33:28 I got qemuarm with net under 900kb Jan 29 14:33:33 ok Jan 29 14:33:39 I think Jan 29 14:33:59 maybe if I put the VGA option rom instead of the i915 and use vesa it would save some space Jan 29 14:34:19 jama hm Jan 29 14:34:42 jama why do we need it execute with shell? Jan 29 14:36:22 woglinde: for pipes between tars Jan 29 14:40:16 hm why is python this inefficent with files at this point so we have to use tar Jan 29 14:40:35 cant we use cpio Jan 29 14:49:17 don't know.. I've sent all details to ML Jan 29 14:54:41 jama hm its only for copying licenses file and preserving owner and rights Jan 29 14:54:53 why preserve owner Jan 29 14:54:56 wtf Jan 29 14:55:17 ah okay Jan 29 14:55:19 deploy dir Jan 29 14:55:22 not image Jan 29 15:00:04 woglinde: and it's in generic sstate-install Jan 29 15:01:06 I wonder why we do not use the python stuff copying there Jan 29 15:01:21 we dont have 5 million license files Jan 29 15:02:22 true but there has to be something else in this particular case Jan 29 15:02:29 it's not so special to fail so badly Jan 29 16:41:59 meh, glib wants libfffi which lives in recipes-gnome... Jan 29 17:37:16 moin Jan 29 17:44:40 t Jan 29 17:44:41 re Jan 29 19:54:56 hello Jan 29 20:13:09 How can I define a python method in a machine config file? Jan 29 20:13:27 I know it seems strange but it is a quite specific need for internal use Jan 29 20:16:48 hm the doesnt work JDK_ARCH = "${@get_jdk_arch(d)}" Jan 29 20:17:03 def get_jdk_arch(d): Jan 29 20:17:38 not in a machine conf Jan 29 20:18:46 hm ah its conf Jan 29 20:18:51 right Jan 29 20:18:56 sorry no idea Jan 29 20:19:00 you probably have to delay evaluation Jan 29 21:22:19 hi, does something like http://docs.openembedded.org/usermanual/usermanual.html but newer exist? Jan 29 21:37:55 I found the yocto docs to be the most helpful thusfar. Jan 29 21:38:38 sr105: thanks for the pointer, i'll look there Jan 29 21:39:09 they don't answer all questions, but they're the most complete IMO **** ENDING LOGGING AT Wed Jan 30 02:59:58 2013