**** BEGIN LOGGING AT Sat Mar 05 02:59:57 2011 Mar 05 03:01:04 have one quick ?: how come there is no recipes to build samples? Lets say I built GTK+, how do I build sample for gtk? Mar 05 04:35:48 03Khem Raj  07master * r3fac2c7d24 10openembedded.git/recipes/net-snmp/ (5 files): Mar 05 04:35:49 net-snmp: Fix build error ./.libs/libnetsnmpmibs.so: undefined reference to `pthread_create' Mar 05 04:35:50 Do not override do_configure and pass endianness information in Mar 05 04:35:50 EXTRA_OECONF, also move common settings from EXTRA_OECONF in recipes to Mar 05 04:35:51 inc file. Mar 05 04:35:52 Let build go in parallel for svn recipe. It works ! Mar 05 04:35:52 Signed-off-by: Khem Raj Mar 05 08:02:10 03Thomas Zimmermann  07org.openembedded.dev * rc9308736b3 10openembedded.git/recipes/shr/shr-settings_git.bb: Mar 05 08:02:10 shr-settings: bump SRCREV Mar 05 08:02:10 This is to update the timezone lable after the timezone has changed. Mar 05 08:02:10 Signed-off-by: Thomas Zimmermann Mar 05 09:50:44 gm Mar 05 09:56:55 hi effem Mar 05 10:34:33 03Andrea Adami  07org.openembedded.dev * ra39c5a9a32 10openembedded.git/recipes/kexecboot/ (kexecboot-klibc_git.bb kexecboot_git.bb): Mar 05 10:34:34 kexecboot: bump to SRCREV 18e5620adf2f2bdf319a363586e923e3bb031848 v.0.5.8 Mar 05 10:34:34 * fixes some hidden bugs Mar 05 10:34:34 Signed-off-by: Andrea Adami Mar 05 11:08:12 is oe alsways stripping binary ? Mar 05 11:13:42 rob_w: I think, that it does, but in -dbg package you can find unstripped version Mar 05 11:15:06 ah nice Mar 05 11:15:23 so i then only need ulimit to get core dumps of a crash ,, thx ynezz Mar 05 11:16:18 03Andrea Adami  07org.openembedded.dev * r96c276ddd7 10openembedded.git/recipes/kexecboot/ (5 files in 3 dirs): Mar 05 11:16:18 kexecboot: remove deprecated kexecboot-logo customization: Mar 05 11:16:19 * old code was doing nothing: Mar 05 11:16:19 * there are now themes in kexecboot Mar 05 11:16:19 Signed-off-by: Andrea Adami Mar 05 11:17:40 03Andrea Adami  07org.openembedded.dev * rb4bbb39728 10openembedded.git/recipes/kexecboot/files/ (angstrom/icon.xpm icon.xpm): Mar 05 11:17:40 kexecboot: rename FILS dir because it's now kexecboot-cfg specific. Mar 05 11:17:40 Signed-off-by: Andrea Adami Mar 05 11:23:58 hmm cant find ulimit busybox cmd .. grrr Mar 05 11:25:47 it should be alias in ash Mar 05 11:26:24 ah Mar 05 11:26:37 its builtin .. no alias Mar 05 12:01:31 * Jay7 have testbuilder 0.4.. Mar 05 13:33:32 hm.. IMAGE_FSTYPES can't be totally overrided Mar 05 13:34:00 it always have "jffs2 tar.gz" appended Mar 05 13:37:43 IMAGE_FSTYPES="jffs2 tar.gz jffs2 tar.gz" Mar 05 13:37:47 this is funny Mar 05 14:05:07 03Yuri Bushmelev  07master * r99510d5479 10openembedded.git/contrib/testing/testbuilder/ (7 files in 3 dirs): contrib/testing/testbuilder: update to 0.4 Mar 05 14:06:42 ynezz: check it ;) Mar 05 14:08:49 oh, you should've done it yesterday, my builds just finished :) Mar 05 14:09:21 is there any option to keep the images somewhere? so it can be boot tested also Mar 05 14:09:42 hm.. no, but this look like good idea Mar 05 14:09:57 I'm doing just the clean builds, inceremental probably has them Mar 05 14:10:58 well, I'll try to implement something like before next testing-next :) Mar 05 14:11:15 btw, you may specify any variable now in build config file Mar 05 14:12:03 I'll rerun the builds, no problem Mar 05 14:12:03 that is wrt your ANGSTROM_LIBC :) Mar 05 14:12:10 yep, i got it Mar 05 14:12:49 and now you have last line from build log right inside tb.log Mar 05 14:13:00 with some attempt to guess build status Mar 05 14:14:18 it should rather create table markup for that crazy mediawiki format :) Mar 05 14:14:28 wow Mar 05 14:14:30 it's insane to edit it Mar 05 14:14:31 nice idea Mar 05 14:14:51 it's pretty possible Mar 05 14:15:26 * Jay7 still can't put all builds results to that table.. Mar 05 14:15:39 why is that? Mar 05 14:15:42 but with ready markup it should be easy Mar 05 14:15:50 ah, the same problem Mar 05 14:15:51 :) Mar 05 14:16:23 ynezz: 10x for idea, I'll try to implement it :) Mar 05 14:20:51 well, some workaround till there's something better Mar 05 14:21:07 like that jenkins or something Mar 05 14:25:40 Jay7: 0.4 looks nice :) Mar 05 16:00:16 03Frans Meulenbroeks  07org.openembedded.dev * r39e87f7315 10openembedded.git/recipes/openssl/ (openssl-0.9.8m/Configure-nios2.patch openssl_0.9.8m.bb): Mar 05 16:00:16 openssl 0.9.8m: added linux-nios2 definition Mar 05 16:00:16 conditional patch for nios2 only Mar 05 16:00:16 as others are not affected, no PR bump needed Mar 05 16:00:16 Signed-off-by: Frans Meulenbroeks Mar 05 16:01:18 03Andrea Adami  07org.openembedded.dev * raa379d1ca9 10openembedded.git/recipes/linux/linux-kexecboot_git.bb: Mar 05 16:01:18 linux-kexecboot_git: bump to b65a0e0c84cf489bfa00d6aa6c48abc5a237100f Mar 05 16:01:18 Signed-off-by: Andrea Adami Mar 05 16:49:07 ERROR: Trying to resolve runtime dependency gdk-pixbuf-loader-gif resulted in conflicting PREFERRED_PROVIDER entries being found. Mar 05 16:50:47 Am I the only one getting this sort of spewage recently? As far as I can tell, I have PREFERRED_PROVIDERS for these entries set correctly (they worked up until recently), but now I get a number of these errors, with futher messages that offer solutions (which don't work). Mar 05 16:51:47 http://pastebin.com/cBfe97w0 Mar 05 16:52:01 ^ Full error text Mar 05 16:52:09 mwester: I've seen that Mar 05 16:52:23 and even reported here Mar 05 16:52:31 ah Mar 05 16:52:32 w/o any answer :) Mar 05 16:52:34 Oh Mar 05 16:52:35 :( Mar 05 16:53:00 Well, then I guess nobody knows the answer :) It's a mystery! Mar 05 16:53:16 nobody tried imho :) Mar 05 16:53:30 or just nobody cares Mar 05 16:54:25 Well, most folks are asleep all weekend, so perhaps I shall poke and prod on Monday to get some guidance. It doesn't seem to break anything, but since I grep the logs for the string "ERROR:" on my autobuilder, it is making my autobuilder messier. Mar 05 16:56:30 hm.. freebsd date tool is much more advanced compared to gnu one.. Mar 05 16:56:58 * Jay7 is looking how to convert unix time to human readable time Mar 05 16:57:29 seems awk is easier way.. Mar 05 17:03:22 Jay7: ? date -d @timetstamp Mar 05 17:03:34 ynezz@ntbk:~$ date -d @1297955283 Mar 05 17:03:34 Thu Feb 17 16:08:03 CET 2011 Mar 05 17:03:38 if you mean that Mar 05 17:04:02 date (GNU coreutils) 8.5 Mar 05 17:04:17 hm.. Mar 05 17:04:26 where should I look to find @123123123? Mar 05 17:05:01 info coreutils 'date invocation' Mar 05 17:05:02 :) Mar 05 17:05:13 ah :) Mar 05 17:05:20 date input formats Mar 05 17:05:31 well.. then I take my words back :) Mar 05 17:05:37 good gnu date, good :) Mar 05 17:08:02 Ah, but does busybox support that? Mar 05 17:08:20 you mean ash? maybe, they're crazy enough to support that :) Mar 05 17:09:19 ah, sorry, not ash Mar 05 17:09:22 * ynezz confused Mar 05 17:10:18 * Jay7 is using gnu date so don't care about busybox one :) Mar 05 17:12:27 hm.. other question Mar 05 17:12:50 how to convert interval in seconds to human readable form like 1:14:23 Mar 05 17:13:12 pfff, maybe you should switch to python :) Mar 05 17:13:15 use timestamp to time conversion in UTC? :) Mar 05 17:13:28 ynezz: yeah.. definitely should one time :) Mar 05 17:14:46 well, I hope no one single build will continue more than 24 hours :) Mar 05 17:16:24 mwester: can u try this patch http://pastebin.com/YDddmntp Mar 05 17:16:45 mwester: for the provider conflict proble Mar 05 17:23:04 Slightly different set of error mesages now. http://pastebin.com/V5494B82 Mar 05 17:23:18 re, gm everyone Mar 05 17:23:43 Note that it selects version 2.20.1 Mar 05 17:23:46 Odd. Mar 05 17:29:12 hello Mar 05 17:29:35 can soemone help me with this error building a package with a homemade recipe ? Mar 05 17:29:38 ERROR: QA Issue with marble: non debug package contains .debug directory:... Mar 05 17:29:55 mwester: in conf/distro/include/preferred-slugos-versions.inc we pin gtk+ an gtk+-native Mar 05 17:30:25 I have this in my recipe "FILES_${PN}-dbg += "${libdir}/marble/plugins/.debug"" Mar 05 17:30:29 mwester: is there a reason to stay at 2.18.6 Mar 05 17:32:00 GarthPS: hmm there is some other regexp which is generic enough in your recipe Mar 05 17:32:18 that gets FILES populated with all this Mar 05 17:32:35 find out what all .debug dirs exist in your install tree Mar 05 17:32:39 khem: so I can remove this line Mar 05 17:32:40 ? Mar 05 17:33:04 um no Mar 05 17:33:12 khem, no reason to pin it any longer. I shall try removing that. Mar 05 17:33:15 this line itself seems ok Mar 05 17:33:25 khem: ok no I understaood. Mar 05 17:33:38 khem: I need to specify the right path that is right ? Mar 05 17:34:03 yes Mar 05 17:38:01 mwester: problem is that we use BBCLASSEXTEND to get native version of recipe but it does not transform the PROVIDES Mar 05 17:41:46 ynezz: new testbuilder is almost ready ;) Mar 05 17:43:17 mwester: hmmm for me it does not give the error MACHINE=nslu2be DISTRO=slugos on git/master Mar 05 17:47:42 03Frans Meulenbroeks  07master * rf14b336a25 10openembedded.git/recipes/binutils/binutils_2.17.50.0.12.bb: (log message trimmed) Mar 05 17:47:42 binutils 2.17.50: fixed nios2 build Mar 05 17:47:42 apparently somewhere along the line the nios build for this pacakge Mar 05 17:47:42 was broken. This fixes it: Mar 05 17:47:42 added -Wno-error as there were a few warnings in the nios2 specific part Mar 05 17:47:42 added LDFLAGS to fix GNU hash QA error Mar 05 17:47:43 both changes are nios2 specific so _append_nios2 Mar 05 17:50:41 khem: thx it is ok now. I have adde a new def of FILES_${PN}-dbg and it works. I need to remember this one.. :) Mar 05 17:55:43 khem, what bitbake version? Mar 05 17:57:12 ynezz: pushed ;) Mar 05 17:57:13 03Yuri Bushmelev  07master * r62c69a90e3 10openembedded.git/contrib/testing/testbuilder/ (3 files in 2 dirs): contrib/testing/testbuilder: update to 0.5 Mar 05 17:57:47 ynezz: look into logs/MW.* files after build Mar 05 17:58:09 there should be 2 files - one for testing table format and one - for release Mar 05 17:58:15 they differs slightly Mar 05 17:59:22 now someone may copy code from testbuilder to testscript.sh to have build results in mediawiki format as well :) Mar 05 18:07:31 * Jay7 gone Mar 05 18:20:47 03Mike Westerhof  07org.openembedded.dev * re209fbe0de 10openembedded.git/conf/distro/include/preferred-slugos-versions.inc: Mar 05 18:20:47 SlugOS: preferred-slugos-versions.inc - remove ipkg and gtk+ preferred versions. Mar 05 18:20:47 Signed-off-by: Mike Westerhof Mar 05 18:20:59 03Mike Westerhof  07org.openembedded.dev * r93eebc86b8 10openembedded.git/recipes/linux/linux-ixp4xx/nslu2/defconfig-2.6.37: Mar 05 18:20:59 SlugOS: kernel defconfig-2.6.37 - update to add device support and remove debug Mar 05 18:20:59 Signed-off-by: Mike Westerhof Mar 05 18:21:01 03Mike Westerhof  07org.openembedded.dev * r7c5d5983d3 10openembedded.git/recipes/meta/slugos-packages.bb: Mar 05 18:21:01 SlugOS: slugos-packages.bb - demote vsftpd until the recipe can be repaired. Mar 05 18:21:01 Signed-off-by: Mike Westerhof Mar 05 19:20:02 mwester: its git/master Mar 05 19:20:59 mwester: oh and with this patch applied http://pastebin.com/HTsRE66n Mar 05 19:21:06 to OE metadata Mar 05 19:21:26 mwester: if you can try this would be nice Mar 05 19:26:19 he khem Mar 05 19:34:55 wog_train: train to ? Mar 05 19:37:16 hmmm I really can't remember, is there some tslib init script Mar 05 19:37:21 how did it work in tslib times Mar 05 19:37:21 back to berlin Mar 05 19:37:21 from cebit Mar 05 19:37:22 ? Mar 05 19:41:54 hi wog_train Mar 05 19:42:04 yews Mar 05 19:42:25 ? Mar 05 19:49:33 *g* Mar 05 19:49:43 lags Mar 05 19:49:47 ok Mar 05 20:07:48 wog_train, do you recall if there was a ts_calibrate script in oe? Mar 05 20:08:02 hm yes Mar 05 20:08:03 like for instance long time ago Mar 05 20:08:13 seppreat package Mar 05 20:08:16 do you have hints for finding it? Mar 05 20:08:25 look at the recipe Mar 05 20:08:25 ok Mar 05 20:08:26 maybee Mar 05 20:08:54 do you know some keywords to grep? Mar 05 20:09:00 or to git+grep Mar 05 20:11:47 there is only a tslib.sh which contain a link for the /dev/touchscreen0 dev node Mar 05 20:13:11 yes Mar 05 20:13:37 I'm looking for an init script that would do that: Mar 05 20:13:53 [ -e /etc/pointercal ] Mar 05 20:13:57 ts_calibrate Mar 05 20:14:31 I forgott the if Mar 05 20:14:41 but I bet you got it anyway Mar 05 21:19:07 hi GNUtoo|laptop check openembedded/recipes/qt4/qt-demo-init/qtdemo-init Mar 05 21:19:56 ok Mar 05 21:20:33 thanks a lot Mar 05 21:21:02 it will be usefull, altough I will have to fork it Mar 05 21:21:18 I was looking for some place where it was already made and I would just have to override it Mar 05 21:21:30 but that doesn't seem to exist Mar 05 21:22:24 anyway my bug 1.x touchscreen have touch broken,so I'll make someone else do the init script instead of me, because I can't test Mar 05 21:22:38 hi btw Mar 05 22:23:28 * mwester curses bluez, gtk, X11, and everything else that creeps into embedded systems. Mar 05 22:23:43 Fork it. Mar 05 22:24:05 And avahi too. Mar 05 22:24:29 What the **** ever happened to Linux? When did even an embedded device become a Windows clone??? Mar 05 22:25:28 mwester: just create bluez-nox11 recipe :) Mar 05 22:25:33 or gstreamer-nox11 Mar 05 22:25:44 mwester, when embedded linux morphed into a device with human inputs and a screen. Mar 05 22:26:50 That's what I'm going to do. Someone will yell, but this won't get fixed unless someone pushes the issue. :) So I shall clean up my tmpdir, and start working on that. Right now I can't even build the feeds for SlugOS anymore; the last gtk+ change results in a build failure (I have no idea how long the autobuilder has been building gtk+ unnecessarily!) Mar 05 22:27:05 ka6sox-away, then it's no longer embedded, is it? Mar 05 22:27:31 Call it "mobile" linux, or "small system" linux, but not embedded. Mar 05 22:27:42 mwester, touche Mar 05 22:28:17 This is all going to get much better when we do the oe-core split. :) Mar 05 22:28:35 I hope Mar 05 22:28:37 I'm so looking forward to that happing. Mar 05 22:28:56 but seems bluez will be out of core :) Mar 05 22:30:24 Jay7, I think bluez is itself broken. It seems to straddle the divide between embedded and mobile, if I can use those terms. There should be a clear way to control the the underlying hardware and manage basic bluetooth *server* services that does not reqiure us to build GTK+... Mar 05 22:31:15 why is everybody blaming gtk+ for all the evil in the world these days ? Mar 05 22:31:31 Because building it is a Royal PITA. Mar 05 22:32:13 dont say that so often Mar 05 22:33:09 GTK+ is marvelous, IF you intend on actually using it in the final image. If not, then it's just many 100's of MBytes of disk space, and several hours of build time, for all the unnecessary userspace libs and dependencies (which go far beyond GTK, but it is GTK that is the broken bit out of the entire collection of unnecessary stuff right now) Mar 05 22:33:09 * rob_w gently touches his gtk buttons , whispering to them Mar 05 22:33:41 It was Pango the last time this came up. Mar 05 22:33:57 I don't recall how I fixed it, but clearly I didn't fix the root cause. :) Mar 05 22:34:26 And before that, back in the openmoko days, it was gstreamer that broke everything. Mar 05 22:34:39 hehe Mar 05 22:34:58 still gstreamer is a good one in his genre Mar 05 22:35:14 I'm sure it is. But hell for dependencies. Mar 05 22:35:27 especially when you need none of them in the final image! Mar 05 22:35:40 So, it is time to fix it *right*. Mar 05 22:36:07 so you would need to set the dependencys on the configure stages new ? Mar 05 22:36:13 Yes. Mar 05 22:36:19 nice Mar 05 22:36:30 actaully "detected" ? Mar 05 22:36:39 No, that's part of the problem. Mar 05 22:37:02 We want to be explicit about it, so we would have to pass in --disable or --without flags to configure. Mar 05 22:37:38 so whats the plan then ? Mar 05 22:38:09 a) find the root cause/recipe Mar 05 22:38:39 b) create -noX variations of that recipe (and dependencies, if required) Mar 05 22:38:44 c) commit the recipes Mar 05 22:39:00 d) respond to the argument that will flare up on the ML :D Mar 05 22:39:01 iirc, that recipes chain is known Mar 05 22:39:11 e) fix it right with feedback from the community. :) Mar 05 22:39:15 bluez -> gstreamer -> something -> x11 Mar 05 22:39:25 can't remember 'something' Mar 05 22:40:06 That's my recollection as well, Jay7 -- but avahi just popped up in my dependency chain some moments ago, so I wonder if something has changed. Mar 05 22:40:15 I want to investigate a bit more. Mar 05 22:40:15 yeah, the right answer to that particular problem is fairly straightforward: bluez shouldn't be a monolithic recipe. Mar 05 22:40:40 in bluez 3.x days, the gstreamer bits were a separate recipe but unfortunately they got munged into the main .bb file when it was updated to 4.x. Mar 05 22:41:54 I shall take a bluetooth USB thingy (<- technical term) with me on my business trip next week, and spend some time hacking on this with my travelling NSLU2. :) Mar 05 22:42:35 * mwester goes rummaging through his drawer of USB thingies. Mar 05 23:25:59 mwester, why not making distro specific --without-x ? Mar 05 23:26:30 like for instance: Mar 05 23:27:28 EXTRA_OECONF_slugos += "--without-x" Mar 05 23:28:33 hm.. DEPLOY_DIR can't be redefined seems Mar 05 23:29:13 * Jay7 is looking for way to change DEPLOY_DIR_IMAGE Mar 05 23:29:47 deploy dir is different between SHR and Angsrtom Mar 05 23:29:56 yes, I know Mar 05 23:30:03 as Angstrom support uclibc too Mar 05 23:30:04 ok Mar 05 23:30:34 so I don't like idea to guess where images are stored Mar 05 23:30:45 better to change deploy dir to known place Mar 05 23:33:08 DEPLOY_DIR ?= "${TMPDIR}/deploy" Mar 05 23:33:10 hm.. Mar 05 23:33:19 it should be possible to redefine then Mar 05 23:35:12 but this doesn't work Mar 05 23:51:17 GNUtoo|laptop, overrides work for *adding* things, but I also need to remove things from the DEPENDS Mar 05 23:51:40 ok Mar 05 23:51:45 you could filter stuff Mar 05 23:51:54 or add a disable Mar 05 23:52:08 hmmm Mar 05 23:52:14 that is complicated Mar 05 23:52:20 because of the autodetection stuff Mar 05 23:52:26 but if you have only 1 recipe Mar 05 23:52:37 you could add --disable-x Mar 05 23:52:42 and still pull X as a dep? Mar 05 23:52:56 I could restructure the whole recipe, and I think that will be how it will end up. But I think it easier to create a parallel recipe, then let the community sort out how to merge the two. Mar 05 23:53:05 It's more than just disabling X. Mar 05 23:53:15 ah? Mar 05 23:53:57 so far, I have modiified bluez4, bluez-libs, and dbus. Mar 05 23:54:05 ok Mar 05 23:54:40 ah, btw, dbus require glib which may depend on X stuff too Mar 05 23:54:56 And I have to disable gstreamer on bluez4. Mar 05 23:55:17 Yes, dbus pulls in glib. And most recently, I think it now pulls in avahi. Mar 05 23:55:49 Or maybe avahi is coming in via python, because dbus depends on python-dbus or something like that. Mar 06 00:00:29 * Jay7 -> sleep Mar 06 00:01:10 Jay7, ping me when you get up :D Mar 06 00:01:39 ka6sox-away: something important? Mar 06 00:02:05 Jay7, nope...continuation. Mar 06 00:02:12 pushing things ahead again. Mar 06 00:02:30 ah.. btw, I've updated a bit wiki page Mar 06 00:02:41 just explained entities Mar 06 00:03:10 the Yocto one? Mar 06 00:03:15 well, I'll be up after ~9 hrs :) Mar 06 00:03:17 ka6sox-away: yes Mar 06 00:03:18 I should "watch" that page Mar 06 00:03:27 Jay7, okay good..I will look Mar 06 00:11:40 03Andrea Adami  07org.openembedded.dev * re0872fd1e2 10openembedded.git/recipes/kexecboot/ (kexecboot-klibc_git.bb kexecboot_git.bb): Mar 06 00:11:41 kexecboot: upgrade git recipes up to version 0.5.9. Mar 06 00:11:41 Signed-off-by: Andrea Adami Mar 06 00:11:53 03Andrea Adami  07org.openembedded.dev * rb1fce1a5b0 10openembedded.git/recipes/linux/linux-kexecboot.inc: Mar 06 00:11:53 linux-kexecboot: bump PR to track new kexecboot 0.5.9 release Mar 06 00:11:53 Signed-off-by: Andrea Adami Mar 06 00:20:51 hmmm what's the best way to add some -I Mar 06 00:20:56 for jni stuff Mar 06 00:21:45 (jni-library.bbclass) Mar 06 00:24:53 I'll add to CFLAGS Mar 06 01:36:35 mwester the autobuilder is setup for you now. **** ENDING LOGGING AT Sun Mar 06 02:59:57 2011