**** BEGIN LOGGING AT Wed Jan 27 02:59:57 2010 Jan 27 03:01:07 So if I unplug it and plug it back in dmesg gives this: Jan 27 03:01:41 http://pastebin.com/m4e97829c Jan 27 03:03:51 it's saying it cannot load zd1211/zd1211b_uphr -- yet it is there. Jan 27 03:05:56 What am I (conceptually) missing? Jan 27 03:07:37 -2 must not mean "file does not exist" then :) Jan 27 03:07:49 Might wanna dig up the errnos. Jan 27 03:12:12 I'm trying to. Jan 27 03:12:58 -ENOENT :) Jan 27 03:13:08 No such file or directory. Jan 27 03:13:56 Anyone know the bitbake fix amend.bbclass depends on? Jan 27 03:17:04 robtow: well, the kernel requests the firmware to load Jan 27 03:17:09 AlTheKiller - but the file IS there, in the /lib/firmware/zd1211 dir. Jan 27 03:17:26 Kernel seems to think otherwise Jan 27 03:17:26 robtow: maybe it is in the wrong path, maybe no firmware helper is set? Jan 27 03:17:33 You need a depmod -a maybe? Jan 27 03:17:45 I dunno how the fw stuff works. Jan 27 03:18:07 Stuff is loaded for the mouse, keyboard, and usb ethernet dongle. Jan 27 03:18:35 robtow: driver and firmware are still two different things Jan 27 03:37:27 nm, found it Jan 27 03:48:54 zecke... guess it's time to compile the driver. Jan 27 03:49:15 enough for tonight. time to drive up the mountain to home. Jan 27 03:49:23 Thanks, all, for the suggestions. Jan 27 05:40:53 I did a git pull of openembedded.dev earlier today, hoping there was a fix to the mplayer issue. Instead, I wound up with a few gigs of new packages and a bunch of new errors. wtffffffffff. is there any way to revert to a previous git, uh, version? Jan 27 05:41:26 doing bitbake omap3-desktop-image, fwwiw Jan 27 05:46:02 penguinece: this is more a git question. right? Jan 27 05:46:30 zecke, it's actually an oe question nested within a less important git question. Jan 27 05:46:56 penguinece: Yes, you can go to previous git revsions. git reset --help Jan 27 05:47:06 penguinece: and git reflog to see your previous revisions Jan 27 05:47:18 zecke, thanks. any insight on my oe woes? Jan 27 05:47:47 penguinece: just general, communicate the issues in a civilized manner. Jan 27 08:40:56 morning Jan 27 08:43:13 hi Jan 27 08:44:03 * XorA looks for a victim Jan 27 08:44:41 victim for what? victim for hi was XorA :-) Jan 27 08:44:48 at least the trigger :-) Jan 27 08:45:10 * XorA has finally gone insane after a night of fixing autofoo Jan 27 08:45:37 * XorA is also waiting for the email from rday as some of the fixes probably have knockon effects Jan 27 08:46:46 XorA: wb from autohell and welcome to rday hell :-D Jan 27 08:50:40 * XorA is hungry Jan 27 09:00:39 good morning Jan 27 09:01:22 hey mckoan Jan 27 09:05:49 hi Jan 27 09:07:05 * XorA hs food :-D Jan 27 09:08:37 XorA: I've sent you an e-mail about glib instead of rday :) Jan 27 09:09:00 JaMa: arse :-D Jan 27 09:09:23 XorA: and I know about one more issue with newest libtool.. but that should be easy to fix too Jan 27 09:09:46 JaMa: issue in general or one particular package? Jan 27 09:10:38 particular.. the same error as transmission had.. but now it complains that macro is from older libtool than 2.2.6b (from python-gst-0.10.17-r0) Jan 27 09:11:35 JaMa: means autoreconf isnt being called Jan 27 09:14:24 good side it means python-gst was always broke :-D Jan 27 09:20:44 XorA: autoreconf was called but looks like libtool in source directory was used somewhere.. if I remove that in do_configure_prepend it seems to work ok Jan 27 09:22:24 JaMa: /usr/lib/glib-2.0/include/glibconfig.h would appear to be correct location for that header Jan 27 09:22:31 JaMa: so I guess your fix is right Jan 27 09:22:43 * XorA ust checked debian package to double check Jan 27 09:25:09 XorA: on gentoo I have also only in /usr/lib/... Jan 27 09:25:32 JaMa: I think we just had an oddity of that custom staging Jan 27 09:27:35 ok.. I'll push that obexpush fix.. Jan 27 09:28:07 and that libtool is strange in python-gst.. it's recreated in ${S} with do_configure because of this: ./gst-python-0.10.17/codegen/Makefile:LIBTOOL = /bin/sh /home/projects/OE/tmpdir-dev-shr/work/armv4t-oe-linux-gnueabi/python-gst-0.10.17-r0/gst-python-0.10.17/common/shave-libtool '$(SHELL) $(top_builddir)/libtool' Jan 27 09:35:40 JaMa: unfortuneately python normally adds another 2 levels of insanity Jan 27 09:42:34 glib-2.0 and gtk+ could do with a good hoover in those directories :-D Jan 27 09:48:07 XorA: hehe.. evas python bindings (which now segfaults on armv4) seems like 5 levels of insanity to me now :/ Jan 27 09:49:22 who uses armv4 :-D Jan 27 09:53:32 heh I know few strangers with armv4t in their phones... Jan 27 09:54:43 armv4t != armv4 ;d Jan 27 09:55:13 JaMa: http://trac.shr-project.org/trac/ticket/885#comment:1 it seems that illume settings is segfaulting too Jan 27 09:57:39 JaMa: crazy phone people :-D Jan 27 09:57:55 Heinervdm: yesterday I tried to rebuild all without multitouch events in EvasMetaObject class.. but didn't test it yet.. Jan 27 10:00:09 hrw: I think that XorA will be laughing at us even with trailing 't' :) Jan 27 10:01:18 JaMa: do not look at me - my only armv4t on desk is not a phone Jan 27 10:02:18 JaMa: I have two neo1973 and one freerunner :-D Jan 27 10:02:39 * XorA has the first neo1973 to ever make a proper phonecall Jan 27 10:03:06 ~curse jabber Jan 27 10:03:07 May you be reincarnated as a Windows XP administrator, jabber ! Jan 27 10:03:13 jabber rocks Jan 27 10:03:17 ~praise jabber Jan 27 10:03:18 All hail jabber! Jan 27 10:03:34 XorA: I have first which had openmoko 2007.2 running Jan 27 10:03:36 Looks like glib doesnt built at the moment. do_configure fails. The logs are gone (colleague removed his tmp) but IIRC it said something about "invalid token PCRE" Jan 27 10:04:03 hrw, XorA: congratz to both :) Jan 27 10:05:06 hrw: I didnt need 2007.2 picocom was good enough to dial a number :-D Jan 27 10:05:30 XorA: I know, I dialed with picocom too Jan 27 10:05:45 http://marcin.juszkiewicz.com.pl/2007/05/29/how-to-check-does-gsm-modem-works/ Jan 27 10:06:20 I dont think when I made the phonecall the LCD was even turned on :-D Jan 27 10:06:52 * hrw -> off Jan 27 10:14:43 XorA, does glib-2.0_2.22.4.bb have any other dependencies on pcre than glib-2.0_2.22.1.bb had? Jan 27 10:15:05 jovox: pardon? Jan 27 10:17:25 XorA, After pulling yesterday it doesnt build. Complaining in do_configure in glib-2.0_2.22.1.bb. Jan 27 10:18:22 jovox: youd be bast asking on mailing list with full details, its hard to guess what you are seeing Jan 27 10:18:54 I'll try to get back the logs Jan 27 11:54:45 morning all Jan 27 11:55:16 morning Jan 27 11:56:43 hey RP Jan 27 12:11:13 good morning Jan 27 12:26:23 morning guys Jan 27 12:46:11 is it possible to have an image depend on another one? I tried adding it to DEPENDS but it doesn't build the other image first. I need an initramfs image inside my rootfs.. Jan 27 12:47:16 or maybe there's a way to package an image up and then depend on that package in my rootfs? Jan 27 12:50:26 strange, an image is just a recipe, so i would expect you can add it to DEPENDS, although I never did Jan 27 12:51:04 you can definitely add additional commands after the rootfs is created (e.g. add or remove files) before the image is created Jan 27 12:52:00 I added this: ROOTFS_POSTPROCESS_COMMAND += "cat ${DEPLOY_DIR_IMAGE}/xenclient-initramfs-image-xenclient-dom0.cpio.gz > ${IMAGE_ROOTFS}/boot/initramfs.gz ;" Jan 27 13:26:54 th1: iirc, DEPENDS just creates a dependency on do_stage by default, which is probably not what you want. try: do_rootfs[depends] = "otherimage:do_image" Jan 27 13:26:59 or some such Jan 27 13:40:06 whats the right way to create a native recipe these days? Jan 27 13:40:17 the swig recipe is hosed wrt to the dev package Jan 27 13:52:03 anyone have a good intro to BBCLASSEXTEND Jan 27 14:01:01 moblin rocks! http://www.openpeak.com Jan 27 14:06:14 lrg: do you happen to have access to a i.mx27ads? Jan 27 14:10:38 zecke: not any more, only i.mx31 Jan 27 14:10:55 I fried my only mx21 Jan 27 14:36:42 lrg: thanks... I hate to fry hardware... I was doing this to hardware from ulitu with a broken PSU :( Jan 27 14:37:29 morning Jan 27 14:38:03 kergoth: morning! it would be awesome if you could land some of the reviewed patches. :) Jan 27 14:38:30 I think most are still lacking second acks, but i'll push the ones that have them today sometime, thanks for the reminder Jan 27 14:38:50 * kergoth also needs to push newcollection.inc Jan 27 14:44:53 morning kergoth Jan 27 14:45:52 kergoth: you have a fairly long list of improvements you would like in OE -- how about getting OE into GSoC this year ? GSoC 2010 announcement just came out. Jan 27 14:46:02 hmm Jan 27 14:46:43 * kergoth reads up on GSoC, never been involved :) Jan 27 14:46:59 * rkirti is going to make use of the planet linuxtogo aggregation privilege given by florian and blog about getting OE in GSoC Jan 27 14:47:36 kergoth: mickey|fsoMeetin asks you to remove R from the channel Jan 27 14:47:47 he can't speak up with his changed nick. :) Jan 27 14:47:54 Time to bash him ;) Jan 27 14:48:03 R.. channel mode? Jan 27 14:48:07 * kergoth 's irc-fu is rusty, clearly Jan 27 14:48:11 kergoth: yes Jan 27 14:48:40 yay Jan 27 14:48:42 i can talk Jan 27 14:48:43 thanks Jan 27 14:48:47 haha Jan 27 14:48:51 R is registered, i assume? Jan 27 14:48:57 YEP Jan 27 14:48:58 (GSoC) there's a low probability we can get in, but we should try anyways Jan 27 14:49:00 the only reason I took up GSoC last year was because I liked OE and amazing people. I know lots of devels here can make awesome mentors. Jan 27 14:49:25 GSoC is a good idea if we could manage it, there are a *lot* of things that would be nice to get done, but which none of us have gotten around to doing Jan 27 14:50:46 kergoth: I would love to take this up , but only if no-one else has the time, and if everyone agrees to me doing it. At the very least, what I can do is talk to the Google about how I got awesome supoort for me OE-related project last year and give her some good prof of how strong an org OE is and capable for such a task Jan 27 14:50:50 mickey|fsoMeetin: ^^ Jan 27 14:51:09 sounds good Jan 27 14:51:26 Last time I saw OE in GSoC, it was under handhelds.org - do we have possible umbrella orgs this time ? Though I ld say OE is big enough now to apply on its own Jan 27 14:51:27 that sounds fine to me. Jan 27 14:51:37 yep Jan 27 14:51:46 well, OE is an eV now, I'd think we could do it Jan 27 14:52:19 * rkirti fires a blog-post about it and gets going to talk to the GSoC admins. Jan 27 14:53:20 the problem with GSOC is they want to see code Jan 27 14:53:29 not recipes or docs Jan 27 14:53:41 so we need to think of code tasks Jan 27 14:54:56 oh is it gsoc time again? Jan 27 14:55:00 soonish Jan 27 14:55:09 Crofton|work: umm..my GSoC was only recipes :-) Jan 27 14:55:17 hmm Jan 27 14:55:23 you were lucky :) Jan 27 14:55:25 well, there are still a lot of tasks Jan 27 14:55:28 a good GSoC would be a fully cross compiled gnome Jan 27 14:55:29 Crofton|work: surely. Jan 27 14:55:35 consider bitbake imporvements, or features that can be implemented as python in classes or .inc files Jan 27 14:55:47 http://pastebin.ca/1768017 Jan 27 14:55:51 kergoth, right Jan 27 14:56:01 I am trying to fix up the swig recipe Jan 27 14:56:05 so swig-dev works Jan 27 14:56:06 we can always split up into two parts right - one to work on recipes, and other to work on the bitbake core or hacking on python/bash Jan 27 14:56:15 and convert to BBCLASSEXTEND Jan 27 14:56:35 The EXTRA_OECONF line looks "troublesome" Jan 27 14:56:36 re Jan 27 14:56:44 any thoughts Jan 27 14:56:48 firefox hates me Jan 27 14:56:57 that do_configure will cause chaos as well if libtool is used Jan 27 14:57:12 the recipe dies if it is not there .. Jan 27 14:57:22 i should really push my autotools_base/autotools split Jan 27 14:57:24 lets try and fix one thing at a time :) Jan 27 14:57:44 NOTE: /home/balister/oe/tmp/staging/x86_64-linux/swig/tcl/tclfragments.swg Jan 27 14:57:45 is bad Jan 27 14:57:49 autotools_base overrides LIBTOOL= in EXTRA_OEMAKE and only updates config.sub/config.guess and runs oe_runconf, no autoreconf Jan 27 14:57:58 looks like you require python-native but don't depend on it Jan 27 15:00:05 these are likely true Jan 27 15:00:32 * kergoth ponders Jan 27 15:00:55 The STAGING_*_NATIVE need to be something else though Jan 27 15:01:45 becease the non native recipe is installing stuff in X*^ that I need to install with the dev package Jan 27 15:02:24 ah, yes, swiglibdir looks very wrong Jan 27 15:04:04 normally, I would fix by getting rid of native, but not sure how to do it in the BBCLASSEXTEND env Jan 27 15:04:42 does anyone know of a good writeup on fixing common autotools buildsystem issues, and crosscompile fixes in general? things like wrapping non-sanity-check AC_TRY_RUNs in AC_CACHE_CHECK. Jan 27 15:04:59 Crofton|work: there's an override for each extended version Jan 27 15:05:00 03Graeme Gregory  07org.openembedded.dev * r227751f13a 10openembedded.git/recipes/gnome/libgdata_0.5.1.bb: Jan 27 15:05:00 libgdata_0.5.1.bb : hit a race problem so disable parallel builds. Jan 27 15:05:00 Some required files were created in one thread and used in another and Jan 27 15:05:00 didnt always get created in time. Jan 27 15:05:00 03Graeme Gregory  07org.openembedded.dev * r9d4b913b6d 10openembedded.git/: Merge branch 'org.openembedded.dev' of git+ssh://git.openembedded.org/openembedded into org.openembedded.dev Jan 27 15:05:16 03Graeme Gregory  07org.openembedded.dev * r95c02d5dbe 10openembedded.git/recipes/pango/pangomm_2.26.0.bb: Jan 27 15:05:16 pangomm_2.26.0.bb : update to latest standards. Jan 27 15:05:16 There is no need to override do_configure or for custom staging anymore Jan 27 15:05:25 03Graeme Gregory  07org.openembedded.dev * r6b7b58d2a2 10openembedded.git/recipes/gtkmm/gtkmm_2.18.1.bb: Jan 27 15:05:25 gtkmm_2.18.1.bb : update to latest standards Jan 27 15:05:25 No need to override do_configure anymore. Jan 27 15:05:36 Crofton|work: _virtclass-native Jan 27 15:06:04 is there a write up anywhere? Jan 27 15:06:05 Crofton|work: I'd make a new var for the swiglibdir arg, set it, override it for virtclass-native, and use it in EXTRA_OECONF Jan 27 15:06:11 XorA, blog mentioned that Jan 27 15:06:13 not that i know of, but maybe.. ask RP :) Jan 27 15:11:00 03Koen Kooi  07org.openembedded.dev * rd96527b442 10openembedded.git/classes/gnome.bbclass: gnome bbclass: change PN to BPN in magic SRC_URI composer to make it work in -native and -sdk cases as well Jan 27 15:11:01 03Koen Kooi  07org.openembedded.dev * r3c6e86517e 10openembedded.git/recipes/gtk+/ (7 files in 2 dirs): gtk+ (-native): add 2.18.6 and convert to new-style staging Jan 27 15:11:02 03Koen Kooi  07org.openembedded.dev * r900153c885 10openembedded.git/recipes/gnome/gnome-keyring_2.28.2.bb: Jan 27 15:11:04 gnome-keyring 2.28.2: add dep on gtk+-native Jan 27 15:11:06 * it needs gtk-builder-convert Jan 27 15:11:08 03Koen Kooi  07org.openembedded.dev * r8c103ba2ea 10openembedded.git/recipes/gtk+/gtk+_2.18.3.bb: Jan 27 15:11:11 gtk+ 2.18.3: enable xkb for angstrom Jan 27 15:11:13 * also remove the displaymigration option that has ceased to exist Jan 27 15:12:17 * kergoth thinks about a blog post summarizing the pains of syncing collections which are subsets of OE with OE Jan 27 15:12:26 hmm Jan 27 15:12:30 hello Jan 27 15:12:36 morning Jan 27 15:13:06 I would like to know what package I have to rebuild so that the /etc/modprobe/* files are regenerated (with alias and option files) ? Jan 27 15:13:43 Crofton|work: unfortunately, this being open source, the documentation is the code i.e. the recipes in Poky Jan 27 15:15:12 i think thats a cop out. many open source projects seem to manage to have decent documentation.. we just need to find a more sane workflow for generating and using it in this project Jan 27 15:15:38 :) Jan 27 15:15:41 right now there's no good way to generate docs from recipes and classes.. we need to fix that.. do something like bbdoc, but which can use docs embedded in comments Jan 27 15:15:57 * kergoth shrugs Jan 27 15:16:00 ok making some progress Jan 27 15:16:14 obviously we need to create a doxygen module :) Jan 27 15:16:38 kergoth: you could try to merge "^#" into the above varflags of the last variable? :) Jan 27 15:17:05 something like that could work, but that wouldn't work for documenting the class as a whole, our files don't have flags Jan 27 15:17:05 kergoth it's not a cop out, I agree that we should have it. I'm just telling Crofton that we don't have it (yet) Jan 27 15:17:27 incandescant: what i meant is, using the fact that it's an open source project as an excuse is a cop out. Jan 27 15:17:35 it's not a valid excuse :) Jan 27 15:17:41 many seem to think it is, but.. Jan 27 15:17:48 well, in my opinion, anyway Jan 27 15:18:38 * incandescant agrees Jan 27 15:19:30 * kergoth stops before he starts ranting about the idiocy of the whole "show me the code" attitude... wait, too late. Jan 27 15:19:38 :) Jan 27 15:20:00 better Jan 27 15:20:38 * incandescant goes back to lurking :-p Jan 27 15:21:06 Crofton: I think a doxygen type thing could be very useful indeed. even just seeing a documentation string for each class could be invaluable in discovering what optional classes are available for you to use.. our discoverability sucks ass right now Jan 27 15:21:16 heh Jan 27 15:21:22 yeah Jan 27 15:21:36 I was just hunting around for valid STAGING_ vars Jan 27 15:21:50 I just made two OECONF vars for the different paths Jan 27 15:22:57 Crofton: that's the way RP has handled it in Poky, for e.g opkg Jan 27 15:32:17 and whats the way to get alist of files in an ipk? Jan 27 15:33:35 got i Jan 27 15:33:37 it Jan 27 15:34:20 hum, it looks like they are generated by the kernel image (in populate_packages_prepend) Jan 27 15:34:22 Crofton: dpkg-deb -c file.ipk Jan 27 15:34:40 heh, I am on Fedora :) Jan 27 15:34:52 Crofton: so build dpkg-native? Jan 27 15:35:21 yeah Jan 27 15:38:15 weird :/ Multiple .bb files are due to be built which each provide curl (/home/projects/OE/dev/recipes/curl/curl_7.18.2.bb /home/projects/OE/dev/recipes/curl/curl_7.19.7.bb). Jan 27 15:38:53 ~curse MACHINE_KERNEL_PR Jan 27 15:38:54 May the fleas of a thousand camels infest your most sensitive regions, MACHINE_KERNEL_PR ! Jan 27 15:39:19 JaMa: thats odd.. someone doing an explicit DEPENDS = curl-7.19.7 or something? Jan 27 15:39:30 ~insult MACHINE_KERNEL_PR Jan 27 15:39:32 MACHINE_KERNEL_PR is nothing but an acidic bag of inferior snake assholes Jan 27 15:39:36 nice. Jan 27 15:39:50 not just regular snake assholes, the inferior kind Jan 27 15:41:09 kergoth: haven't seen that yet.. but I'll check.. it's while building task-python-everything.. Jan 27 15:41:24 git grep "DEPENDS.*curl-" Jan 27 15:41:26 perhaps? Jan 27 15:41:33 * kergoth doens't have any other brilliant ideas, sorry to say Jan 27 15:41:58 * incandescant hugs git grep Jan 27 15:42:00 JaMa: use "-g"? Jan 27 15:42:17 * kergoth hugs it too Jan 27 15:42:28 -g's a good idea Jan 27 15:43:06 heh Jan 27 15:43:09 morning kergoth Jan 27 15:44:01 * incandescant is a fan of the depexp ui if you're using bitbake from git "-g -u depexp" Jan 27 15:44:41 hey pb__ Jan 27 15:47:05 anyone have any opinion on the do_rebuild task? In my opinion the way it is now its of limited usefulness, and could actually cause confusion if you attempt to use it.. better to just kill it off to avoid the confusion Jan 27 15:47:16 hrw, kergoth: RDEPENDS = "python-core curl (>=${PV})" Jan 27 15:47:50 in python-pycurl_7.18.2.bb Jan 27 15:47:55 JaMa: that;s not that Jan 27 15:48:05 bitbake do not have versioned depends yet Jan 27 15:48:07 thats runtime, thats fine Jan 27 15:48:21 ah. runtime takes care in opkg level Jan 27 15:48:23 hrw: Missing or unbuildable dependency chain was: ['task-python-everything', 'python-pycurl', 'curl-7.18.2'] Jan 27 15:48:43 hrw: you can do a version specific depends, but its not handled specially, its just another provider. PROVIDES includes ${PN}-${PV} Jan 27 15:49:01 its pretty useless, actually, and can cause issues, since it can easily cause multiple versions of osmething to be built in a single run.. Jan 27 15:49:04 heh Jan 27 15:49:14 kergoth: I know bout pn-pv Jan 27 15:49:16 but rdepends are pulled to deptree for bitbake tree, aren't they? Jan 27 15:49:29 hrw: see above, i think thats what his problem is Jan 27 15:49:34 JaMa: yes, but the version components aren't Jan 27 15:50:07 and last time I tried "curl (>=${PV})" in rdepends it ended in .ipk file like this rdepends = "curl, >=0.17.4" with , between.. Jan 27 15:50:27 weird Jan 27 15:51:06 the kernel recipes does generate the files for /etc/modprobe.d/* where I have the options for the modules, but why are they included in no package ? In what package are they supposed to be ? Jan 27 15:51:57 Longfield: what makes you think they aren't in a package? Jan 27 15:52:02 kernel.bbclass, line 432 Jan 27 15:55:09 kergoth: here is what I tried with RCONFLICTS.. http://www.mail-archive.com/shr-devel@lists.shr-project.org/msg01766.html Jan 27 15:56:15 hmm. Jan 27 15:56:24 * kergoth adds researching that to his list Jan 27 15:56:25 kergoth: yeah I have read this code too. The files are generated and are present in linux-2.6.xx/image/etc/modprobe.d/, but when I list the files in linux-2.6.xx/packages-split, I see these /etc/modprobe.d/* files nowhere. Jan 27 15:56:27 oh no....http://pastebin.com/d4aac782f not libtool....I hope I won't have to spend several days debuging that Jan 27 15:56:51 Longfield: well, that code adds the files to FILES_. Jan 27 15:57:02 which is how files are added to a package. Jan 27 15:57:06 * kergoth shrugs Jan 27 15:58:18 I'm building Angstrom distr, I've added this http://pastebin.ca/1768060 but it seems, that it isn't never run Jan 27 15:58:52 is it correct or how can I debug this? Jan 27 16:00:02 this rootfs_postprocess function should end in run.do_rootfs script, right? Jan 27 16:01:16 GNUtoo: I have patch for that.. Jan 27 16:01:45 JaMa, wow...when will it be commited? does it need to be reviewed? Jan 27 16:02:13 GNUtoo: http://jama.homelinux.org/org.openembedded.shr.images/om-gta02/0001-libsdl-image-remove-all-.m4-macros-from-source-dir.patch Jan 27 16:02:25 thanks Jan 27 16:02:46 JaMa, do you need some review in the ml for that patch,in other word if it works should I report? Jan 27 16:03:09 GNUtoo: I have no idea if it's proper patch.. I only know that it compiles fine.. so please report Jan 27 16:03:22 ok Jan 27 16:03:28 I had libtool issues in the past Jan 27 16:03:30 GNUtoo: similar patch can be used for python-gst.. Jan 27 16:04:04 I don't remember how exactly they were solved Jan 27 16:04:14 what I fear is a messed staging Jan 27 16:06:23 GNUtoo: no.. I had those issues after rebuild from scratch.. it's because those packages have m4 macros for libtool in their sources AFAIK Jan 27 16:06:34 ahh ok Jan 27 16:06:48 thanks so I'll try the patch instead of trying to fix my staging dir Jan 27 16:07:13 transmission was complaining about older libtool used from staging then for .m4 generation and now older packages are complaining that libtool is too new :) Jan 27 16:07:37 "This wave is experiencing some slight turbulence and may explode. If you don't want to explode please reopen the wave. Some recent changes may not be saved." .. awesome.. Jan 27 16:09:25 03Graeme Gregory  07org.openembedded.dev * r57c564dae4 10openembedded.git/: Merge branch 'org.openembedded.dev' of git+ssh://git.openembedded.org/openembedded into org.openembedded.dev Jan 27 16:09:36 03Graeme Gregory  07org.openembedded.dev * r3919422f00 10openembedded.git/recipes/gnome/ (2 files in 2 dirs): Jan 27 16:09:36 gnome-system-tools_2.27.1.bb : remove building the help docs Jan 27 16:09:36 I cant get xml2po to work on these sources so remove doc directory Jan 27 16:09:36 from being built. This made recipe much simpler. Jan 27 16:10:23 JaMa: were possible we want to use m4 macros from OE in preference to ones in source so your probably right Jan 27 16:11:46 XorA: is there some easy way to check which m4 macros should be used from OE (and are available)? reading diff for those macros is not funny.. and usually not so clear which one is "better" to me Jan 27 16:12:26 * kergoth experiments with shoving his random oe ideas list into a google wave Jan 27 16:13:09 XorA: on that python-gst is also "install -m 0644 ${S}/common/m4/*.m4 ${S}/m4/" in preppend and there is A LOT of m4 macros some gst-* are maybe missing in OE staging.. but there are also conflicting lt* Jan 27 16:14:23 btw: any tip for new SRC_URI for http://www.threekong.com/download/dweba/dweba-1.0.0-pre14.tar.gz ? This domain name expired on 01/07/10 and is pending renewal or deletion Jan 27 16:14:47 bye all Jan 27 16:15:43 JaMa: seen witpa? Jan 27 16:17:27 JaMa: they are in staging/usr/lib/aclocal Jan 27 16:19:20 kergoth: what/who is witpa Jan 27 16:19:22 ? Jan 27 16:20:21 its a utility someone wrote to assist with dealing with sources with oe.. it can search for a file on the net with google and verify the md5 matches up for you, to find mirrors Jan 27 16:20:27 figured it might be useful Jan 27 16:20:31 google it Jan 27 16:22:19 JaMa, your patch worked: NOTE: Running task 790 of 1139 (ID: 299, /home/embedded/oe/org.openembedded.dev/recipes/libsdl/libsdl-image_1.2.10.bb, do_install) Jan 27 16:26:20 I'll comit a PN->BPN change Jan 27 16:27:44 kergoth: ah.. thanks.. I tried to google the source manually and only found was on angstrom mirror which I'm not sure is right one for recipe (in case someone will remove it there expecting all needed sources to be redownloaded automatically during world build) Jan 27 16:30:26 ok done Jan 27 16:30:31 03Denis 'Gnutoo' Carikli  07org.openembedded.dev * r269e3d5199 10openembedded.git/recipes/pngcrush/pngcrush_1.6.4.bb: (log message trimmed) Jan 27 16:30:31 pngcrush_1.6.4.bb : use BPN(fixes pngcrush-native install) Jan 27 16:30:31 The problem is that pngcrush-native fails during the installation Jan 27 16:30:31 because of this code in pngcrush_1.6.4.bb: Jan 27 16:30:33 install -m 755 ${PN} ${D}${bindir} Jan 27 16:30:35 that is in the do_install function. Jan 27 16:30:37 This code is inherited by pngcrush-native because Jan 27 16:30:41 03Martin Jansa  07org.openembedded.dev * rfba1e60aaf 10openembedded.git/recipes/libsdl/libsdl-image_1.2.10.bb: libsdl-image: remove all .m4 macros from source dir Jan 27 16:30:52 oops Jan 27 16:30:54 sorryt JaMa Jan 27 16:30:56 I revert? Jan 27 16:31:10 I totally forgeten that I had am-ed this thing in the tree Jan 27 16:31:26 I'm doing too much things at the same time....sorry Jan 27 16:31:31 JaMa, ping what should I do? Jan 27 16:32:24 GNUtoo: no problem.. Jan 27 16:32:34 JaMa, ok should I revert? Jan 27 16:32:43 only the fba1e60aaf obviously Jan 27 16:32:52 imo no.. Jan 27 16:32:55 ok Jan 27 16:33:43 if there is some problem, someone will screem on ML.. and if there is no problem better to fix libsdl-image before someone notices.. Jan 27 16:33:46 what I did was that I did not create a temporary branch because after talking to various irc channels I forgot that I had this commit Jan 27 16:33:56 ok Jan 27 16:34:45 as the commit didn't have my name in it It didn't trigger an alarm in my head Jan 27 17:02:45 have a nice rest of the day Jan 27 17:17:34 I'm still trying to understand why module_conf_* generated files are not installed but not packaged Jan 27 17:17:55 sorry, installed but included in no package Jan 27 17:30:37 hmm, what's a good book to read on UI work for someone with very little experience with no, zero knowledge of usability, user experience, ui design, etc? anyone happen to have any ideas off the top of their heads? (OT) Jan 27 17:31:21 sorry no Jan 27 17:31:31 only usefull ui is qt for me Jan 27 17:31:39 * kergoth googles for some recommendations Jan 27 17:31:55 I've mostly stuck to backend work, but i think its past time to spread out a bit Jan 27 17:31:58 heh Jan 27 17:32:14 hm get familar with qtcreator Jan 27 17:32:22 or qtdesigner Jan 27 17:32:27 yes, Qt is great Jan 27 17:32:39 the output ui's you can even use with pyqt Jan 27 17:32:47 which is really great Jan 27 17:32:52 yeah, qt is pretty nice, i remember messing with it when hacking on opie, and also used qt/e on the land warrior project Jan 27 17:33:00 * kergoth never learned the whole mvc thing with qt4 though Jan 27 17:35:11 well, you're not forced to use MVC, but it saves you from writting a lot of code :) Jan 27 17:36:09 separating implementation from UI is straightforward enough, but never quite wrapped my head around splitting into 3 pieces instead of two :) Jan 27 17:37:12 Can anybody with an account create a page on the OE wiki ? Jan 27 17:37:41 * kergoth wonders if he still has an account :) Jan 27 17:37:47 *g* Jan 27 17:40:37 kergoth: by 3 pieces you mean, model, view and delegates? Jan 27 17:42:56 rkirti: usually, you don't even have to have an account Jan 27 17:58:52 Longfield: add those to FILE_ += "" Jan 27 17:58:56 gm all Jan 27 17:59:26 hey khem Jan 27 17:59:32 hi Jan 27 17:59:40 kergoth: decided on i5 or i7 :) Jan 27 17:59:54 heh, i'll decide once i have the cash to buy it :) Jan 27 18:01:27 microcenter has good deals Jan 27 18:01:40 under 700 for full system Jan 27 18:09:19 I just installed the battery on my beagle Jan 27 18:13:25 for the RTC? Jan 27 18:13:35 yeah Jan 27 18:13:48 It is a nuisance having time warp Jan 27 18:14:35 hi khem Jan 27 18:14:42 I use RO /root so I don't care :) Jan 27 18:16:09 well, but RTC would be handy anyway Jan 27 18:16:59 hi woglinde Jan 27 18:44:26 WTF is meld.mvista.com? Jan 27 18:45:20 MS's own social network? Jan 27 18:45:41 I mean MV's Jan 27 18:46:41 just noticed they are talking about me :-) Jan 27 18:48:08 So they don't censor swearing then :) Jan 27 18:48:18 hehe Jan 27 18:48:50 yeah its the embedded Linux forum dealy MV started Jan 27 18:49:13 just noticed the incoming links to my blog Jan 27 18:49:41 ah Jan 27 18:49:53 oh, i linked a couple coworkers to your post about recipe creation Jan 27 18:49:54 * rkirti tries to recall the user-demands/wishlist page for OE .. Jan 27 18:50:13 kergoth: no bothers, just wondered what it was Jan 27 18:50:17 * kergoth nods Jan 27 18:50:50 kergoth, I am pushing the time stamp patch Jan 27 18:51:04 k Jan 27 18:51:20 * XorA wonders if any OE guys can remeber the userboice url Jan 27 18:51:24 the patch applied from patchwork :) Jan 27 18:51:45 rkirti: well, you can do feature requests with voting in the bts i think, there's also http://openembedded.uservoice.com/, but i always forget it exists Jan 27 18:51:45 heh Jan 27 18:52:24 kergoth with his elephant memory :-) Jan 27 18:52:40 searchign the OE wiki failed Jan 27 18:52:48 oh hell no, that was the result of two googles for customer feedback and feature request sites Jan 27 18:52:50 :) Jan 27 18:53:02 kergoth: thanks Jan 27 18:53:05 np Jan 27 18:53:17 we need to promote the uservoice page more Jan 27 18:53:19 I guess you can say we dont read it :-) Jan 27 18:54:21 its such a nice little site, no complexity, just does what it does well.. i just forget that damn name Jan 27 18:54:24 "uservoice" is not helpful Jan 27 18:54:34 should set up a subdomain that forwards to it if we want to use the thing Jan 27 18:58:04 03Koen Kooi  07org.openembedded.dev * ra902c271e1 10openembedded.git/conf/checksums.ini: checksums: add missing gtk+ checksum, thanks to Graeme for noticing it Jan 27 19:17:53 * kergoth thinks 'XKBTOGGLE' in gtk+ should die in favor of xkb in DISTRO_FEATURES Jan 27 19:21:09 03Rolf Leggewie  07org.openembedded.dev * r96b3f7e82d 10openembedded.git/recipes/gtk+/gtk+-native_2.12.11.bb: gtk+: fix typo Jan 27 19:23:38 hmm, can anyone on wave get to https://wave.google.com/wave/#minimized:nav,minimized:contact,minimized:search,restored:wave:googlewave.com!w%252BN9dP0PCnA ? experimenting :) Jan 27 19:23:50 anybody can explain me if coming to Brussel is really important for improving my openembedded knowledge? I'll see some important setup? Jan 27 19:24:33 03Petr Štetiar  07org.openembedded.dev * r16535dde42 10openembedded.git/recipes/udev/ (udev-141/init udev_141.bb): (log message trimmed) Jan 27 19:24:33 udev: get rid of 'timestamp in the future' tar messages Jan 27 19:24:33 This patch tries to get rid of spew of the useless tar messages durring boot Jan 27 19:24:33 process on machines without RTC, such as the Beagleboard. Example: Jan 27 19:24:33 Please wait: booting... Jan 27 19:24:36 kergoth: and do what ? Jan 27 19:24:37 Starting udev Jan 27 19:24:39 tar: dev/.udev/uevent_seqnum: time stamp 1970-01-01 01:00:21 is 14.122497521 s in the future Jan 27 19:24:42 destroy everything ? :D Jan 27 19:24:58 It is ease to install oe udev to another old distro without it? Jan 27 19:25:09 Crofton: thanks Jan 27 19:25:15 hehe, nothing in particular. just figured I'd start moving my random ideas into one place, and wave is collaborative, so seemed a good possible place to gather it together Jan 27 19:25:20 np Jan 27 19:25:24 and wasn't sure if making it public worked Jan 27 19:25:57 it worked :) Jan 27 19:26:02 works great, but i noticed that really big waves can cause problems Jan 27 19:26:18 really slow everything down, where it shouldnt Jan 27 19:26:23 tomorrow I have to make my boss crazy in order to come to Brussel? Jan 27 19:26:52 kergoth: okay, its neat that i can see as you are typing.. :) Jan 27 19:26:58 :) Jan 27 19:27:29 ok, nobody interested in me .. :( Jan 27 19:27:34 recalcati, mostly you get to meet other OE people Jan 27 19:27:41 and help promote OE to others Jan 27 19:27:53 Crofton: thx for reply Jan 27 19:28:41 I also like CELF in San Francisco and Nuremberg. But I'm working on OE so first I have to come to Brussel. Jan 27 19:28:44 I try. Jan 27 19:29:53 FOSDEM is relatively cheap Jan 27 19:30:05 and it is a good experience Jan 27 19:31:12 Crofton|work: ok. I always decide last days. Relativeky cheap what means? Jan 27 19:31:35 * kgilmer curses icedtea build errors Jan 27 19:33:06 the entrance fee is free to whatever you can afford to donate :) Jan 27 19:33:13 kergoth: but you had a good page earlier. /me link-eyed to it in this post: http://rkirti.wordpress.com/2010/01/28/getting-openembedded-accepted-for-gsoc-2010/ Jan 27 19:33:43 Crofton|work: ok.. instead mobile congress of Barcellona in not absolutely free.... Jan 27 19:33:44 yeah, I'm just looking for one thats easier to maintain.. I've been using omnioutliner, which is nice and all, but not collaborative, and not easy to manipulate from a mobile device Jan 27 19:34:00 ok Jan 27 19:36:54 you are able to edit waves on a mobile device ? Jan 27 19:40:04 rofl Jan 27 19:40:11 mlip: wave crashed safari on my ipod touch Jan 27 19:40:14 that's a neat trick Jan 27 19:40:19 kergoth: argg it cant be opened in konqueror Jan 27 19:40:23 bah Jan 27 19:40:26 stupid google Jan 27 19:40:56 bet it works on android... :) Jan 27 19:41:05 the first time i tried wave it crashed in chromium but worked fine using firefox ;) Jan 27 19:41:25 well, maybe in the android browser, but I cant find a client for it in the market Jan 27 19:41:55 kergoth: trying with FF.... ok it opens up Jan 27 19:42:15 cool.. well, that experiment was not a huge success ;) Jan 27 19:42:18 kergoth: I have wave account Jan 27 19:42:35 I wonder if it can be made accessible to non wavers Jan 27 19:42:43 It asked me to login into it before Jan 27 19:42:59 hmm, don't know Jan 27 19:45:07 kergoth: going thru your laundry list I have one done already Jan 27 19:45:11 Add a nice qemu launch script and qemu machine/distro Jan 27 19:45:12 hehe, nice Jan 27 19:45:35 right now qemu works well for arm mips and x86 on uclibc and glibc Jan 27 19:45:36 * kergoth needs to add some more stuff, moving it over from his main doc which has been collecting crap in a giant disorganized pile for years Jan 27 19:45:41 and I have the script Jan 27 19:45:55 I was wondering what would be best place to commit them to Jan 27 19:45:59 may be contrib Jan 27 19:46:22 the script is simple enough you call run-qemu Jan 27 19:46:40 | arm-angstrom-linux-gnueabi-ld: no machine record defined Jan 27 19:46:42 .. this has to do with a defconfig/.config mistake ? Jan 27 19:46:43 and you might have to adjust few paths to fetch root file system and kernel Jan 27 19:46:58 khem: as to qemu scripts, doesnt poky already have them ? Jan 27 19:47:06 CosmicPenguin: hehe, watching people edit it live is nifty indeed Jan 27 19:47:12 mlip: this is a ld bug I never came to hunt down Jan 27 19:47:13 There we go - now its all collaborative and stuff.. :) Jan 27 19:47:17 mlip: reboot your box Jan 27 19:47:17 hehe Jan 27 19:47:33 khem: so it is related to the host machine ? Jan 27 19:47:33 mlip: it has to do with memory Jan 27 19:47:38 mlip: yes Jan 27 19:47:45 I guess you have ubuntu Jan 27 19:48:00 khem: k, just tried to create a new recipe using -rt patch and use a defconfig (igepv2) from 2.6.28 Jan 27 19:48:04 rkirti: yes poky has them too Jan 27 19:48:10 did not work for me Jan 27 19:48:12 no , actually I use a quite uptodate gentoo Jan 27 19:48:24 mlip: then it could be something common Jan 27 19:48:48 mhm , I'm trying a reboot before i will hunt down anything in defconfi Jan 27 19:49:18 mlip: the error is that linker is not getting machine type Jan 27 19:49:24 its nothing to do with defconfig Jan 27 19:49:38 * khem should debug it one day Jan 27 19:50:09 mlip wait.. Jan 27 19:50:34 khem: i have a little qemu script too, mine calls bitbake to get the paths it needs, you just tell it the image (i.e. console-image) Jan 27 19:50:42 * kergoth hasn't done much with it beyond that, though Jan 27 19:50:53 kergoth: oh nice Jan 27 19:51:01 I can integrate that into mine Jan 27 19:51:07 mine is pure shell stuff Jan 27 19:51:17 groks stuff from where you tell it to Jan 27 19:51:20 thatd be good, lets get something into contrib so we can work on it in a public area Jan 27 19:51:34 kergoth: ok Jan 27 19:51:41 stupid la files Jan 27 19:51:46 I will commit the three needed scripts and a README Jan 27 19:51:49 poor Crofton|work Jan 27 19:51:56 because one needs to make that bridge Jan 27 19:52:00 :) Jan 27 19:52:05 I forgot to delete them before running configure Jan 27 19:52:25 kergoth: had nothing to do with my box ;/ Jan 27 19:52:44 Crofton|work: I'd make a little class, add to INHERIT, make it rm -f them all at the end of do_install or something :) Jan 27 19:53:04 * kergoth thinks mlip2 may have mistabcompleted :) Jan 27 19:53:04 yeah Jan 27 19:53:16 kergoth: like a native script Jan 27 19:53:20 I'll wory about it after I fix the stuff i haven't figured out yet :) Jan 27 19:53:22 native package Jan 27 19:53:40 ~~ Jan 27 19:53:41 i guess ~ is the key Jan 27 19:54:36 kergoth: indeed, sorry Jan 27 19:54:45 khem: -ld fun had nothing to do with my box Jan 27 19:55:03 mlip2: ok Jan 27 19:55:25 mlip2: then I think its defconfig prolly not passing right machine Jan 27 19:55:36 or the linker invocation Jan 27 19:55:43 check the linker commandline Jan 27 19:55:45 make V=1 Jan 27 19:56:11 where? (never used V=1) Jan 27 19:56:24 kernel compile Jan 27 19:56:38 that should print the verbose messages Jan 27 19:56:59 ah , without bitbake? Jan 27 19:59:11 yeah or you can modify the run.do_compile script generated by bitbake and add it there Jan 27 19:59:19 in your workdir Jan 27 19:59:41 just running in tmp doesnt work because some path's are missing Jan 27 19:59:59 as I said try that Jan 27 20:01:59 arm-angstrom-linux-gnueabi-ld: unrecognized option '-Wl,-rpath-link,/home/mlippautz/oe/tmp/staging/armv7a-angstrom-linux-gnueabi/usr/lib' Jan 27 20:05:41 khem ? ;) Jan 27 20:25:16 khem: it is a defconfig issue; investigating the config now Jan 27 20:26:55 03Petri Lehtinen  07org.openembedded.dev * rf2387e3dd5 10openembedded.git/ (3 files in 3 dirs): Jan 27 20:26:55 clish: Bump to 0.7.3 Jan 27 20:26:55 Signed-off-by: Petri Lehtinen Jan 27 20:26:55 Signed-off-by: Rolf Leggewie Jan 27 20:42:12 mlip2: -Wl,-rpath-link is only valid for gcc Jan 27 20:42:26 mlip2: somehow your LDFLAGS seems to be messed up Jan 27 20:42:48 i know, the weird thing is that using an other defconfig fixes also the LDFLAGS Jan 27 20:43:23 I c Jan 27 20:43:37 i just found out that CONFIG_MACH_BEAGLE is available in the general omap kernel tree's, while igepv2 is not Jan 27 20:43:42 and i think this is the problem Jan 27 20:43:46 I think the OE's default LDFLAGS are being passed in Jan 27 20:43:50 for the faulty case Jan 27 20:44:22 something is messed up here Jan 27 20:44:26 :) Jan 27 20:44:29 yes Jan 27 20:44:42 you need to check LDFLAGS set up Jan 27 20:44:53 check in the run.do_compile Jan 27 20:44:56 how it is setup Jan 27 20:45:04 well , but using the exact same checkout and patchset with the other defconfig just works fine Jan 27 20:45:14 thats ok Jan 27 20:45:23 it could be kernel kconfig problem Jan 27 20:45:25 so a wrong defconfig (mb machine setup) messes around with LDFLAGS ? Jan 27 20:45:39 that in one case the LDFLAGS are not overwritten or something Jan 27 20:45:39 weird Jan 27 20:46:09 you can compare ld commandline for both configs Jan 27 20:46:22 and see where the difference is Jan 27 20:46:28 but this doesnt solve my problem with the machine not being available in the kernel ;) Jan 27 20:46:43 i think (unless I patch it in) I have to stick with the 2.6.28 kernel Jan 27 20:46:54 yes Jan 27 20:47:27 §"$(%"%& ... kernel 2.6.28 lacks -rt patches Jan 27 20:48:04 either way work is involved so see what fits the bill :) Jan 27 20:48:30 well, work has to be done till tomorrow 9 am ;) Jan 27 20:51:05 * khem runs off to meeting Jan 27 20:52:05 hello Jan 27 20:58:58 are the /etc/modprobe.d/* files generated by module_conf_* packaged by the kernel class ? If yes in which package should they be in ? Jan 27 21:00:05 Because for me the files are generated and installed, but they do not make it into a package ... and then I have to copy them by hand on my install, which I would like to avoid Jan 27 21:04:48 kergoth: I don't understand bug 5037. Can you take a look to see if it's still valid? Jan 27 21:04:49 http://bugs.openembedded.org/show_bug.cgi?id=5037 Jan 27 21:05:18 * kergoth thinks a bug squashing weekend should get put together again at some point.. do we do those regularly yet, or are they just whenever things? Jan 27 21:08:16 Hello all Jan 27 21:08:49 Laibsch: looks like that situation still exists, as far as i can tell. it shouldn't be closed, it hasn't been fixed, as far as i can see. not an expert on the kernel/module classes anymore, but.. :) Jan 27 21:08:54 heh Jan 27 21:09:00 OK Jan 27 21:09:01 * kergoth shakes head at his grammar Jan 27 21:09:01 thanks Jan 27 21:09:12 Can you leave such a comment and remove the NEEDSINFO status? Jan 27 21:10:07 re the bug squashing weekends. I meant to call them, but I've generally got other things to do and did not want to call one if I can't participate (frankly for fear that there may be nobody else participating) Jan 27 21:10:29 I really don't get why OE devs don't use the tracker more often Jan 27 21:10:43 Yes, I know, it's fixing other people's problems Jan 27 21:13:02 heh, well, i expect many are too busy fixing shit that's broken for them to have time to fix things that are broken for other people, due to our qa issues.. hopefully that'll improve Jan 27 21:13:20 for me, i just haven't acquired the habit of checking the tracker when i have the time to do so Jan 27 21:14:59 Can someone tell me if oe can be configured to build and use a mingw32 target sdk? Jan 27 21:22:44 I would like to use oe for building a cross-platform app. Have used it for Linux but never to target win32. Any ideas? Jan 27 21:54:31 hmm Jan 27 21:55:07 I just did a build of my image and suddenly it didn't install networkmanager. I have no idea why, but the files/packages are missing in the image even though it's listed in the task I depend on Jan 27 21:55:23 I've made a lot of changes but none to networkmanager Jan 27 21:56:03 I'd check the do_rootfs task log, and possibly the testlab info for the image Jan 27 21:56:29 I checked the do_rootfs log it doesn't even mention networkmanager Jan 27 21:56:51 okay, try a bitbake -g foo-image and see if its listed in the dependency graph at all? :) Jan 27 21:58:57 hmm Jan 27 21:59:06 I'll try that if it still fails. I bumped the PR of networkmanager :) Jan 27 21:59:12 and it's building it now Jan 27 22:01:21 actually it fails in postprocess image Jan 27 22:01:34 because I have a postprocess command that modifies one of the NetworkManager config files Jan 27 22:01:51 it just failed again, even though it did build networkmanager Jan 27 22:03:47 "task-xenclient-dom0" -> "networkmanager" [style=dashed] Jan 27 22:03:49 so yeah it's listed Jan 27 22:04:34 hm Jan 27 22:04:38 it's only installed a fraction of the packages Jan 27 22:04:50 the image dir is 27MB it's normally 120 or so Jan 27 22:05:27 03Denys Dmytriyenko  07org.openembedded.dev * re2de494b97 10openembedded.git/classes/base.bbclass: Jan 27 22:05:27 base.bbclass: pre-create SRC_URI checksums to include in the recipe Jan 27 22:05:27 Signed-off-by: Denys Dmytriyenko Jan 27 22:05:27 Acked-by: Khem Raj Jan 27 22:05:27 Acked-by: Frans meulenbroeks Jan 27 22:07:08 ah Jan 27 22:07:26 03Stanislav Brabec  07org.openembedded.dev * r532806deef 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Jan 27 22:07:29 the error is much much further up in the log file :) Jan 27 22:07:36 03Stanislav Brabec  07org.openembedded.dev * r7d685461b7 10openembedded.git/recipes/gnome/ (gnome-doc-utils.inc gnome-doc-utils_0.19.2.bb): gnome-doc-utils: Fix of python path moved to do_install_append, so xml2po calls correct python instance during build. Jan 27 22:07:37 it just stopped because it couldn't find lvm2 package Jan 27 22:08:20 how .. helpful Jan 27 22:08:25 yep Jan 27 22:08:30 it's because I changed the version of lvm2 Jan 27 22:08:32 to an older one Jan 27 22:08:45 but my task package specifies the newer version for some reason :/ Jan 27 22:08:57 I guess it'll work now then :) Jan 27 22:09:20 really dependencies should work with versions Jan 27 22:09:50 and it should use the same dependencies in the packages that it uses in the build system Jan 27 22:09:56 that can easily poison a build, has done for me several times Jan 27 22:13:30 yeah.. hmm, i think i forgot to add that to the reorganized list Jan 27 22:19:02 oh ffs Jan 27 22:19:09 it just refuses to build that image Jan 27 22:23:04 hmm, think i finally got most of the oe bits into the wave... wait, more usability stuff and use cases need to go over Jan 27 22:23:15 oh, and more high level architectural thoughts Jan 27 22:23:17 hmm Jan 27 22:31:53 kergoth, where can I look at that? Jan 27 22:32:24 https://wave.google.com/wave/#minimized:nav,minimized:contact,minimized:search,restored:wave:googlewave.com!w%252BN9dP0PCnA Jan 27 22:32:36 * kergoth isn't sure if there's an anonymous view or anything Jan 27 22:33:41 it's "everyone has access to this wave" Jan 27 22:33:48 basically like a wiki page then :) Jan 27 22:34:48 yeah, pretty much. just experimenting with it Jan 27 22:34:54 * kergoth had it in an omnioutliner doc previously Jan 27 22:35:09 feel free to add ideas if you think of anything Jan 27 22:41:29 kergoth, I won't add it but .. "rewrite the BitBake parser in something faster than python" :) Jan 27 22:42:47 heh, to be fair, the parser is just weak in general, i don't how how much python is to blame, I'd be more inclined to blame the fact that i didn't know python worth a damn when i wrote the original version :P Jan 27 22:43:47 the parser's a good idea for the list though. zecke's made good progress, but there's more to be done Jan 27 22:44:05 there are bigger fish to fry in the performance, than the parse itself, though. Jan 27 22:44:19 like running all those anonymous python functions after each of the parsers Jan 27 22:44:23 s/ers/es/ Jan 27 22:45:08 it's mostly if you changed a config file and it has to reparse everything. that's several minutes of single-threaded evilness Jan 27 22:45:17 when you have build servers with 32 cpu's that becomes very boring :) Jan 27 22:45:37 yeah, thats quite annoying. that's part of why enabling variable reference tracking is in the list Jan 27 22:46:03 if we know what components a variable consists of, and what files contributed to it, it would be easier to avoid unnecessary reparses. that, coupled with zecke's AST, anyway Jan 27 22:47:27 maybe there could be a whitelist of variables that it doesn't care about .. Jan 27 22:47:41 and if only those are changed no reparing Jan 27 22:47:54 maintaining that would be a giant pain in the ass :) Jan 27 22:48:07 today, its not easy at all to tell what variables are actually *used* in a build Jan 27 22:48:18 we need to work on that, ideally we'd be able to track the input and output of all of our tasks.. imo, anyway Jan 27 22:49:09 ah well.. lots of shit to think about doing Jan 27 22:58:45 03Philip Balister  07org.openembedded.dev * rcdfded760f 10openembedded.git/recipes/swig/ (7 files): Jan 27 22:58:45 swig : Fix recipe so it works on target, convert to BBCLASSEXTEND and .inc file. Jan 27 22:58:45 * native packages now use BBCLASSEXTEND method. Jan 27 22:58:45 * Now installs need files into target. Jan 27 22:58:45 * include file to eliminate duplicate portions of the recipe. Jan 27 23:01:24 hehe Jan 27 23:01:33 BBCLASSEXTEND comes more and more Jan 27 23:02:11 * khem is baffled by a gcc warning Jan 27 23:10:32 * ynezz finaly found that exquisite bastard *** ECORE ERROR: Ecore Magic Check Failed!!! Jan 27 23:14:02 * kergoth smacks stanislov for not using pull --rebase :) Jan 27 23:14:40 kergoth: don't forget to smack xora :) Jan 27 23:14:44 heh :) Jan 27 23:20:52 git hooks should just disallow such pushes Jan 27 23:20:53 :) Jan 27 23:21:23 It is not hard to do so. If I have some free time ever I will write one Jan 27 23:23:48 good book http://progit.org/book/ch7-4.html Jan 27 23:29:09 woglinde, I had to touch the recipe, so I fixed several issues Jan 27 23:32:16 crfoton hehe Jan 27 23:43:46 * RP is pleased to see the rise of BBCLASSEXTEND for OE Jan 27 23:44:07 * RP spent a while blaming it for bugs today, only to find it innocent :/ Jan 27 23:44:30 03Rolf Leggewie  07org.openembedded.dev * r35211e52f1 10openembedded.git/recipes/hal/ (files/20hal hal_0.5.14+git.bb hal_0.5.9.1.bb hal_git.bb): hal: if necessary, create $CACHEDIR before starting the daemon. Closes: #3740. Jan 27 23:49:06 kgilmer: still waiting for your comment on http://bugs.openembedded.org/show_bug.cgi?id=4983 Jan 27 23:49:13 what's the status? **** ENDING LOGGING AT Thu Jan 28 02:59:56 2010