**** BEGIN LOGGING AT Wed Dec 30 02:59:57 2009 Dec 30 07:36:00 khem: /topic Dec 30 07:36:03 er Dec 30 07:42:27 03Martin Jansa  07org.openembedded.dev * rb657b3004d 10openembedded.git/recipes/kexecboot/linux-kexecboot-2.6.32+2.6.33-rc2/spitz/defconfig: Dec 30 07:42:35 linux-kexecboot-2.6.32+2.6.33-rc2: update spitz defconfig a bit to support of booting from internal hdd Dec 30 07:42:42 Signed-off-by: Martin Jansa Dec 30 07:43:07 http://pastebin.ca/1731587 shouldn't a native version or qemu be used here? Dec 30 07:59:45 03Khem Raj  07org.openembedded.dev * r6f3c454e90 10openembedded.git/ (146 files in 6 dirs): Dec 30 07:59:48 gcc-svn: Move SRCREV to recipe. Dec 30 07:59:50 * Delete the unused patches Dec 30 07:59:52 * Add --with-system-zlib to configure options. Dec 30 07:59:54 * Add patches needed for uclibc build. Dec 30 07:59:56 Signed-off-by: Khem Raj Dec 30 08:46:28 i'm going to rebuild glibc again, to make it support profiling, anyone knows what's the step to just rebuild glibc? last time i meet some errors and rebuild the whole system :( Dec 30 08:47:49 leslie: clean both glibc and glibc-initial Dec 30 08:48:18 JaMa: oh, ok, thanks ! Dec 30 08:49:04 sometimes I had to rebuild also gcc* recipes because of some missing headers (like stdio.h, limits.h) not sure why but I've seen it only with eglibc.. Dec 30 08:50:14 JaMa: leslie: gcc-initial and (e)glibc share some header files... glibc is overwriting them in the staging area Dec 30 08:51:22 zecke: i have already do the clean Dec 30 08:51:32 zecke: i have to rebuild all again? Dec 30 08:51:53 leslie: not all, glibc, gcc-cross-initial as well Dec 30 08:52:33 zecke: ... ok, i'm scared that moment ... it takes too long time Dec 30 08:52:45 zecke: was it always like this? I haven't seen it few years ago iirc Dec 30 08:52:52 zecke: to rebuild all like xorg ang both qt stuff Dec 30 08:53:34 leslie: only gcc-cross-initial and glibc Dec 30 08:53:45 JaMa: I think it was always the case :} Dec 30 08:54:21 leslie: I use this (but as zecke said its probably not minimal and you are using different versions probably): for i in recipes/gcc/*4.4.2*bb recipes/eglibc/*2.10*bb; do echo $i; bitbake -c clean -b $i; done; bitbake -k shr-image Dec 30 09:00:39 zecke: any idea why this fails? ${@base_version_less_or_equal("${PV}", "2.6.30", "", "CONFIG_INITRAMFS_COMPRESSION_LZMA=y", d)} Dec 30 09:01:19 PV is "2.6.26" Dec 30 09:04:06 ant_work: no idea Dec 30 09:04:20 ant_work: it looks like bbclass code, couldn't you use KERNEL_VERSION instead of PV? but PV should probably work too in all cases Dec 30 09:04:43 no, KERNEL_VERSION is "2.6" Dec 30 09:05:13 ant_work: thats KERNEL_MAJOR_VERSION Dec 30 09:05:28 JaMa: the point is: it's a string Dec 30 09:06:39 ant_work: RRECOMMENDS_kernel-module-hostap-cs += '${@base_version_less_or_equal("KERNEL_VERSION", "2.6.17", "", "apm-wifi-suspendfix", d)}' Dec 30 09:06:51 ant_work: not sure why its not between ${}.. Dec 30 09:06:55 I followed that example but just wont work Dec 30 09:07:06 try grepping in classes Dec 30 09:07:51 yup that was from kernel.bbclass Dec 30 09:08:06 kernel.bclass mangles KERNEL_VERSION Dec 30 09:08:33 but we can rely on PV (in linux-kexecboot.inc) Dec 30 09:08:47 ant_work: but if you read def base_version_less_or_equal, then first parametr is variable name Dec 30 09:08:49 or play with @get_kernelversion Dec 30 09:08:51 ant_work: not variable value Dec 30 09:09:12 a, the pointer you mean Dec 30 09:09:40 def base_less_or_equal(variable, checkvalue, truevalue, falsevalue, d): if float(bb.data.getVar(variable,d,1)) <= float(checkvalue): Dec 30 09:09:54 try it with "PV" instead of "${PV}" :) Dec 30 09:11:34 JaMa: I tried Dec 30 09:11:36 ${@base_version_less_or_equal("KERNEL_VERSION", "2.6.30", "", "CONFIG_INITRAMFS_COMPRESSION_LZMA=y", d)} Dec 30 09:11:57 ant_work: and that KERNEL_VERSION should be available in linux-kexecboot.inc as it inherits kernel.bbclass too through linux.inc Dec 30 09:11:57 and got Dec 30 09:11:59 NOTE: :empty string for float() while evaluating: Dec 30 09:12:39 JaMa: just ned to play with it a bit more ;) Dec 30 09:13:01 ok Dec 30 09:20:55 zecke: it seems workable to clean glibc and glibc-initial, to rebuild glibc Dec 30 09:22:21 zecke: the compilation of bitbake glibc finished, and seems correct Dec 30 09:26:23 03Koen Kooi  07org.openembedded.dev * rd419d5722a 10openembedded.git/contrib/angstrom/build-feeds.sh: angstrom feed builder: add geany Dec 30 09:26:34 03Koen Kooi  07org.openembedded.dev * r0462d99080 10openembedded.git/contrib/angstrom/sort.sh: angstrom feed sorter: add archos5 machine Dec 30 09:26:40 03Koen Kooi  07org.openembedded.dev * r84c277962c 10openembedded.git/recipes/gcc/gcc-package-target.inc: Dec 30 09:26:43 gcc-package-target: set INSANE_SKIP is true on the objc package to mask missing GNU_HASH Dec 30 09:26:49 * it was suggested on the oe-devel mailinglist that it doesn't matter for libobjc Dec 30 09:28:16 hi, Dec 30 09:28:25 mmm abiword also use doltlibtool Dec 30 09:30:35 we should find a better solution than -DOLT on each configure.in file Dec 30 09:34:00 leslie_: great Dec 30 09:43:46 03Frans Meulenbroeks  07org.openembedded.dev * r65a4b38bbc 10openembedded.git/recipes/perl/libhtml-treebuilder-perl_3.23.bb: libhtml-treebuilder-perl: updated dependency Dec 30 09:43:56 03Frans Meulenbroeks  07org.openembedded.dev * rbd3d9db65d 10openembedded.git/recipes/perl/perl_5.8.8.bb: Dec 30 09:44:02 perl: bumped PR to force rebuild; libextutils-makemaker-perl can have overwritten MM_Unix.pm in staging Dec 30 09:44:08 (see commit c595d241fb4d23c714dc61a1c9d281b9bbb133cd) a rebuild fixes this. Dec 30 09:44:10 03Frans Meulenbroeks  07org.openembedded.dev * rc595d241fb 10openembedded.git/recipes/perl/ (2 files): Dec 30 09:44:13 libextutils-makemaker-perl: removed; with the new staging it breaks perl by overwriting MM_Unix.pm Dec 30 09:44:17 It is not needed anyway as makemaker is already created when perl is build, so remove Dec 30 10:08:55 anybody knows how to change OE_QMAKE_CFLAGS in oe? Dec 30 10:10:22 leslie_: like any other variable as well. :) Dec 30 10:10:33 leslie_: git grep in classes/ for their initial values Dec 30 10:12:18 zecke: thanks a lot~~ :-) Dec 30 10:52:04 for staging a binary should I append to the staging process? Dec 30 10:53:39 (I want to stage libwmf-config) Dec 30 10:57:51 like do_stage_append(){ install ${bindir}/libwmf-config ${STAGING_BINDIR} Dec 30 10:57:59 something like that Dec 30 10:58:06 with the good install arguments Dec 30 10:58:28 GNUtoo|oeee: that is old skool staging Dec 30 10:58:36 lol ok Dec 30 10:58:37 something like that would work, though with the new staging system I think you should be able to just delete do_stage() altogether. Dec 30 10:58:41 GNUtoo|oeee: you would have to check the mail archives for the new skool Dec 30 10:58:51 ok lol thanks Dec 30 10:59:03 I thought the new school was kind of automatic Dec 30 10:59:15 I'll re-read the mails about that Dec 30 10:59:23 yes, it is. you just need a small photo of RP stuck to your monitor, and everything should be done for you. Dec 30 11:38:57 zecke: what's the status with the oestats waterfall at the moment? Dec 30 11:40:35 pb_: I suck. Laibsch attempted to build a package, the installed server had "issues" he was not very specific and I planned to create a virtualbox with the distro to test it myself Dec 30 11:41:15 righto Dec 30 11:41:50 pb_: maybe I manage to do it next year Dec 30 11:44:33 fingers crossed Dec 30 11:45:07 I am still happy to install oestats on ltg if it is too difficult/annoying to get it working on melo. Dec 30 11:45:58 brb... need to pull the power plage from the taiwanese plastique router... Dec 30 11:46:05 heh Dec 30 11:48:28 pb_: it can't be that difficult(tm) Dec 30 11:48:53 right, just a case of pressing the correct keys on your computer Dec 30 12:07:19 pb_: okay, I'm installing a vm now Dec 30 12:10:26 excellent Dec 30 12:10:30 go zecke! Dec 30 12:49:10 03Koen Kooi  07org.openembedded.dev * r234d22a2e9 10openembedded.git/ (conf/checksums.ini recipes/bluez/bluez4_4.59.bb): bluez4: add 4.59 Dec 30 12:56:23 mmm in the porting guide there is no simple way to add a binary Dec 30 12:56:48 binaty means ${bindir}/libwmf-config Dec 30 12:57:06 what porting guide are you referring to? Dec 30 12:57:20 oh, you mean the new-staging porting guide? Dec 30 12:57:22 [oe] Staging - time to end the current mess - now with patches Dec 30 12:57:26 yes Dec 30 12:58:02 I don't think you should need to do anything special to add a binary. It ought to "just work", as far as I know. Dec 30 12:58:19 in my case it doesn't Dec 30 12:58:59 sysroot-destdir only has the libs and the headers Dec 30 12:59:04 not that binary Dec 30 12:59:11 is your "make install" installing the binary into the right place? Dec 30 12:59:19 yes Dec 30 13:00:03 because the binary is in package and in image Dec 30 13:00:21 and I realy need that binary for abiword...else CROSS_COMPILE BADNESS Dec 30 13:00:22 er Dec 30 13:00:29 is this a target binary you are trying to stage? Dec 30 13:00:35 normally you would only stage native binaries. Dec 30 13:00:47 ouch.... Dec 30 13:00:51 yes Dec 30 13:01:00 mmm Dec 30 13:01:21 I don't think you will get any useful effect if you install a target binary into staging. There is no infrastructure in oe that would allow you to run it. Dec 30 13:01:24 "cross"-compiling for x86 target makes me do some confusion Dec 30 13:01:33 indeed Dec 30 13:01:35 sorry Dec 30 13:01:52 so I should create a native package which stage the binary or something like that? Dec 30 13:01:56 right Dec 30 13:02:15 here's what that binary does: Dec 30 13:03:22 if you make a native package, the staging infrastructure should stage the binary automatically. it's inhibited for target packages. Dec 30 13:03:49 libwmf-config --cflags => -I/usr/include/freetype2 -I/usr/include Dec 30 13:04:13 if that's all it does, it might be worth investigating whether you can throw it away and use pkg-config instead. Dec 30 13:04:35 ok thanks Dec 30 13:06:36 no pc$ file Dec 30 13:07:18 I fear that if I compile a native version it would give native flags Dec 30 13:13:26 maybe I found a better solution...I use prefix: Dec 30 13:13:48 libwmf-config --prefix=/home/embedded/oetmp/staging/i686-angstrom-linux --libs => -L/home/embedded/oetmp/staging/i686-angstrom-linux/lib -lwmf -lwmflite -lfreetype -lz -lX11 -lexpat -ljpeg -lpng -lz -lm Dec 30 13:13:50 etc... Dec 30 13:14:10 I'll make this native recipe Dec 30 13:16:30 maybe that is even better: Dec 30 13:16:31 --with-libwmf-config Dec 30 13:17:38 but we still should find a fix for dotlibtool Dec 30 13:23:10 maybe I should write to the ml about that Dec 30 13:41:33 03Martin Jansa  07org.openembedded.dev * rcc4ed9b2bc 10openembedded.git/conf/machine/om-gta02.conf: Dec 30 13:41:35 om-gta02.conf: set MACHINE_DRI_MODULES = glamo only for shr as it needs mesa-dri_git.bb Dec 30 13:41:37 Signed-off-by: Martin Jansa Dec 30 14:24:26 03Marcin Juszkiewicz  07org.openembedded.dev * r12cb70dff8 10openembedded.git/recipes/linux/linux_2.6.31.bb: linux 2.6.31: updated to .9 Dec 30 14:24:28 03Marcin Juszkiewicz  07org.openembedded.dev * rbc6231c101 10openembedded.git/recipes/linux/linux_2.6.27.bb: linux 2.6.27: updated to .42 Dec 30 14:24:34 03Marcin Juszkiewicz  07org.openembedded.dev * r661b3a9860 10openembedded.git/recipes/linux/linux_2.6.32.bb: linux 2.6.32: updated to .2 Dec 30 14:24:42 03Marcin Juszkiewicz  07org.openembedded.dev * r014b5c206a 10openembedded.git/conf/checksums.ini: checksums.ini: added checksums for kernel patches Dec 30 14:27:16 pb_: short update: I do have hardy installed, I have a django stat package too Dec 30 14:38:12 hello Dec 30 14:38:29 does anybody use .debs for their distro here? Dec 30 14:39:06 a problem I am running into is that the package index is getting rebuilt too frequently Dec 30 14:39:40 there seems to be provisions for a "DEB_PACKAGE_INDEX_CLEAN" stamp, which indicates when the package cache is clean Dec 30 14:40:10 .. but the presence of this file is not tested before regenerating the package index (unlike in package_ipkg.class) Dec 30 14:44:55 zecke: very good Dec 30 14:48:55 morning Dec 30 15:03:37 mickey|sofa: good afternoon Dec 30 15:03:48 hey pb_ Dec 30 15:08:31 hey mickeyl Dec 30 15:08:34 hey mickey|sofa Dec 30 15:09:11 hey XorA Dec 30 15:09:17 got any gadgets for xmas? :) Dec 30 15:09:30 mickey|sofa: no, Im safely gadget free :-) Dec 30 15:09:35 hehe Dec 30 15:09:36 good for you Dec 30 15:09:48 actually not 100% true, I bought a r3600 nvidia Ion based device Dec 30 15:10:27 sounds powerful Dec 30 15:10:30 but... what is that? Dec 30 15:10:43 its an intel atom/nvidia geforce combo Dec 30 15:10:44 pb_: I need to ping Laibsch to get his django packet Dec 30 15:10:54 so low power for x86 but will play 1080p video Dec 30 15:10:55 cool. so a netbook? Dec 30 15:11:08 like the EEE boxes, but with some grunt int he gfx dept Dec 30 15:11:09 zecke: ah, isn't it the standard one from ubuntu? Dec 30 15:11:20 but, anyway, do what you need to do :-) Dec 30 15:11:58 just need to wait for the Win7 dvd to arrive now Dec 30 15:14:13 hmm, windows? Dec 30 15:14:16 tsss Dec 30 15:14:35 * mickey|sofa waits for the next MBP 13" update Dec 30 15:16:08 mickeyl: turns out linux VDPDAU support sucks :-) Dec 30 15:16:48 JaMa: ping Dec 30 15:16:59 eFfeM: pong Dec 30 15:17:43 jama, got your email, i'm not really a perl wiz either but most of the things you forwarded are new recipes and so pretty harmless. Dec 30 15:18:19 i see no problem committing them even though my perl is not good enough to test it (i mainly needed perl for some mythweb related functionality and kinda "made it work" Dec 30 15:18:44 only remark i had was that the perl modules typically reside in recipes/perl and do not get their own dir Dec 30 15:19:55 and probably i would prefer committing on a per module basis (as that allows for easier rollback) (and I think checksums are missing too) Dec 30 15:20:18 do you feel making those changes that way is a good way forward Dec 30 15:20:20 ? Dec 30 15:20:33 eFfeM: ok.. I'll ask them for move and proper credits in it.. because I'm not sure as there is different combination of author/submiter a bit strange.. Dec 30 15:21:02 i can commit on their behalf, that is what often happens with new developers Dec 30 15:21:18 http://wiki.openembedded.net/index.php/GitPhraseBook#Commiting_someones_else_work Dec 30 15:21:39 but it does expose their email address in the git log Dec 30 15:21:52 eFfeM: yes.. but there is 3 different "realnames" for that work (1 author I guess) so not sure which one to use.. Dec 30 15:22:14 ah ok, then better ask them and either send to patchwork or the mailing list Dec 30 15:22:30 parilo, chuck norris and anton Dec 30 15:22:36 preferably not in one big chunk but per module Dec 30 15:23:02 oki Dec 30 15:23:25 (btw thought they were only from parilo, didn't see the other names) Dec 30 15:26:42 eFfeM: other names are in patch body and in submitter in 2nd patch.. Dec 30 15:27:00 ah ok, didn't review in that much detail Dec 30 16:05:06 03Jeremy Lainé  07org.openembedded.dev * r280420fa7a 10openembedded.git/recipes/linux/linux-2.6.31/qemux86/defconfig: linux-2.6.31: enable 8139cp for networking in qemux86 Dec 30 16:05:46 hi Dec 30 16:07:08 anybody notice that iptables is broken? Dec 30 16:07:37 the /usr/libexec/xtables/*.so files are no longer shipped in "iptables", for some reason they have been moved to "iptables-dbg" Dec 30 16:07:53 the corresponding commit is:http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=da29e9aaa051f40665f818fe5884725047288a2f Dec 30 16:09:26 jeremy_laine hm ah typical plugin problem Dec 30 16:10:13 that commit was made while asleep at the wheel from looks of things Dec 30 16:10:33 yep, looks like the line: FILES_${PN}-dbg += "${libexecdir}/xtables" Dec 30 16:10:42 should be: FILES_${PN}-dbg += "${libexecdir}/xtables/.debug" Dec 30 16:10:46 jeremy_laine: just fix that like, bump PR and commit Dec 30 16:11:33 XorA: I was wondering why I could not longer get shell access to my devices :) Dec 30 16:11:49 jeremy_laine: hehe Dec 30 16:12:00 XorA: so I had to rebuild my image for qemux86 to debug Dec 30 16:12:29 XorA: which is when I realised networking was broken for qemux86.. long turnaround time for a fix! Dec 30 16:17:35 bye Dec 30 16:17:40 * XorA curses thunderbird for not having an account export option Dec 30 16:21:44 jeremy_laine: I have fixed qemumips havent got around to qemux86 Dec 30 16:22:04 woglinde, hi, do you have an openoffice recipe? Dec 30 16:22:36 hmmm Dec 30 16:23:03 gnutoo at work Dec 30 16:23:08 XorA: are you switching your mail client ? Dec 30 16:23:17 hi khem Dec 30 16:23:23 gnutoo next week Dec 30 16:23:24 woglinde: hello.. Dec 30 16:23:40 khem: considering it, but I cant be arsed switching accounts on all my machines Dec 30 16:25:14 XorA: I usually use mutt with IMAP works well for multiple machines Dec 30 16:26:03 ok Dec 30 16:26:07 khem: mutt keeps segfaulting/locking up on me when network connection dies and so is annoying me :-) Dec 30 16:26:38 btw about mutt...what's the most usable mail client in oe? Dec 30 16:26:51 GNUtoo: mutt :-) Dec 30 16:26:55 claws has an issue with screen size with 800x480 when composing mails Dec 30 16:26:56 ok Dec 30 16:27:05 and gnus is too dificult to setup Dec 30 16:27:07 GNUtoo: claws has a small screen mode Dec 30 16:27:14 that's what I used Dec 30 16:27:18 the rest is fine Dec 30 16:27:37 it's only the compose winows that go beyong the screen size Dec 30 16:28:19 odd, must be new breakage, I used to use it Dec 30 16:29:30 ok Dec 30 16:29:55 basically it is usable if you put a lot of breaklines...at the end of the mail you send Dec 30 16:30:01 XorA: hmm I never faced this situation thus far most of the time I am connected to network Dec 30 16:30:04 but that's not good Dec 30 16:30:13 XorA: may be use pop or disconnected IMAP Dec 30 16:31:00 khem: yeah, will have to figure out running offlineimap on windows Dec 30 16:33:16 XorA: or use Andrew File system for storing mails :) Dec 30 16:33:35 afs always looks like a disaster waiting to happen to me Dec 30 16:33:36 then you can mount it from other machines and it will sync Dec 30 16:34:29 03Jeremy Lainé  07org.openembedded.dev * rd49b36e8cd 10openembedded.git/recipes/iptables/iptables.inc: iptables: fix packaging of ${libexec}/xtables modules Dec 30 16:36:43 XorA: windows means thunderbird is your best bet Dec 30 16:37:14 and thunderbird 3 has that really fucking annoying auto conf stuff that always gets it wrong, but doesnt seem to have a disable :-( Dec 30 16:38:42 * khem uses IMAP with apple mail,kmail and mutt and is happy thus far Dec 30 16:39:13 I just object to typing the same info 7-8 times for 7-8 accounts Dec 30 16:40:25 you could copy .thunderbird dirs around :) Dec 30 16:42:35 morning kergoth Dec 30 16:44:57 hey Dec 30 16:46:13 hi pb Dec 30 16:46:17 hi kergoth Dec 30 16:46:19 hi ph5 Dec 30 16:46:23 hi woglinde Dec 30 16:46:31 hi all Dec 30 16:46:54 jeremy_laine: thx for fixing the iptables thing which I broke :) Dec 30 16:59:06 * pb__ waits patiently for lvm to migrate his data Dec 30 17:00:16 khem: you must promise to drink more coffee before commits :-) Dec 30 17:00:55 actually heh yes Dec 30 17:04:28 XorA: heh yes Dec 30 17:28:26 Does anyone else get this error? Dec 30 17:28:36 Requested 'x11 >= 1.1.99.1' but version of X11 is 1.1.5 Dec 30 17:28:42 With libxext-1.1.1 Dec 30 17:28:56 B_Lizzard which distribution? Dec 30 17:29:13 jlime, I know, it's not Angstrom Dec 30 17:29:13 problem is the different xorg packages now in oe Dec 30 17:29:27 use the sane-xorg reivison files Dec 30 17:29:35 for distribution Dec 30 17:29:46 Clean build with clean openembedded workdir Dec 30 17:30:05 I tried with and without preferred-xorg-versions-X11R7.5.inc Dec 30 17:30:47 yeah preferred-xorg-versions-X11R7.5.inc is a bit broken at the moment Dec 30 17:30:49 minimal should also work Dec 30 17:30:57 se 7.4 and 7.4 updates Dec 30 17:30:59 If I "fix" configure.ca to accept 1.1.5, it fails at do_configure Dec 30 17:31:00 it used 7.4 Dec 30 17:31:19 I see that Angstrom doesn't have that preferred-version thing Dec 30 17:31:20 no Dec 30 17:31:24 dont fix the configures Dec 30 17:31:30 Yeah, I know Dec 30 17:31:35 (I meant do_compile) Dec 30 17:31:49 hm I thought one commit fixed the xorg7.5 Dec 30 17:31:57 seems I need to test it again Dec 30 17:32:07 B_Lizzard: mix and match will get you a dalmatian Dec 30 17:32:21 I know :) Dec 30 17:32:34 Thing is, it compiled for me a couple of days ago. Dec 30 17:32:49 hm Dec 30 17:32:52 yes and lot of updates from martin were pushed recently Dec 30 17:32:54 look at the git-log Dec 30 17:33:01 the openembedded workdir must've been a week old from now. Dec 30 17:33:17 B_Lizzard: week is too old for .dev Dec 30 17:33:32 Yeah, I know. Dec 30 17:33:38 At least it narrows it down. Dec 30 17:33:57 The error seems pretty clear-cut, and that's what worries me. Dec 30 17:37:03 hms Dec 30 17:37:21 libudev-dev is still broken Dec 30 17:37:30 why noboy fixing such lowhanging fruits Dec 30 17:38:00 the following files were installed but not shipped in any package: Dec 30 17:38:04 NOTE: /usr/lib/libudev.so Dec 30 17:38:07 NOTE: /usr/lib/libvolume_id.so Dec 30 17:38:10 NOTE: /usr/lib/pkgconfig/libudev.pc Dec 30 17:38:14 NOTE: /usr/lib/pkgconfig/libvolume_id.pc Dec 30 17:39:21 It must be one of Martin's commits 8 days ago. Dec 30 17:42:15 It must be this one http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=fba77bf1ca0257e3e795b9c514b3420fc9cf5fed Dec 30 17:42:34 He added newer xext versions Dec 30 17:42:57 Which get picked up no matter if or if you don't have preferred-xorg-versions-X11R7.5.inc Dec 30 17:43:24 I'm guessing he forgot to add corresponding libx11 and diet-x11 versions. Dec 30 17:44:13 Or, at least dietx11 Dec 30 17:44:19 As that what jlime uses. Dec 30 17:44:34 b_lizzard do you can make a patch? Dec 30 17:44:53 Sure, if if the recipes don't change much Dec 30 17:44:56 Lemme see. Dec 30 17:45:16 I'm not sure why we use diet-x11, it sucks. Dec 30 17:45:20 But ehm Dec 30 17:46:37 b_lizzard space? Dec 30 17:47:10 Well, the devices we target aren't as space-challenged as they are memory-challenged Dec 30 17:47:24 But libx11 proper isn't such a hog Dec 30 17:47:53 And damn me to hell if I can't have xkb on a machine with a proper laptop-like keyboard (Jornada) Dec 30 17:47:56 :) Dec 30 18:11:13 Generally, is it good to require preferred-xorg-versions-X11R7.5.inc? Dec 30 18:11:23 You said, for instance, that it is buggy. Dec 30 18:11:37 hm seems jama fixed all Dec 30 18:11:44 so its a good idea Dec 30 18:11:46 ...and I see that Angstrom doesn't use that either. Dec 30 18:11:57 and it seems dietx is the fault Dec 30 18:12:03 Yeah, I know. Dec 30 18:12:11 But what about Angstrom not using the file? Dec 30 18:12:15 b_lizzard the "not invented here syndrom" for angstroem Dec 30 18:12:21 I guess Dec 30 18:12:21 I see ;) Dec 30 18:12:47 Oh well. Dec 30 18:16:29 B_Lizzard: not nihs imho.. but angstrom prefer always latest version available.. Dec 30 18:16:40 B_Lizzard: and that is easier without any include Dec 30 18:16:56 git-native suckz with 4 gb of compile space Dec 30 18:17:49 B_Lizzard: I did the same in shr.conf, because I wanted pixman_git.bb and that is difficult to prefer if you have PREFERRED_VERSION_pn-pixman from some include then DEFAULT_PREFERRENCE wont help you.. Dec 30 18:18:17 I see. Dec 30 18:18:35 Well, unless it proves troublesome, I'll just require it. Dec 30 18:19:17 fuck Dec 30 18:19:27 git-native with angstroem over 7 gig tmp sapce Dec 30 18:19:57 woglinde: ggdb3 :) Dec 30 18:20:05 khem yeah Dec 30 18:20:16 I hate it every time Dec 30 18:20:46 I dont understand -O2 and -ggdb3 Dec 30 18:21:06 two conglicting things as far as gcc is concerned Dec 30 18:21:30 khem and gcc-4.4.2 Dec 30 18:21:40 mabyee 4.2 is smaller Dec 30 18:21:54 gcc 4.4.2 generates rich debug info Dec 30 18:22:01 gcc 4.2 is lame Dec 30 18:22:39 but the copying of over 7 gig suckz Dec 30 18:22:45 agrred Dec 30 18:23:34 I think gcc generates the info but how useful this will be is a question that remains to be seen Dec 30 18:25:45 something is pulling in external-toolchain-csl.bb hmmm Dec 30 18:26:11 khem which distro? Dec 30 18:26:16 which image Dec 30 18:26:21 but I can bet which Dec 30 18:26:23 woglinde: its my own crap Dec 30 18:26:29 There's nothing particularly contradictory about having both -O2 and -ggdb3; it's legitimate to want to both optimize your program and debug it. I'm not very sure that -ggdb3 is worthwhile in terms of space though, compared to -ggdb. Dec 30 18:26:44 hm for native? Dec 30 18:26:54 its crazy koen Dec 30 18:27:02 pb__: yes true however I wonder how useful it is Dec 30 18:27:10 as of now Dec 30 18:27:20 yeah, personally I would stick at -O2 -ggdb. Dec 30 18:27:31 in an ideal compiler it should generate best debug info for best optimized code Dec 30 18:29:46 indeed, and gcc is actually pretty close to ideal in that respect: -g and -O are pretty much independent. Dec 30 18:30:10 obviously the more optimisation you have, the harder it gets to debug, but adding -g shouldn't cause you to get worse code. Dec 30 18:30:13 pb__: there are certain opts which are harsh on debugging and some not so we could hatch out an option set which has opts yet is suitable for debugging Dec 30 18:30:48 pb__: no gcc loses a lot of information with many opt phases Dec 30 18:31:09 that it can not generate correct debug info Dec 30 18:32:19 anyway, each distro can choose the set of cflags that works best for them. if angstrom wants to use -ggdb3 then they are, obviously, welcome to do so. Dec 30 18:33:34 * khem nods Dec 30 18:35:17 woglinde: I had inadvertently changes += for a PREFERREED_PRVIDED to = in my local conf Dec 30 18:35:36 and that caused the stuff to go awry Dec 30 18:35:53 khem abiword is problem for uclibc Dec 30 18:36:15 03Koen Kooi  07org.openembedded.dev * r3e891e35d2 10openembedded.git/recipes/webkit/ (3 files in 2 dirs): webkit-gtk: bump to 1.1.18+svnr52207 Dec 30 18:36:26 03Koen Kooi  07org.openembedded.dev * r282fd53120 10openembedded.git/ (conf/checksums.ini recipes/gtk-webcore/midori_0.2.2.bb): midori: add 0.2.2 Dec 30 18:36:30 03Koen Kooi  07org.openembedded.dev * rf2d165f2fd 10openembedded.git/ (3 files in 2 dirs): libsoup: add 2.28.2 and 2.29.3 Dec 30 18:36:50 woglinde: hmmm what happens ? Dec 30 18:37:10 look at the recipe Dec 30 18:37:41 hm ah not abiword Dec 30 18:37:46 was the office programm Dec 30 18:38:00 pb__: I am getting gcc 4.5/svn to work well then I would like to test LTO in OE Dec 30 18:38:21 yeah, that would be interesting Dec 30 18:38:39 woglinde: qemu is such a nice thing Dec 30 18:38:42 I'm not sure that gold works well enough yet to do that on arm, but it would be worth a try for x86. Dec 30 18:39:13 pb__: yeah gold is good on x86 as well as x86_64 Dec 30 18:39:13 khem hm some of the gnome office suite drags in libc locales packages Dec 30 18:39:23 yuck Dec 30 18:39:51 woglinde: doesnt it have some knobs to say I dont want locales Dec 30 18:40:12 having said that office might be a little too much for uclibc :) Dec 30 18:40:43 I would stop at firefox beyond that its too big a ask for such a small library Dec 30 18:41:14 wahahaha I thought the strip git-native bug was fixed Dec 30 18:41:17 woglinde, ah you're fixing abiword too? Dec 30 18:41:34 gnutoo no Dec 30 18:41:47 ah ok Dec 30 18:44:58 btw does someone has a clue about fixing all theses dolt thing Dec 30 18:45:04 or should I write to the ml? Dec 30 18:45:15 gnutoo hm which package it was? Dec 30 18:45:23 GNUtoo: whats wrong with it ? Dec 30 18:45:58 oh you mentioned yesterday .. Dec 30 18:46:07 khem, it fallbacks on "libtool" instead of "i686-angstrom-libtool" Dec 30 18:46:09 re kgilmer Dec 30 18:46:22 gnutoo hm Dec 30 18:46:31 I looked at the m4 code Dec 30 18:46:37 shouldnt be that hard to fix Dec 30 18:47:03 GNUtoo: that would be dolt not knowing/caring cross builds Dec 30 18:47:13 ok Dec 30 18:47:34 but I bet they include the m4 in the source of the package Dec 30 18:47:35 no? Dec 30 18:48:19 khem: as you're talking about low level stuff, I noticed info about "LIBM_BIG Option Group" in recent eglibc... what do you think about that? Dec 30 18:49:19 ./abiword-2.8.1/dolt.m4 Dec 30 18:50:21 JaMa: I have local patches to make eglibc configurable here Dec 30 18:51:40 JaMa: I think its useful for many embedded systems who do not care so much about floating point precision Dec 30 18:51:45 I haven't checked how many apps are using double precision where single precision would be enough.. but I guess that it could be also a bit faster on arm? not only because of smaller libc.. Dec 30 18:52:42 yes I think so Dec 30 18:53:30 but I guess its not so different as using integer based math instead of double (ie when using tremor instead of libvorbis for ogg decoding) Dec 30 18:54:06 I think it will be somewhere in middle Dec 30 18:56:04 woglinde: isnt dolt added to libtool itselg Dec 30 18:59:07 no it replaces it Dec 30 19:00:54 I think if added to libtool there should be no need to add it to packages ideally Dec 30 19:02:33 basically -DOLT fixes it but if too much recipes start using it (I identified 3 but they are hard to identify) patching them is not very productive Dec 30 19:04:11 and they ship dolt.m4 inside the sources of each packages Dec 30 19:07:00 re Dec 30 19:19:11 re Dec 30 19:19:33 hm opkg needs a fix for micro too Dec 30 19:26:02 what's wrong with it? Dec 30 19:26:35 micro doesn't use O_P_M so opkg is largely irrelevant; it only needs to work for image construction Dec 30 19:28:31 hm oh? Dec 30 19:28:52 ipkg is used? Dec 30 19:29:15 or shouldnt you be able to install stuff? Dec 30 19:34:12 there's no support for installing new packages on a running system. Dec 30 19:34:25 okay Dec 30 19:34:43 wondered why opkg got installed Dec 30 19:34:49 it shouldn't. Dec 30 19:34:53 maybee my fault for the image Dec 30 19:34:59 if opkg is landing in your rootfs then that would be a bug somewhere. Dec 30 19:35:34 a recipe_git.bb defines PV=x.y+svnr${SRCPV} now how can I choose it in PREFERRED_VERSION for that package ? Dec 30 19:35:41 in my local.conf Dec 30 19:36:57 re Dec 30 19:37:33 patch it to stop using SRCPV and then set P_V appropriately :-} Dec 30 19:37:43 or, alternatively, if you can predict what SRCPV will come out as, you are all set. Dec 30 19:38:16 jo hrw Dec 30 19:38:17 yeah I will hardcode it Dec 30 19:46:10 dman why bitbake not respect my BB_ENV_EXTRAWHITE = "CCACHE_DIR" Dec 30 20:32:07 hmmm I can not call mpc for http://www.multiprecision.org/ Dec 30 20:32:23 because we already have a recipe for musicpd/mpc Dec 30 20:32:34 * khem renames to libmpc Dec 30 20:33:23 btw. we have mpfr and gmp recipes I think they could also be renamed to libmpfr and libgmp Dec 30 20:33:57 they are libraries only so prefixing lib is not such a bad idea Dec 30 20:34:40 maybe, but the tarball is named "mpfr-x.x.x.tar.gz" and the project calls itself MPFR. Dec 30 20:35:40 debian also names the source package "mpfr", not libmpfr, though I guess that is not a very strong argument. Dec 30 20:36:25 mpc is also named mpc-x.x.tar Dec 30 20:36:46 but there is naming clash musicpd/mpc existed before Dec 30 20:37:56 pb__: opensuse and mandriva prefix them with lib :) Dec 30 20:38:07 for now Dec 30 20:38:14 I will leave them as it is Dec 30 20:38:20 only change mpc Dec 30 20:38:23 to libmpc Dec 30 20:38:54 that sounds best. renaming an existing package is a fairly disruptive thing to do, I don't think it would be a good idea to do that just because you can. Dec 30 20:39:09 yep agreed Dec 30 20:39:15 obviously that doesn't apply to new packages, so renaming mpc is fine Dec 30 20:39:49 right I did not hit this until I built target gcc Dec 30 20:40:08 because cross gcc just depended on mpc-native and that name was unique Dec 30 20:40:37 Fedora's naming is like OE, they have mpfr gmp and libmpc Dec 30 21:02:16 mckoan|away: ping Dec 30 21:19:32 re Dec 30 21:20:05 looks like Guo Hongruan generated lot of noise on OEML Dec 30 21:20:06 re hrw Dec 30 21:20:18 hrw jupp Dec 30 21:20:37 do you know him? Dec 30 21:20:45 nope Dec 30 21:22:18 Hi hrw Dec 30 21:24:08 woglinde: anyway it is much easier to filter OEML - just marked all threads started by that guy as read Dec 30 21:57:09 hrw: Did you get my mail about ltg planet? Dec 30 21:57:40 florian: got it, did not had a time to check Dec 30 21:57:50 florian: feed validator says that all is ok Dec 30 21:58:21 * florian shoots planet Dec 30 22:06:47 has anyone here attempted to use ts_lib in your image after the recent xorg version bump? Dec 30 22:23:54 sakoman_, yes I have it on the bug + on the htcdream Dec 30 22:24:01 xf86-input-tslib Dec 30 22:24:11 be shure to use the last version Dec 30 22:25:10 let me check to see which version of xf86-input-tslib is being used Dec 30 22:26:23 0.0.6-r7 Dec 30 22:29:11 that should be fine Dec 30 22:29:19 what's the problem exactly? Dec 30 22:29:57 on first boot I get a black screen with blinking cursor after the ts_calibrate Dec 30 22:30:20 seems to hang somewhere in the Xinits Dec 30 22:30:41 ah ok Dec 30 22:30:47 if I reboot from the console the boot completes and I get a normal gpe screen Dec 30 22:31:05 gpe??? I thought you were doing some gnome work Dec 30 22:31:20 but the ts_lib input is bad -- almost seems like it is taking relative inout rather than absolute Dec 30 22:31:34 one of the images for gumstix is gpe based Dec 30 22:31:40 ok Dec 30 22:31:45 I'll boot and look Dec 30 22:31:46 for their lcd based products Dec 30 22:31:52 ok Dec 30 22:32:16 running ts_test from the command line shows perfect calibration Dec 30 22:32:35 so it is som mismatch between latest xorg stuff and tslib Dec 30 22:33:32 sakoman_, ah so it's not xorg.conf? Dec 30 22:33:58 I'm using the same xorg.cong I've been using for the last year or so Dec 30 22:34:01 here's htcdream config: http://pastebin.com/m6e97f680 Dec 30 22:34:09 ah ok Dec 30 22:34:13 mmm Dec 30 22:34:25 I did ts_calibrate and it worked for me Dec 30 22:35:11 else if you didn't do rm_work you can do man /path/to/the/tslib/man/file to look Dec 30 22:35:19 prior to this I didn't have a "InputDevice" section -- things "just worked" :-) Dec 30 22:35:19 maybe they changed some option Dec 30 22:35:31 I'll look my xorg version Dec 30 22:35:35 mmm Dec 30 22:35:51 so you used xorg before? Dec 30 22:35:58 yes, for years Dec 30 22:36:18 (seems like anyway) Dec 30 22:36:44 http://pastebin.com/m5405ea2d seem sufficient for bug Dec 30 22:36:47 mmm Dec 30 22:37:01 this broke with the version bump from 1.6.1 to 1.7.3 Dec 30 22:37:04 else you could look in the logs Dec 30 22:37:05 ok Dec 30 22:37:13 Xorg.log.0 or something like that Dec 30 22:37:47 I bet I'm not realy helpfull Dec 30 22:37:50 yeah, I tried an "InputDevice" section much like that -- didn't help :-( Dec 30 22:38:00 yes but you need to activate it Dec 30 22:38:07 like that: Dec 30 22:38:49 http://pastebin.com/m15cf297e Dec 30 22:39:02 but the key thing is to look into the logs Dec 30 22:39:11 it should tell you how it is recognized Dec 30 22:39:20 right, I did that :-) Dec 30 22:39:28 and found nothing...:( Dec 30 22:39:30 ah, a clue in the log :-) Dec 30 22:39:34 ah ok Dec 30 22:39:50 (EE) Failed to load module "tslib" (module requirement mismatch, 0) Dec 30 22:39:54 also if you have corepointer.... Dec 30 22:39:57 ouch Dec 30 22:40:01 ok Dec 30 22:40:12 I'll check xorg versions Dec 30 22:40:31 I see the issue! Dec 30 22:40:35 ah no I've an old xorg Dec 30 22:40:36 sorry Dec 30 22:40:58 the version bump did not trigger rebuilds of the xf86 stuff Dec 30 22:41:13 they are still compiled for 1.6.1 :-( Dec 30 22:41:20 but on the other device: X.Org X Server 1.7.3 Dec 30 22:41:28 ok Dec 30 22:41:36 so perhaps just manually forcing a rebuild will do the trick for me Dec 30 22:41:37 and it works on the 1.7.3 device Dec 30 22:41:48 mmm Dec 30 22:41:54 maybe there are other things than PR Dec 30 22:41:59 like inc_PR or similar Dec 30 22:42:20 I never had to use them...so I don't know how they work Dec 30 22:43:07 for the record, here is the error you get when you have 1.7.3 xorg and older xf86 stuff: http://pastebin.com/m7f327eeb Dec 30 22:43:44 ok Dec 30 22:44:07 back to fixing abiword then Dec 30 22:44:12 :-) Dec 30 22:44:16 good luck! Dec 30 22:44:35 thanks Dec 30 22:44:39 but it's quite simple Dec 30 22:44:42 there are 3 issues: Dec 30 22:44:55 one with a program generating cflags and similar Dec 30 22:45:00 is fixed locally Dec 30 22:45:08 one for dolt Dec 30 22:45:26 one for the plugins becuase we do not run autogen.sh Dec 30 22:45:43 (autogen.sh runs a custom script which I'll import and run) Dec 30 22:45:52 so no collab etc... Dec 30 23:11:16 mmm maybe we could teach to remove DOLT to a class? Dec 30 23:11:57 gnutoo abiword is using dolt? Dec 30 23:12:04 s/a class/autotools.bbclass Dec 30 23:12:04 yes Dec 30 23:12:12 3 packages identified so far Dec 30 23:12:17 I bet there are more Dec 30 23:12:25 remove just the marco invokation in configure.in/ac Dec 30 23:12:29 that should be enough Dec 30 23:12:34 the 2.8.1 is using it Dec 30 23:12:39 normal libtoll should take over Dec 30 23:12:44 yes that's I wanted to do Dec 30 23:12:46 as far as I read dolt.m4 Dec 30 23:12:53 but my question was where to do it Dec 30 23:13:09 if I do it in autotools.bbclass is it ok(would need review) Dec 30 23:13:24 and the fix work(removing DOLT) Dec 30 23:14:41 no Dec 30 23:14:46 please fix per recipe Dec 30 23:15:06 you dont know the software uses dolt Dec 30 23:15:29 sed ? Dec 30 23:16:10 what I'm afraid is if every autotool based program start using dolt Dec 30 23:16:14 what is dolt is used in a private .m4 file? Dec 30 23:16:20 no Dec 30 23:18:18 ok thanks a lot Dec 30 23:18:41 I'll patch or sed inside the recipe themselves and not in autotools.bbclass Dec 30 23:18:50 why sed? Dec 30 23:18:55 I would make a patch Dec 30 23:19:05 ok I'll make a patch Dec 31 00:24:00 03Andrea Adami  07org.openembedded.dev * ra45b6d5b03 10openembedded.git/recipes/kexecboot/linux-kexecboot.inc: linux-kexecboot.inc: favour lzma for initramfs.cpio for k. >= 2.6.32 Dec 31 00:24:12 03Andrea Adami  07org.openembedded.dev * r572abc3fdd 10openembedded.git/classes/kernel.bbclass: Dec 31 00:24:17 kernel.bbclass: copy gz, lzo, lzma initramfs cpio images. - since 2.6.30 there Dec 31 00:24:19 is choice between gzip and lzma for the initramfs.cpio - in patchwork there are Dec 31 00:24:21 pending patches for lzo compressor, so we add lzo too - patch is only concerning Dec 31 00:24:25 initramfs images Dec 31 00:25:58 03Khem Raj  07org.openembedded.dev * r83d170ac1b 10openembedded.git/recipes/mpc/ (mpc-native_0.8.1.bb mpc.inc mpc_0.8.1.bb): Dec 31 00:26:01 libmpc: Rename newly added mpc to libmpc. Dec 31 00:26:03 * there already exists a recipe called mpc Dec 31 00:26:05 so call this one as libmpc to avoid DEPENDS Dec 31 00:26:07 collision. Dec 31 00:26:09 Signed-off-by: Khem Raj Dec 31 00:26:25 03Khem Raj  07org.openembedded.dev * rd83ab1a2b1 10openembedded.git/recipes/libmpc/ (libmpc-native_0.8.1.bb libmpc_0.8.1.bb): Dec 31 00:26:30 libmpc: Rename require filenames in recipes to reflect recent name Dec 31 00:26:37 change. Dec 31 00:26:41 Signed-off-by: Khem Raj Dec 31 00:26:45 03Khem Raj  07org.openembedded.dev * rcae8d0ad28 10openembedded.git/recipes/gcc/ (6 files in 2 dirs): Dec 31 00:26:54 gcc-cvs: DEPEND on libmpc instead of mpc. Dec 31 00:26:58 * Define BINV to 4.5.0 Dec 31 00:27:02 * Add fortran-cross-compile-hack.patch. Dec 31 00:27:06 * Bump SRCREV. Dec 31 00:27:08 Signed-off-by: Khem Raj Dec 31 00:39:50 is there a way to tell bitbake to do verbose mode or something?> Dec 31 00:39:57 caues stuff is erroring out without any real message Dec 31 00:46:16 bitbake -DDD Dec 31 00:49:30 cool thanks Dec 31 00:51:57 man I need to get a laptop or something Dec 31 00:52:02 that I Can install native debian Dec 31 00:52:10 cause virtualbox kind of kills my CPU Dec 31 00:53:35 hmm Dec 31 00:53:36 what the hell Dec 31 01:04:38 erroring out on building libxml Dec 31 01:04:42 make[1]: *** No rule to make target `include/lhbxml/HTMLparser.h', needed by `SAX.lo'. Stop. Dec 31 01:04:56 would that be because I ned to install something else to get those headers? Dec 31 01:05:21 wait Dec 31 01:05:25 why is it lhbxml... Dec 31 01:05:32 makefile has a typo methinks Dec 31 01:06:04 hm strange Dec 31 01:08:42 hmm Dec 31 01:08:45 its not in the makefile Dec 31 01:08:52 it syas its needed by SAX.lo Dec 31 01:08:59 but SAX.lo doesn't have any references to it either Dec 31 01:09:39 what's an .lo file? Dec 31 01:13:09 yeah Dec 31 01:13:21 SAX.Plo in .deps Dec 31 01:13:25 has a typo Dec 31 01:13:38 is there any way to like... notify someone about this? Dec 31 01:32:21 Aditya1 look upstream if it is fixed Dec 31 01:32:54 "look upstream"? Dec 31 01:33:12 yeah were the soft ware is developed Dec 31 01:33:21 we are mostly only packaging it Dec 31 01:33:28 hm Dec 31 01:33:38 well when I changed the one typo it worked fine Dec 31 01:33:53 you are saying I should talk to libxml people? Dec 31 01:34:18 you should lookup if it is fixef Dec 31 01:34:20 fixed Dec 31 01:35:12 sorry to be dense, but I am still alittle confused, how would I check that? Dec 31 01:35:42 in the scm tool they are using? Dec 31 01:35:49 svn git mercurial cvs Dec 31 01:36:12 Aditya1 are you familar with freesoftware? Dec 31 01:36:14 re zecke Dec 31 01:36:41 umm Dec 31 01:36:47 I use linux somewhat frequently Dec 31 01:36:53 but I am new to OE Dec 31 01:37:07 so you know what a scm is? Dec 31 01:40:26 I know what git, subversion etc are if thats what oyu mean Dec 31 01:41:00 then use the webfront of the scm of this xml lib Dec 31 01:41:07 look if they fixed it upstream Dec 31 01:42:56 hm k Dec 31 01:43:36 if it is fixed upstream we might need a newer version Dec 31 01:43:51 otherwise we can make a real patch out of your solution Dec 31 02:10:18 http://git.gnome.org/browse/libxml2/tree/ **** ENDING LOGGING AT Thu Dec 31 02:59:57 2009