**** BEGIN LOGGING AT Fri Aug 13 02:59:57 2010 Aug 13 03:00:09 Remember when you are young.... Aug 13 03:00:15 and the life is long Aug 13 03:48:13 kergoth, nope. i just didn't realise Aug 13 03:53:28 :) Aug 13 03:54:13 khem, 100mins for angstrom/x11-image Aug 13 04:03:21 grg: thats from ground zero ? Aug 13 04:03:40 khem, yup Aug 13 04:03:43 grg: wow I am impressed then Aug 13 04:03:58 grg: now I think I should get something like that Aug 13 04:04:03 its good price/perf Aug 13 04:04:07 indeed Aug 13 04:04:19 one could buy i7 980X Aug 13 04:04:29 but its 1000 dollars just for chip Aug 13 04:04:38 I dont know how much will I gain Aug 13 04:05:04 920 is around 280 dollars Aug 13 04:05:11 which is way less Aug 13 04:05:17 probably better off investing that extra money for when you decide to upgrade in 3 years Aug 13 04:05:26 yup Aug 13 04:05:36 exactl Aug 13 04:05:50 Anyone know the Motorola A1200 ? Aug 13 04:06:03 and as long as my motherboard has the given socket Aug 13 04:06:08 I can put them in Aug 13 04:06:21 so I think for now 920 makes sense Aug 13 04:06:28 im developing a new operational system to it! i need oe to compile opie! Aug 13 04:06:45 and all 32nm that are 1000 usd today will be 200usd in 2 years :) Aug 13 04:06:56 and I can easily upgrade Aug 13 04:09:04 grg: for your gnome-doc-utils patch can you use inherit gettext instead ? and see if that works **** ENDING LOGGING AT Fri Aug 13 04:13:39 2010 **** BEGIN LOGGING AT Fri Aug 13 04:14:14 2010 Aug 13 04:23:24 kergoth, khem: bitbake gnome-doc-utils-native fails with inherit gettext. Aug 13 04:24:12 (same failure as in my email) Aug 13 04:27:30 gettext.bbclass doesn't add a dependency for virtual/native Aug 13 04:27:45 virtclass-native i mean Aug 13 04:33:14 The filtering out of gettext from DEPENDS in the case that USE_NLS="no" wouldn't make sense for gnome-doc-utils anyway. The configure script barfs out because of a missing gettext.m4 macro. Aug 13 04:34:38 why would it use gettext macros if it doesn't use gettext, and therefore have the gettext configure arguments? Aug 13 04:34:44 that doesn't make any sense Aug 13 04:35:01 oh, i see what you mean.. nevermind :) Aug 13 04:35:11 we should fix this in multiple ways, i'd htink Aug 13 04:35:28 hmm Aug 13 04:35:33 well, nevermind, your patch is good for now Aug 13 04:36:41 grg: we should probably just remove that conditional dep bit from the bbclass, unless we're going to patch in gettext.m4 to everything Aug 13 04:36:43 heh Aug 13 04:47:14 Hi! I'm looking into conversion of htmldoc recipes to BBCLASSEXTEND usage. Aug 13 04:47:14 There are 2 htmldoc recipes both using htmldoc-native for generation of their docu. Aug 13 04:48:17 I see no example in recipes where someone with BBCLASSEXTEND is using it's own -native. Is this possible? Aug 13 04:48:52 yep, should work fine. worst case you can override the depends for virtclass-native to ensure it doesn't get a circular dep, but that might not be necessary Aug 13 04:49:14 and which recipe should get BBCLASSEXTEND? htmldoc_x.x or htmldoc-gui_x.x? Aug 13 04:50:51 * kergoth has no idea what htmldoc-gui is Aug 13 04:53:36 Ah, it's htmldoc with compile option to compile a GUI. Needs X installed which customers machine didn't have. Aug 13 04:54:07 So it's either htmldoc or htmldoc-gui, they should RCONFLICT Aug 13 04:54:33 ah, indeed Aug 13 04:55:04 and you only need the extend in the htmldoc, not the gui Aug 13 04:55:42 Hmm, OK. I will try to get it working. Thank you for your help! Aug 13 05:03:35 np Aug 13 06:14:32 hi Aug 13 06:14:38 I am trying to get opencv-dev Aug 13 06:14:46 but when I bitbake for beagleboard Aug 13 06:14:49 it says "Nothing PROVIDES 'opencv-dev'" Aug 13 06:14:56 does anyone know what I can do? Aug 13 06:15:01 I need libcv and the like Aug 13 06:33:45 aditya_1010: bitbake opencv Aug 13 06:34:27 that will also create the dev package (might require a flag in local.conf or so, but thought it was default behaviour to also create -dev packages Aug 13 06:34:31 btw gm everyone Aug 13 06:38:14 morning Aug 13 06:38:18 yeah it does opencv fine Aug 13 06:38:23 but the libs never get made Aug 13 06:38:26 guess I'll check teh recipe Aug 13 06:39:10 EXTRA_OECONF = "--disable-debug ... that might be a problem heh Aug 13 06:39:53 might be Aug 13 06:40:03 although that might be like debugging info Aug 13 06:40:04 hmm Aug 13 06:40:23 btw I know some people in #beagle also are working on opencv Aug 13 06:40:42 although none might be there now, still only 8.40 in europe and 7.40 in UK Aug 13 06:42:07 damn you europeans Aug 13 06:42:12 why do you have to live in a different timezone Aug 13 06:46:12 We ? It are the people in US who had to move from the UK and GMT to some far away TZ ? Aug 13 06:46:23 * eFfeM_work assumes aditya_1010 is in US Aug 13 06:46:27 yes Aug 13 06:46:29 unfortunately Aug 13 06:46:37 in possibly the most boring state... ever Aug 13 06:46:45 eFfeM_work: it's 9:40 here in the EEST zone :) Aug 13 06:47:21 do you guys know if Mplayer build for OE/beagleboard does anything special ? Aug 13 06:47:28 dm8tbr: what country? Aug 13 06:47:34 .fi Aug 13 06:47:36 aditya_1010: not that I am aware of Aug 13 06:47:42 by special I mean use the DSP or something Aug 13 06:48:01 I am just trying to see if I can mimic what they are doing for capturing video from a webcam Aug 13 06:48:10 hehe.. Aug 13 06:48:12 dm8tbr: ah somehow I assumed .fi to be in the same tz as the other nordic countries Aug 13 06:48:15 TZ flame :) Aug 13 06:48:23 unless you guys know of better ways to do this Aug 13 06:48:31 Jay7: nah, for that the solution is UGT Aug 13 06:48:48 let's stop the Earth :) Aug 13 06:48:49 actually did you know that although there are only 24 hrs in a day the distance between the most eastern and western TZ is 26 or 27 hrs Aug 13 06:48:59 ~ugt Aug 13 06:49:00 well, ugt is Universal Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html Aug 13 06:49:15 before someone asks: what is ugt Aug 13 06:49:48 :) Aug 13 06:49:58 * dm8tbr had the link at hand :) Aug 13 06:51:47 do you guys know if there is a sort of recommended way to capture images from a webcam? Aug 13 06:51:55 for beagleboard/OMAP Aug 13 06:51:56 What version of bitbake do you use now? I see 1.8.18 is the latest on the webpage, but I have 1.10, and don̈́'t remember where I found it... Aug 13 06:52:19 tasslehof, cgit.openembedded.org Aug 13 06:52:27 bitbake 1.10 branch Aug 13 06:53:49 aditya_1010: not sure if thre is a recommended way, in the past I used vidcat to capture a picture, it is part of the w3cam package Aug 13 06:53:58 * eFfeM_work also used motion Aug 13 06:59:35 hm Aug 13 06:59:47 mplayer seems to do a bangup job of capture Aug 13 06:59:56 which is why I was trying to mimic it Aug 13 07:16:54 03Martin Jansa  07org.openembedded.dev * r8786dbcabd 10openembedded.git/recipes/pidgin/pidgin_2.7.2.bb: Aug 13 07:16:55 pidgin: 2.7.2->2.7.3 Aug 13 07:16:55 Signed-off-by: Martin Jansa Aug 13 07:26:05 holy jesus Aug 13 07:26:16 installing opencv-dev on bboard installs about 2012873198 packages Aug 13 07:27:15 aditya_1010, install with --nodeps Aug 13 07:27:26 too late now lol Aug 13 07:27:32 its already installed about 38479347 Aug 13 07:27:38 so figure I'll wait for the rest Aug 13 07:28:07 * grg shrugs. Its your time and sdcard(?) space Aug 13 07:28:30 hmm Aug 13 07:28:36 maybe yeah I should do nodeps Aug 13 07:29:49 hmm Aug 13 07:29:58 8 FPS on average with opencv Aug 13 07:30:03 thats not terribly fast Aug 13 07:34:34 can i have multiple locol.cnf? Aug 13 07:35:17 can i have multiple local.conf? Aug 13 07:43:35 eFfeM_work: should I stay stable, or use 1.10? any opinions? Aug 13 07:45:12 use head :) Aug 13 07:45:17 I mean head version Aug 13 07:45:31 you will help to test it ;) Aug 13 07:49:49 did some knwon how can I ceate a ipk server? Aug 13 07:49:54 for the host and the target Aug 13 08:00:11 Jay7: I will use the head and my head :) Aug 13 08:10:42 hi, is there any known issue with opkg-0.1.8+svnr543-r1 ?? Aug 13 08:11:13 my ./configure doesn't get that it is a cross compiled env: Aug 13 08:11:17 configure: error: C compiler cannot create executables Aug 13 08:13:00 ceyusa: read config.log Aug 13 08:13:31 ceyusa: and if it's some strange problem there, then -c clean whole toolchain and build again (this almost always helps) Aug 13 08:15:19 JaMa: thanks, reading confing.log ... the whole toolchain!? :( Aug 13 08:32:59 khem: can you check your ./x86_64-linux/usr/armv5te/bin/arm-angstrom-linux-gnueabi-klcc if you have $prefix pointing to ${WORKDIR}/.../klibc/image? I do and as soon as rm_work removes klibc workdir it fails to find its crtO.o Aug 13 08:33:24 khem: and after this replace works ok again :) sed -i 's#\\/OE\\/tmpdir\\-dev\\/work\\/spitz\\-angstrom\\-linux\\-gnueabi\\/klibc\\-1\\.5\\.18\\-r18\\.0\\/image\\/#\\/OE\\/tmpdir\\-dev\\-shr\\/sysroots\\/armv5te\\-oe\\-linux\\-gnueabi\\/#g' /OE/tmpdir-dev/sysroots/x86_64-linux/usr/armv5te/bin/arm-angstrom-linux-gnueabi-klcc Aug 13 09:03:41 Hi eFfeM_work Aug 13 09:03:59 hi Aug 13 09:04:31 are you going to send a new patch for genboot-native ? Aug 13 09:04:58 I need to ask you one thing. I send a patch on openembedded.dev list Aug 13 09:05:17 ask Aug 13 09:05:38 do i have to add PR although it is not required Aug 13 09:06:00 secondly i got a feedback to run oe_stylize.py on recipe Aug 13 09:06:17 when i ran it is saying to remove tab with spaces Aug 13 09:06:26 if there is no PR in the recipe better add one, and set it to r1, better safe than sorry Aug 13 09:06:30 the tabs are already there i didnt add those tabs Aug 13 09:07:08 so what should i do, should i change the tab with spaces.other recipes also have tabs Aug 13 09:07:14 well oe-stylize will do that for you, what I do is run oe-stylize, redirect output to a file, compare files with diff, if needed rebuild and remove the # comment on tabs Aug 13 09:07:36 i have no strong preference here, but apparently the policy is to use spaces Aug 13 09:07:57 can any of the oldtimers comment on it Aug 13 09:08:17 but the other recipes have tabs. Aug 13 09:08:31 i know Aug 13 09:09:02 mickeyl, JaMa, hrw, denix etc any preference on tabs vs spaces ? Aug 13 09:09:25 spaces are preferred afaik Aug 13 09:09:25 secondly it also gives error on NATIVE_INSTALL_WORKS Aug 13 09:09:41 it says unknown variable/routine Aug 13 09:10:08 you do not need it I think Aug 13 09:10:09 eFfeM_work: check with oe-stylize, but every time I notice tabs, I replace those with spaces (they are usually only in part of recipe) Aug 13 09:10:28 noor, there is your answer ^^^ Aug 13 09:10:30 iirc there was some discussion years ago...we concluded that indents are not standardized (some are 4, other 8) Aug 13 09:10:53 kernel standardises on 8, but I am happy with spaces Aug 13 09:11:00 iirc Python too is 8 Aug 13 09:11:32 jama I've been considering doing a big sed tab to 8 spaces on all recipes, but I was not 100% sure it would not break anything Aug 13 09:12:40 http://www.secnetix.de/olli/Python/block_indentation.hawk Aug 13 09:12:41 norr you do not need NATIVE_INSTALL_WORKS Aug 13 09:13:41 hm patchwork is dropping patches again, the canutils patches are not there Aug 13 09:17:55 is there a good way to retrieve a patch from gmail ? Aug 13 09:18:45 eFfeM_work: there is "Show original" in right menu Aug 13 09:19:12 yeah used that, only gives the msg not the thread Aug 13 09:19:26 would like to have it in mbox fmt, oh well Aug 13 09:19:34 just need to readd the ack Aug 13 09:21:13 grr, save original gives the file but git am then says patch format detection failed Aug 13 09:22:23 03Frans Meulenbroeks  07org.openembedded.dev * r4b2780d203 10openembedded.git/conf/distro/ (include/preferred-om-2008-versions.inc openmoko.conf): (log message trimmed) Aug 13 09:22:23 openmoko: removed distro Aug 13 09:22:23 As discussed on #oe on Aug 10, 2010. Aug 13 09:22:23 Removed because this is not used or maintained any more. Aug 13 09:22:23 A message Aug 13 09:22:24 * Aug 13 09:22:24 * Aug 13 09:22:25 03Vitus Jensen  07org.openembedded.dev * r06aea27b18 10openembedded.git/recipes/socketcan/canutils_4.0.4.bb: Aug 13 09:22:26 canutils: remove old version 4.0.4 Aug 13 09:22:26 Signed-off-by: Vitus Jensen Aug 13 09:22:27 Acked-by: Frans Meulenbroeks Aug 13 09:22:27 03Vitus Jensen  07org.openembedded.dev * r3fa4b5ff01 10openembedded.git/recipes/socketcan/libsocketcan_0.0.7.bb: Aug 13 09:22:28 libsocketcan: remove old version 0.0.7 Aug 13 09:22:28 Signed-off-by: Vitus Jensen Aug 13 09:22:29 Acked-by: Frans Meulenbroeks Aug 13 09:22:29 03Vitus Jensen  07org.openembedded.dev * r8ffa417ea2 10openembedded.git/recipes/socketcan/canutils_4.0.5.bb: Aug 13 09:22:48 OK then I'll re submit the genboot path with spaces and PR Aug 13 09:23:00 *patch Aug 13 09:23:07 eFfeM_work: can you ACK my preferred-xorg-version removals? Aug 13 09:23:15 hm, not sure why it now pushes the om removal, thought that was done before Aug 13 09:23:30 JaMa: yes, if I have it somewhere Aug 13 09:24:51 eFfeM_work: Message-Id: <1277552779-12490-1-git-send-email-Martin.Jansa@gmail.com> Aug 13 09:25:18 as shown in http://patchwork.openembedded.org/patch/2262/ Aug 13 09:25:39 JaMa: i don't have the message any more so cannot reply on it Aug 13 09:26:27 eFfeM_work: just write right subject and add "In-Reply-To" header Aug 13 09:26:48 read the patch on patchwork is fully compliant with what I proposed Aug 13 09:27:47 i'm not sure I can do that in gmail, but I think it is ok if I formally ack it here Aug 13 09:28:12 so JaMa for the xorg patch: Acked-by: Frans Meulenbroeks Aug 13 09:28:26 ok :) Aug 13 09:28:30 if you like an ack by mail please fwd the msg to me and I'll reply with the list on cc Aug 13 09:28:32 to ro Aug 13 09:28:35 or to Aug 13 09:34:03 03Martin Jansa  07org.openembedded.dev * r83c3934bd5 10openembedded.git/conf/distro/include/ (7 files): (log message trimmed) Aug 13 09:34:03 preferred-xorg-versions: remove include files for old releases Aug 13 09:34:03 * git grep shows that all distributions (in OE tree) moved to Aug 13 09:34:03 preferred-xorg-versions-X11R7.5.inc already Aug 13 09:34:03 * would be nice to encourage everybody to use most tested X11R7.5 or Aug 13 09:34:04 no preferred-xorg-versions at all (we can keep latest versions working Aug 13 09:34:05 and use D_P -1 for stuff which needs more testing first) Aug 13 09:34:14 good morning Aug 13 09:35:23 hi florian Aug 13 09:37:36 I agree preferred_versions list should be as short as possible, then D-P Aug 13 09:38:26 there are some more unused one which I suggested to remove Aug 13 09:39:35 because no one uses them (e.g. gpe, om-2009, maemo) Aug 13 09:39:58 btw I think bluelightning is trying to regroup opie recipes Aug 13 09:40:33 yeah, he told me there was going to be a new version, he planned to remove some of the older ones Aug 13 09:40:40 we have long long OpenZaurus legacy Aug 13 09:41:55 Poky split the recipe dirs in meta-* iirc Aug 13 09:42:14 this would allow to move most cruft in a separate dir Aug 13 10:05:49 hmm, has anyone noticed a problem, than when updating packages that have rc.d scripts, the script is not being updated? Aug 13 10:06:06 I'm not on the newest dev though, gotta browse through the commit messages Aug 13 10:10:41 eFfeM_work: when run the updated recipe without NATIVE_INSTALL_WORKS populate_sysroots runs Legacy staging Aug 13 10:10:50 so i need to have this varaible in my recipe Aug 13 10:15:00 but oe_stylize.py is giving me warning and does not show that variable in its output Aug 13 10:17:09 ooops I am wrong it is showing that variable Aug 13 10:17:54 but y it is giving warning tha unknown variable/routine "NATIVE_INSTALL_WORKS" Aug 13 10:22:34 03Koen Kooi  07org.openembedded.dev * rea353d70c3 10openembedded.git/conf/distro/include/sane-srcrevs.inc: opkg: bump SRCREV to 550, fixes a segfault and some symlink truncation issues Aug 13 12:26:37 is NATIVE_INSTALL_WORKS needed for native recipes (looking at the new version of serload-native) Aug 13 12:26:40 ? Aug 13 12:32:05 eFfeM_work: when I dont give varaible for genboot do_stage function is called Aug 13 12:33:27 and genboot-native recipe failes in do_stage function Aug 13 12:45:19 eFfeM_work: it doesn't need to there for serlod-native recipe Aug 13 12:45:25 i'll update it Aug 13 12:45:41 bit it does need for genboot-native recipe Aug 13 12:48:47 aaahhh I am worng Aug 13 12:48:56 i'll update that recipe as well Aug 13 12:50:36 helllo Aug 13 12:50:39 hello* Aug 13 12:51:34 im getting this error when bitbake opie-tictac : http://pastebin.com/sSJCbdur Aug 13 12:58:05 hi? Aug 13 13:07:38 hello???? Aug 13 13:10:33 hello Aug 13 13:10:37 anyhelp here? Aug 13 13:11:26 shit Aug 13 13:11:31 no one help me Aug 13 13:11:34 i m cry Aug 13 13:11:36 -) Aug 13 13:13:20 help Aug 13 13:13:20 help Aug 13 13:13:27 anyone can help Aug 13 13:13:56 Angelox_123, be patient. Aug 13 13:16:36 Angelox_123, I don't use uglibc images so I don't know exactly what's happening. Just wondering did you built the toolchain using "bitbake meta-toolchain" ? Aug 13 13:17:43 no..i just run bitbake opie-tictac and it compiled toolchain Aug 13 13:17:43 s/did you built/did you build/ Aug 13 13:18:19 i just run bitbake opie-tictac and it compiled toolchain Aug 13 13:18:42 I see, maybe some piece of the cross compiler is missing, try building meta-toolchain first Aug 13 13:19:01 i need erase the tmp directory? Aug 13 13:21:07 Angelox_123, I don't think so. Aug 13 13:21:35 ok..lets try Aug 13 13:27:02 thanks all help! Aug 13 13:27:23 `Running task 376 of 917` slow..... =) Aug 13 13:29:41 so definitely there were pieces missing :) Aug 13 14:12:28 I'm very curious about rubygems sqlite3 gem. The whole time I thought there is something wrong with my filesystem since gem couldn't find the sqlite3.h in the /usr/include directory. Now it seems to be a problem with my environment. Aug 13 14:13:44 If there a possibility to install gcc on the embedded device? A better way would be to crosscompile the rubygem on my dev machine. But I don't have any clue how/where to begin. Can anybody help me with it? Aug 13 14:14:19 Taalas..compile gcc on embedded is possible! Aug 13 14:15:03 to* Aug 13 14:16:21 Angelox_123 ok. And how? Aug 13 14:17:28 Have a toolchain...and configure correctly the GCC using --target= param Aug 13 14:18:16 you want gcc *on* the device? there are tasks / images in OE that include development tools Aug 13 14:18:52 i think meta-toolchain build Aug 13 14:30:36 meta-toolchain builds an sdk with a crosscompiler and libs/headers and the like, it has nothing to do with development *from* the target Aug 13 14:56:26 same error Aug 13 14:56:38 http://pastebin.com/sSJCbdur Aug 13 15:11:54 help here? Aug 13 16:05:54 pushed a bitbake-1.10.0.tar.gz to berlios and pushed a corresponding tag in the git repository Aug 13 16:07:12 any idea why we see lots of errors about ERROR: '[]' RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'virtual/arm-oe-linux-gnueabi-gcc-2.95' but it wasn't found in any PACKAGE or RPROVIDES variables Aug 13 16:07:16 Unknown Event: Aug 13 16:07:19 ? Aug 13 16:07:24 I know it's from 2.4 kernels Aug 13 16:07:39 I'm trying to blacklist those in shr.conf now to get rid of that error message Aug 13 16:07:51 but something changed in bitbake/metadata that it's shown now Aug 13 16:07:57 it wouldn't be issuing that message unless something in your build is trying to pull in something that depends on it Aug 13 16:08:26 kergoth_: it's shown while parsing recipes I think, but I'll recheck Aug 13 16:08:45 kergoth_: and it's not fatal.. only shown Aug 13 16:09:15 hmm Aug 13 16:10:21 also those 2: Aug 13 16:10:22 Missing or unbuildable dependency chain was: ['virtual/arm-oe-linux-gnueabi-depmod-None'] Aug 13 16:10:25 DEBUG: File '/OE/dev/recipes/linux/linux-ml403-mvista-2.6.x_git.bb' is unbuildable, removing... Aug 13 16:10:28 DEBUG: File '/OE/dev/recipes/linux/linux-davinci_git.bb' is unbuildable, removing... Aug 13 16:10:54 And it's build for om-gta02 so no reason to depend on linux-davinci afaik Aug 13 16:15:19 kergoth_: maybe it's because now it's checking also runtime providers? After blacklisthing those I see "ERROR: Nothing RPROVIDES 'collie-kernel-64-0'" Aug 13 16:16:10 kergoth_: and also the warning about multiple providers which RPROVIDES the same is quite new Aug 13 16:16:17 strange. Aug 13 16:22:47 hey. any of you guys here have done jtag debugging on a beagleboard? Aug 13 16:23:07 i need some advice in that area :) Aug 13 16:23:32 didn't think the header was exposed Aug 13 16:31:56 kergoth: 36b980c16bf74b3c2066cc120f9f27e11f189a63 in bitbake master maybe? Aug 13 16:35:03 I'm failing build of squashfs-tools-native-4.0-r3.3 b/c its looking for usr/include/attr/xattr.h in STAGING_DIR - what would provide attr/xattr.h? I don't see any recipe for libattr Aug 13 16:50:29 kergoth: hmm and they are pulled to build probably because ie c7x0-kernels-2.4-embedix depends on some and PROVIDES = "virtual/kernel" Aug 13 17:03:24 03Martin Jansa  07org.openembedded.dev * r1443fb7bcc 10openembedded.git/conf/distro/shr.conf: Aug 13 17:03:24 shr: blacklist all 2.4 kernels, because of multiple ERRORs it shows in every build Aug 13 17:03:24 Signed-off-by: Martin Jansa Aug 13 17:08:50 tharvey, that's an old version Aug 13 17:08:52 been fixed since Aug 13 17:09:01 i'd swear, anyhow Aug 13 17:09:05 since there's a dep on attr-native Aug 13 17:09:28 and i went and updated the recipe yesterday and i don't have that header on my host Aug 13 17:09:35 Tartarus, old ver of squashfs-tools-native? Aug 13 17:17:20 Tartarus, found my issue - I needed libattr1-dev on my build host (which is a freshly installed ubuntu 10.04 system) - I have found in the past that there are several lib*-dev pkgs needed on the host for various OE recipes which is a little disconcerting... seems like if an OE recipe depends on something it should build a native version of that or display a clear error that you need that pkg (such is done for things like autoconf texinfo gawk subverison cvs, Aug 13 17:17:20 etc) Aug 13 17:21:15 damn... thats still not it - libattr1-dev used to be required, perhaps that got fixed improperly and that's what your patch is for Tartarus? still looking for your change Aug 13 17:43:53 re Aug 13 17:51:34 hiyo Aug 13 17:51:45 anyone here working with OE/beagleboard with a webcam? Aug 13 17:51:57 I am trying to figure out what the reccomended way of doing this is Aug 13 17:52:15 I am talking about capturing images from a wbecam in a program for realtime processing Aug 13 17:52:26 I have tried gstreamer and opencv and both give me about 8 FPS Aug 13 17:52:36 mplayer somehow does at around 15FPS Aug 13 17:52:49 not entirely sure what thats doing Aug 13 18:13:40 my issue with squashfs-tools-native is fixed by http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=12e8838d9e154e4b99de381ef4be93ecf81a759c - seems horrible that some recipes pull from repos without specifying SRCDATE's Aug 13 18:13:59 seems like there should be something that warns you about all pkgs that do that Aug 13 18:14:10 something in sanity.bbclass perhaps Aug 13 19:10:59 what purpose have variable ONLINE_PACKAGE_MANAGEMENT? Aug 13 19:11:45 what is purpose of this variable? Aug 13 20:43:27 I'm trying to add a patch for a SRC_URI in a kernel module package only for a specific kernel provider. Trying via a python do_unpack function here: http://pastebin.mozilla.org/767398 - I find that the patch is getting fetched, but its not applied. Any ideas on how to fix it or is there a better way? Aug 13 20:45:03 tharvey: why not use override Aug 13 20:46:13 OVERRIDE doesn't support providers does it? Aug 13 20:47:03 tharvey: no it doesnt but we can make somehting Aug 13 20:47:15 when do you want this patch to be applied Aug 13 20:47:49 you mean like addsomething to my overirdes? OVERRIDES += "${PROVIDER}" ? Aug 13 20:48:05 no but add machine override Aug 13 20:48:08 which use this kernel Aug 13 20:48:16 or something like hat Aug 13 20:48:18 that Aug 13 20:49:16 the situation is that I have a compat-wireless.bb recipe that builds compat wireless for linux-omap-psp_2.6.32 kernel which needs a patch specific to that kernel so SRC_URI_append_omap3 += "file://linux-omap-psp.patch" isn't appropriate Aug 13 20:49:22 or you could make do_unpack_prepend Aug 13 20:49:58 well thats what I did... I created a do_unpack http://pastebin.mozilla.org/767398 override but it doesn't work - copies the patch but it doesn't get applied Aug 13 20:49:59 ok it this patch intrusive to others Aug 13 20:50:19 you redefined do_unpack Aug 13 20:50:48 yes, this patch would fail the recipe on vanilla kernels (linux-omap-psp_2.6.32 pulled in some patches from 2.6.33 that wireless compat collides with) Aug 13 20:51:21 ok Aug 13 20:51:34 true I did redefine do_unpack but I thought that the bb.build.exec_func("base_do_unpack", d) called the prior - guess thats my issue? (let me test) Aug 13 20:53:33 hm.. Aug 13 20:53:37 incidently I noticed that a compat-wireless recipe was posted to maillist some time back and it was never integrated - seems unfortunately that people post patches that never get pulled in :( Aug 13 20:53:48 do we have problems with libjpeg62? Aug 13 20:54:31 http://tinderbox.openembedded.net/public/logs/task/7127666.txt Aug 13 20:55:02 tharvey: actually your patch seems ok from topline Aug 13 20:59:11 tharvey: is it that its not applying your patch or is it that its not applying any patch Aug 13 20:59:32 Jay7: libjpeg was fixes recently check commit logs Aug 13 21:00:45 hm.. may be I should clean something before.. Aug 13 21:01:07 khem, I think it was not applying just my new patch, let me test again Aug 13 21:02:19 btw you should make a copy of metadata before you make changes Aug 13 21:02:30 localdata = bb.data.createCopy(d) Aug 13 21:02:38 and then operate upon localdata Aug 13 21:03:53 khem, why is that? In this case I'm trying to append something to metadata - is there a better way? Aug 13 21:04:39 for data consistency Aug 13 21:05:24 and since you will be completing the whole task using localdata it will work Aug 13 21:05:46 bb.build.exec_func("base_do_unpack", localdata) Aug 13 21:05:54 should be the last call Aug 13 21:06:00 try this and see if it helps Aug 13 21:08:01 do I want bb.build.exec_func("base_do_unpack", localdata) or 'do_unpack'? I found this from a similar example somewhere... don't really understand why I need to do the exec_func, esp if I use do_unpack_prepend Aug 13 21:09:32 not with prepend of course Aug 13 21:09:45 it was for if you were overriding do_unpack Aug 13 21:11:09 oh I see Aug 13 21:11:25 base_do_unpack operates on global metadaya Aug 13 21:11:32 and creates its own copy and stuff Aug 13 21:11:38 so this approach wont work Aug 13 21:12:09 but you can do is bb.data.update_data(localdata) before calling base_do_unpack Aug 13 21:14:44 tharvey: try something like this http://pastebin.com/iGFrxYkP Aug 13 21:19:16 ping mickeyl Aug 13 21:22:36 did any of you guys see my query about capturing from a webcam Aug 13 21:22:37 ? Aug 13 21:31:58 khem, looks like changing to do_unpack_prepend and eliminating the bb.data.exec call works fine thx Aug 13 21:32:37 oki Aug 13 21:32:40 enjoy Aug 13 21:36:35 khem, not sure why but the other method (like you pasted) doesn't work - in that case the patch gets copied to $WORKDIR but does not get applied as patch Aug 13 21:39:26 I actually like hte do_unpack_prepend method better though, it seems more intuitive - I didn't think of it before Aug 13 22:35:13 tharvey: may be add apply=yes Aug 13 22:37:14 hey khem Aug 13 22:39:21 ant_: hello Aug 13 22:42:25 What's a good way to troubleshoot autoconf-based recipes? I'm doing a fresh build, and it just died on liborc during the linking stage with a bunch of "../arm-angstrom-linux-gnueabi-libtool: line 4641: func_append: command not found" messages. Aug 13 22:59:58 I must not be doing what I think I'm doing -- somehow, cleaning it first allowed it to build. Weird, considering I'm building this in a new TMPDIR. Aug 13 23:07:27 Hi everyone Aug 13 23:08:17 is someone aware of the libtool patches here ? Aug 13 23:08:38 you're going to have to be more specific Aug 13 23:09:52 there is 2 patchs for the libtool-cross recipe Aug 13 23:10:07 what about them? Aug 13 23:10:12 khem, so we have a bad prefix in klcc, pointing to klibc workdir Aug 13 23:10:55 see staging.patch Aug 13 23:10:56 kergoth: I tried them outside of the OE environment Aug 13 23:11:04 and we set +export prefix = $(INST) Aug 13 23:11:08 kergoth: with buildroot (to not mention it) Aug 13 23:11:42 khem, and export INST=${D} Aug 13 23:11:45 kergoth: but it hasn't fixed the problem I had Aug 13 23:11:53 they aren't magic. Aug 13 23:11:59 they do some specific things Aug 13 23:12:18 in particular, keeping libtool from adding the -L for the libdir in .la files you link against, and from adding the -L for the rpath Aug 13 23:12:21 kergoth: I'm always having -L/usr/lib parameters added by libtool Aug 13 23:12:24 are the main ones, if i remember correctly Aug 13 23:12:37 you've confirmed libtool is the one adding it? it isn't passed to the libtool command? Aug 13 23:12:45 it's quite common for the problem to be on the autotools end as well Aug 13 23:12:52 -L$(libdir) Aug 13 23:13:08 there is a -rpath given to libtool Aug 13 23:13:17 because of a .la file Aug 13 23:13:33 libtool converts it to -L Aug 13 23:14:04 anyway, I found another patch this morning Aug 13 23:14:08 from stlinux.com Aug 13 23:14:12 http://pastebin.com/gEDjPWw9 Aug 13 23:14:51 is someone able to comment on this patch ? Aug 13 23:15:20 seems reasonable, that's just a different approach Aug 13 23:15:27 ours tended to just remove the -L adding entirely Aug 13 23:15:34 whereas theirs obeys a prefix variable Aug 13 23:15:39 but seem similar in concept Aug 13 23:15:46 ok Aug 13 23:15:57 looks like they may catch a case or two we don't, it's possible, libtool is a giant maze of duplicated code Aug 13 23:16:04 i'm not sure though, i'd have to double check our patches Aug 13 23:16:29 * kergoth did the first versions of our patches back shortly after OE was started, but it's been a while, and they were improved/redone when libtool 2.0 came around Aug 13 23:16:51 I'm thinking to parse possible --sysroot options given to libtool to avoid the LIBTOOL_PREFIX_BASE variable Aug 13 23:16:56 i like their approach, using a sysroot Aug 13 23:17:11 that'd be nice, if you also add a good way to add that argument to the libtool calls generated in the makefile rules Aug 13 23:17:22 modify the libtool m4 macros or so to add a configure argument Aug 13 23:17:41 the original patches predated the existance of the 'sysroot' concept, really Aug 13 23:18:32 so you're in the business since a while ;) Aug 13 23:19:07 my first exposure to embedded was when i started the openzaurus distro, which is what led to me being one of the OE founders, to replace its buildroot with something more scalable Aug 13 23:19:22 pretty funny, my whole professional coding career got its start in hacking on a zaurus pda Aug 13 23:19:30 :) Aug 13 23:19:55 nice ;) Aug 13 23:20:28 so, yeah, i think that patch is probably a nicer solution than ours in the long term, and a sysroot argument as you suggest is better yet Aug 13 23:20:43 * kergoth would love to see libtool die in a fire, but sadly we're stuck with it Aug 13 23:21:10 * kergoth looks over our patches again Aug 13 23:22:03 our prefix patch is really completely unnecessary Aug 13 23:22:18 all it does is makes it really obvious when a build isn't using our patched libtool, due to the different filename Aug 13 23:22:41 indeed Aug 13 23:29:23 does OE run autoreconf on all autotools packages ? Aug 13 23:34:25 yep Aug 13 23:34:30 unless you stop it :) Aug 13 23:34:31 unless they override do_configure, of course Aug 13 23:34:33 its just the default Aug 13 23:34:37 * Crofton hears an echo Aug 13 23:34:50 :) Aug 13 23:34:52 some peoples configure does not match the atuoconf .... Aug 13 23:35:13 'k great Aug 14 00:22:43 yep Aug 14 00:22:51 this patch looks awesome Aug 14 00:24:40 hi Aug 14 00:24:53 anyone know why gstreamer might not recognize v4l2src? Aug 14 00:25:01 I have gst-plugins-good and gst-plugins-bad installed **** ENDING LOGGING AT Sat Aug 14 02:59:57 2010