**** BEGIN LOGGING AT Wed Apr 20 02:59:58 2011 Apr 20 03:15:42 I think I got a bug on testlab class Apr 20 03:15:56 it seems it puts the package epoch in the filename Apr 20 03:16:00 on installed-packages Apr 20 03:16:05 opkg-cl ${IPKG_ARGS} info $pkg | grep -B 7 -A 7 "^Status.* \(\(installed\)\|\(unpacked\)\)" | awk '/^Package/ {printf $2"_"} /^Versio$ Apr 20 03:16:37 this is the line it does it. The filter for Version is only to print $2 and to not drop the epoch Apr 20 03:16:44 and so it sends it to the filename Apr 20 07:15:03 Hi. Will "bitbake -i" come to life again ?? Apr 20 07:17:46 quick answer is no :) Apr 20 07:18:07 long answer - if someone will reimplement it Apr 20 07:35:25 JaMa|W, thx for reviewing the meta-oe patches so fast ... I totally missed the PRs Apr 20 07:37:24 mlip: yw, thanks for working on meta-oe Apr 20 07:37:37 we really need more man power there :) Apr 20 07:38:10 gm ant Apr 20 07:38:20 moin Apr 20 07:38:52 heh...overnight build still failed because of schroedinger_1.0.9 :p Apr 20 07:39:16 missing int16_t Apr 20 07:39:23 JaMa|W, yeah , but the layer works quite well, I switched from org.oe.dev when I wanted to re-active my old gumstix, and realized that the provide git repo from gumstix is once again not in the best shape Apr 20 07:39:47 so i just created my own gumstix-layer and started using the layers ;) Apr 20 07:40:14 mlip: yes, works quite well, biggest problem now from my POV are missing recipes Apr 20 07:41:05 true Apr 20 07:41:17 so with more devs using meta-oe we get more maintainers for recipes Apr 20 07:41:22 I have about ~20 copied ones currently lying around waiting for reviews Apr 20 07:42:21 mlip oha Apr 20 07:42:38 jama hm Apr 20 07:44:15 woglinde: yesterday I noticed GNUtoo saying that he wont switch to meta-oe before there are recipes for games there :) I think this is bad sign.. Apr 20 07:44:45 woglinde: as it should be vice-versa.. switch to meta-oe soon, to migrate recipes for games if I care about games Apr 20 07:46:38 JaMa problem is now tsc Apr 20 07:46:44 we need a proper sign from it Apr 20 07:47:29 agreed Apr 20 07:50:04 woglinde: gm Apr 20 07:50:09 hi effem Apr 20 07:50:32 woglinde: wrt your question yesterday, my orc test was with neek which is nios2 and only has gcc 4.1.2 iirc Apr 20 07:50:41 so gcc version seems to play a role too Apr 20 07:50:43 effem native Apr 20 07:50:50 target wasn problem Apr 20 07:51:00 because there is nothing executed at build time Apr 20 07:51:16 effem the problem is orcc which parses orc files Apr 20 07:51:32 and when you have two dest commands in it Apr 20 07:51:42 woglinde: I did build both orc and orc-native Apr 20 07:51:46 the parsing into the internal structs goes beserk Apr 20 07:51:49 or should I have build a package that uses orc-native Apr 20 07:51:56 no Apr 20 07:51:58 yeah, read your message on it Apr 20 07:52:06 orc-native itself runs orc Apr 20 07:52:08 for examples Apr 20 07:52:11 and testsuites Apr 20 07:52:18 orcc Apr 20 07:52:36 actually the parsing is okay Apr 20 07:52:36 someone should please push the schroedinger updates Apr 20 07:52:46 besides the parser coe is guly Apr 20 07:52:49 ups ugly Apr 20 07:52:58 ant which one? Apr 20 07:53:04 1.09 -> 1.10 Apr 20 07:53:23 JaMa|W: maybe not all users are able to move over at the moment (e.g. because the distro they are using is not there, or the bsp) Apr 20 07:53:56 effem start making a bsp for your machine Apr 20 07:54:33 eFfeM_work: that's the point.. try to create such bsp layer or distro layer Apr 20 07:55:11 actually tried to add a bsp to yocto twice (sheevaplug and mpc8313ds iirc) in both cases got feedback that yocto people were also working on it/would be looking at it; not really sure on the best way to progress Apr 20 07:55:19 and my time is fairly limited too Apr 20 07:55:58 Hi all Apr 20 07:56:19 * eFfeM_work would like the tsc to provide some guidance on how we want to deal with bsp's distro's etc Apr 20 07:56:28 is there yocto documentation about how to add machine specific patches to a recipe? Apr 20 07:56:53 the only concrete examples of BSP are in Yocto iirc (RP work) Apr 20 07:57:06 mrAlmond: there is a doc Apr 20 07:57:23 ant you dont really need a bsp right? Apr 20 07:57:24 mrAlmond: http://www.yoctoproject.org/docs/bsp-guide/bsp-guide.html parts about .bbappend Apr 20 07:57:56 jama kernel recipe is still enugh? Apr 20 07:58:28 * woglinde looks at shr and angstroem again Apr 20 07:58:46 yes, about kernel... each bsp will add own files (and versions)? Apr 20 07:59:00 woglinde: ? Apr 20 07:59:57 oe are we supposed to adapt/use the linux-yocto kernel? Apr 20 08:00:02 *or Apr 20 08:00:15 woglinde: meta-shr is not completely cleaned from distro overrides and machine specific files (even has stuff which will be moved to separate bsp layers later) Apr 20 08:00:38 JaMa|W: Thank you!! Apr 20 08:01:11 jama ah thanks Apr 20 08:01:17 JaMa|W: for bsp's my first attempt was in nov: https://lists.yoctoproject.org/pipermail/yocto/2010-November/000183.html got the message from Bruce that there should be an official version in a week or so, but apparently it has not surfaced yet (or I cannot find it) Apr 20 08:03:08 hm Apr 20 08:03:16 I should start one for the simpad Apr 20 08:03:29 hm whats with opie? Apr 20 08:03:40 should opie be now on an own layer Apr 20 08:03:47 I will ask paul later Apr 20 08:04:03 .o(good he works for intel/linuxfoundation) Apr 20 08:04:23 hi florian Apr 20 08:04:26 hi florian_kc Apr 20 08:05:10 hi all Apr 20 08:05:24 eFfeM_work: seems like that thread is dead long enough.. I would try to ping Bruce if there are some news about it Apr 20 08:05:27 woglinde: another blocker are the overrides like e.g. _simpad in some recipes. I'm cleaning out several for Zaurus right now but unfortunately some cruft remains in oe-core Apr 20 08:05:59 I expect my layer to patch a 'vanilla' recipe, without overrides Apr 20 08:06:13 ant hm which overrides? Apr 20 08:06:23 seedirs in udev (my last commit) Apr 20 08:06:31 or initscripts Apr 20 08:06:33 JaMa|W: yeah probably will, actually had a 2nd attempt with kirkwood last jan or so, but didn't get too much feedback on that one either, so somewhat became discouraged Apr 20 08:07:06 woglinde: cleaning for Zaurus I've seen many overrides for ipaq and some for simpad Apr 20 08:07:13 hm Apr 20 08:07:17 let me see Apr 20 08:07:24 check my last commits :p Apr 20 08:07:30 I think I should remove the older simpad kernel anyway Apr 20 08:09:15 and all remaining 2.4 kernels too :) Apr 20 08:09:45 as agreed on ML half year ago :) Apr 20 08:09:59 * ant_work was thinking...let's take advantage of the current rather 'unmantained' situation and do soma massive purge for kernel/udev/... rercipes Apr 20 08:10:04 eFfeM_work: ^^^ Apr 20 08:10:36 silence Apr 20 08:10:38 :) Apr 20 08:11:06 we should put on OE site 'Don't use git master now!' :) Apr 20 08:11:08 ant_work: I have no problems with it, but I am not going to do it myself (but I happily will ack the patches) Apr 20 08:11:31 no plans to burn my fingers again on doing something that I myself do not need Apr 20 08:11:48 yesterday I was almost removing 90% of udev Apr 20 08:11:51 Jay7: why? for lots of stuff git master is still the best resource Apr 20 08:11:53 then I limited Apr 20 08:12:31 eFfeM_work: I mean ant_work's idea :) Apr 20 08:12:52 jay7 hm Apr 20 08:12:53 "we are working for you" Apr 20 08:12:56 please wait Apr 20 08:13:04 It seems like the maillist is realy slow. It has been several minutes since I sent and email to it. Anyone else have the same problem ? Apr 20 08:13:04 ah ok Apr 20 08:13:15 ant_work: yes, your text is better :) Apr 20 08:13:42 sgh: check ML archive Apr 20 08:13:58 Jay7: ok. Apr 20 08:14:08 sgh, half an hor ago it was super fast; like i hit the send button and 3 secs later it was on the ml Apr 20 08:15:05 JaMa|W: I'd remove linux-rp too... Apr 20 08:15:28 I haven't used it for few years, so I'm fine with it :) Apr 20 08:15:50 mlip: hmmmm ... I get all the emails from the list. Just not my own. Apr 20 08:18:37 gm mickeyl Apr 20 08:19:25 sgh: I see your email Apr 20 08:20:05 Jay7: yeah ...... 10 minutes of delay. That's bad I think. Apr 20 08:20:20 for LTG it's ok imho :) Apr 20 08:22:29 ant_work: What is happening with schroedinger - is a patch on its way ? Apr 20 08:25:23 yes, three patches iirc Apr 20 08:26:41 well, 5 http://patches.openembedded.org/project/oe/list/ 2429, 2433, 2435, 2437, 2439 Apr 20 08:26:58 I applied those while debugging Apr 20 08:27:12 I'll ack those Apr 20 08:32:56 great Apr 20 08:34:55 03Martin Jansa  07master * r6499b3d160 10openembedded.git/recipes/freesmartphone/ (fso-specs_git.bb libfso-glib_git.bb): Apr 20 08:34:55 fso-specs, libfso-glib: bump SRCREV and PV Apr 20 08:34:55 Signed-off-by: Martin Jansa Apr 20 08:35:03 03Martin Jansa  07master * r302bd03f2d 10openembedded.git/recipes/freesmartphone/cornucopia.inc: Apr 20 08:35:03 cornucopia: bump SRCREV Apr 20 08:35:03 Signed-off-by: Martin Jansa Apr 20 08:35:07 03Martin Jansa  07master * rb1f08c510a 10openembedded.git/recipes/freesmartphone/libgisi_git.bb: Apr 20 08:35:07 libgisi: bump SRCREV Apr 20 08:35:07 Signed-off-by: Martin Jansa Apr 20 08:35:09 03Martin Jansa  07master * r5334e565e3 10openembedded.git/recipes/freesmartphone/msmcomm.inc: Apr 20 08:35:09 msmcomm: bump SRCREV Apr 20 08:35:09 Signed-off-by: Martin Jansa Apr 20 08:35:18 03Martin Jansa  07master * raffb144ca7 10openembedded.git/recipes/freesmartphone/frameworkd_git.bb: Apr 20 08:35:18 frameworkd: bump SRCREV Apr 20 08:35:18 Signed-off-by: Martin Jansa Apr 20 08:35:25 03Martin Jansa  07master * r872bf75c59 10openembedded.git/recipes/freesmartphone/msmcomm.inc: Apr 20 08:35:25 msmcomm: add LIC_FILES_CHKSUM and sync with meta-shr Apr 20 08:35:25 Signed-off-by: Martin Jansa Apr 20 08:46:29 morning everyone Apr 20 08:47:15 morning bluelightning Apr 20 08:47:28 hello bluelightning Apr 20 08:47:41 hi bluelightning Apr 20 08:48:04 hi bluelightning Apr 20 08:48:15 bluelightning will opei become its own layer? Apr 20 08:48:20 ups opie Apr 20 08:48:30 ant_work: shroedinger still fails to build for me, even with 2429, 2433, 2435, 2437, 2439 applied Apr 20 08:48:30 woglinde: I've been thinking about that recently Apr 20 08:48:56 one of my concerns is all of the recipe history will be lost Apr 20 08:48:56 jama dev or core? Apr 20 08:49:04 dev Apr 20 08:49:12 then again I guess one can just go back to oe.dev and look there Apr 20 08:50:07 JaMa|W: interesting..I built it with older orc but hacking staged header Apr 20 08:50:13 but opie in its own layer makes a lot of sense, I'm sure having all of that stuff filling up people's recipe dir is annoying :) Apr 20 08:50:44 JaMa|W: and build failed overnight (no extra patches) Apr 20 08:50:45 bluelightning good I would happly testing it Apr 20 08:50:58 ant_work, woglinde: this was build from scratch with new orc and later with those patches from patchwork applied Apr 20 08:51:25 ok, fwiw builfd from scratch fails with schroedinger_1.0.9 too Apr 20 08:51:29 trying here to build xbmc Apr 20 08:51:31 woglinde: I have a skeleton layer set up already... will let you know when I have something to look at Apr 20 08:51:46 but it will last a while to schroedinger now Apr 20 08:52:04 ant_work, woglinde: here is log http://paste.pocoo.org/show/374964/ Apr 20 08:52:24 thi sis th e'old' error Apr 20 08:52:27 omg Apr 20 08:52:37 this is the 'old' error Apr 20 08:53:51 jama shure you have orc-0.4.13? Apr 20 08:55:35 yes.. but looks like #ifndef _ORC_INTEGER_TYPEDEFS_ is false and stdint is not #included Apr 20 08:56:02 just got it building after moving that include outside this ifndef Apr 20 08:56:50 they have abandoned stdin types in favor of own orc-types Apr 20 08:56:57 but something is still wrong... Apr 20 08:56:58 just checked orc-native-0.4.13-r0.do_build orc-0.4.13-r0.do_build Apr 20 08:57:23 I am at orc now Apr 20 09:00:58 who maintains oe-core ML, I think that GPG signed content-types should be enabled there too "Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="cvVnyQ+4j833TQvp"" Content-Disposition: inline Apr 20 09:01:16 I always forget to disable default sign when posting something there :/ Apr 20 09:24:57 okay I hitting the schroedinger bugs now too Apr 20 09:26:05 woglinde: maybe that second condition about STDC is false? Apr 20 09:26:22 woglinde: as I have checked orc headers and they all use same ifdefs Apr 20 09:26:46 but I didnt apply the patches from patchwork yet Apr 20 09:28:22 the problem is Apr 20 09:28:38 schro.c dont include schro.h Apr 20 09:28:39 hm hm Apr 20 09:44:47 jama there is going somethind odd on Apr 20 09:45:02 the orc variant should have the orc types Apr 20 09:45:14 and the .c file should include the header Apr 20 09:45:31 instead of orc.h Apr 20 09:45:33 hm Apr 20 09:46:42 i have the strange feeling regenarating for the files isnt enabled Apr 20 09:47:19 but I will apply the patches now Apr 20 10:01:04 hi gnutoo Apr 20 10:01:09 hi Apr 20 10:02:34 woglinde, do you have any idea for debugging emacs-x11 compilation(qemu segfaults) Apr 20 10:04:34 hello Apr 20 10:04:54 can anyone help me how to bypass kernel compilation in oe... Apr 20 10:10:19 03Denis Carikli  07org.openembedded.dev * r394f1a819f 10openembedded.git/recipes/xserver-nodm-init/xserver-nodm-init_2.0.bb: (log message trimmed) Apr 20 10:10:19 xserver-nodm-init 2.0: remove DEFAULT_PREFERENCE = "-1" Apr 20 10:10:19 That fix was made in order to prevent distributions like Angstrom2010 Apr 20 10:10:19 from picking the 1.0 version: Apr 20 10:10:19 Angstrom 2008 picked the 2.0 version while Angstrom 2010 use the 1.0 version. Apr 20 10:10:19 That can prevent Xorg session from starting on the target device(like bug20 in my case): Apr 20 10:10:20 The 1.0 lacks the following session start: Apr 20 11:32:22 okay stupid schroedinger devs Apr 20 11:32:34 we have to rengerate some files with recent orcc Apr 20 11:36:09 ah, that was the issue Apr 20 11:37:02 he he, seems fixes are in master but not in 1.10 Apr 20 11:37:07 yes the dont regnerate from schroo.orc Apr 20 11:37:22 instead coping preconverted files Apr 20 11:37:32 will fix it this evening Apr 20 11:37:42 wow, that was nasty :/ Apr 20 11:38:01 all coming from bluez, isn't? Apr 20 11:40:21 no? Apr 20 11:40:25 xbmc needs it Apr 20 11:40:28 as gst-ti as well Apr 20 11:44:09 why does my poor Zaurus buid that? :P Apr 20 11:44:28 ant_work: I'm experimenting with openwrt init scripts (+ their additional stuff) when I've time Apr 20 11:44:37 I like it a lot more for my use cases Apr 20 11:44:53 there is space for improvements Apr 20 11:44:57 they've hook based event system Apr 20 11:45:06 so you can hook there and there if you need to Apr 20 11:45:17 and you don't actually modify the base scripts Apr 20 11:46:05 we have three knobs: IMAGE_INIT_MANAGER, IMAGE_INITSCRIPTS, IMAGE_DEV_MANAGER Apr 20 11:46:29 yep, I'm not that far yet :) Apr 20 11:46:57 we *already* have that Apr 20 11:47:14 this is just for the integration Apr 20 11:47:53 ant_work your zaurus has pcmcia? Apr 20 11:48:50 yes Apr 20 11:57:00 03Koen Kooi  07org.openembedded.dev * rf1d5702637 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3.inc libgles-omap3/rc.pvr): Apr 20 11:57:00 libgles-omap3: abstract ES handling and share it across all targets Apr 20 11:57:00 Signed-off-by: Koen Kooi Apr 20 11:57:46 woglinde: fwiw those could be reverse deps of schroedinger http://gentoo-portage.com/media-libs/schroedinger/RDep#ptabs Apr 20 12:01:11 I have a newbie question about prebuild toolchain use in OE. I seem from local.conf.sample that one needs to tell OE that libc and gcc are already provided, that seems fair enough. Would it be possible though for me to write a recipe that unpacks my toolchain tarball and have OE do that as a first port of call? If so, how would I depend on that? Apr 20 12:01:23 ant_work -> pcmcia-> bluez->bluezsound->gstreamer->schroedinger Apr 20 12:01:29 ant thats like usb Apr 20 12:01:31 ^_^ Apr 20 12:01:38 and known and discusses serval times Apr 20 12:01:52 it's a new addition, though Apr 20 12:02:19 no? Apr 20 12:02:30 I have never seen it failing before Apr 20 12:02:49 started last week Apr 20 12:03:50 because 0.4.12 was added Apr 20 12:03:54 orc ^ Apr 20 12:05:27 JaMa: did you see my comment about testlab class? Apr 20 12:05:35 no Apr 20 12:05:47 JaMa: it seems to be putting version epoch on the package filename Apr 20 12:06:00 JaMa: any guess why? Apr 20 12:06:15 testlab? I thought it's normal Apr 20 12:06:34 JaMa: the generated file has no epoch on it Apr 20 12:14:28 PN_PEah I have testlab enabled and don't see this here Apr 20 12:15:58 but you're right I think I have seen it before (long time ago..) Apr 20 12:16:43 otavio: isn't it just debian.bbclass working right for you? are you sure it's testlab? Apr 20 12:27:40 JaMa: dunno ... the problem is it breaks my build since we check the logs ;-) Apr 20 12:28:15 can you show how exactly does it break logs? Apr 20 12:28:34 s/logs/build/ Apr 20 12:30:48 I bet it's debian naming Apr 20 13:01:32 JaMa: debian? but debian changes the package name but not version, does it? Apr 20 13:01:55 JaMa: sure. It fails to check the package size since the file does not exist Apr 20 13:02:08 I have no idea what error you see.. Apr 20 13:02:58 + for file in '$(cat ${TESTLAB_DIR}/installed-packages.txt)' Apr 20 13:02:59 + head -n1 Apr 20 13:02:59 ++ find /home/otavio/hacking/ossystems/oe/tmp/deploy/glibc/ipk -name xserver-xorg-xephyr_2:1.5.3-r4.oss.1.1_i686.ipk Apr 20 13:03:01 + du -k Apr 20 13:03:04 du: write error Apr 20 13:04:43 and testlab does: | awk ... /Version/ {printf $2"_"} ... Apr 20 13:07:03 good, then drop improve this awk to ignore .*: Apr 20 13:07:09 -drop Apr 20 13:07:38 JaMa: but I doesn't get why I am the only one seeing it Apr 20 13:07:43 JaMa: it most be something else Apr 20 13:08:14 and, I don't know how to change the awk to ignore it heheh Apr 20 13:11:26 ATAL: Error when parsing []: file classes/angstrom.bbclass not found in did i just fuck up the paths ? Apr 20 13:21:01 otavio: grep -B 7 -A 7 "^Status.* \(\(installed\)\|\(unpacked\)\)" | awk '/^Package/ {printf $2"_"} /^Version/ {sub(/.*:/,"", $2); printf $2} /^Archi/ {print $2".ipk"}' works here Apr 20 13:21:25 JaMa: do you have the epoch on the filename? Apr 20 13:22:17 JaMa: you missed the "_" Apr 20 13:22:36 See? ... /^Version/ {sub(/.*:/,"", $2"_") Apr 20 13:23:20 sorry fixed it only locally Apr 20 13:24:53 sending to list Apr 20 13:25:01 JaMa: hold Apr 20 13:25:04 it is wrong Apr 20 13:25:35 ... /^Version/ {sub(/.*:/,"", $2); print $2"_"} Apr 20 13:25:46 seems to be the correct one Apr 20 13:25:48 what's wrong? Apr 20 13:25:56 that's what I have Apr 20 13:26:03 * JaMa doesn't see difference Apr 20 13:26:37 see this http://paste.pocoo.org/show/375111/ Apr 20 13:28:29 JaMa: good; seems fine Apr 20 14:21:55 I have a toolchain which has binaries named like: "mipsel-linux-uclibc-gcc". I'm tring to use it with OE by setting ASSUME_PROVIDED for libc and gcc in my local.conf. But at build time OE is tring to use "mips-oe-linux-uclibc-gcc". Aside from the incorrect endianness, I'd like to drop the "-oe" part of the name. Any idea how? Apr 20 14:26:22 why am i missing classes/angstrom.bbclass but oe is looking for it ? Apr 20 14:27:39 rob_w: you have combined old and new OE Apr 20 14:27:51 rob_w: I renamed that class a few weeks ago Apr 20 14:30:13 ok so then my error is in my overlay Apr 20 14:30:15 thx Apr 20 14:31:21 change to angstrom-blacklist.bbclass Apr 20 14:39:10 ^^ to answer my own question, "-oe" in the t/c name is set by TARGET_VENDOR. In my case, the micro distro included sane-toolchain.inc, which sets TARGET_VENDOR="-oe" Apr 20 14:39:37 thhp isnt that documented in local.conf? Apr 20 14:40:47 not in the version I have from release-2011.03 -- possibly upstream Apr 20 15:09:05 hi CosmicPenguin Apr 20 15:09:21 yello Apr 20 15:30:17 [balister@octopus setup-scripts (master)]$ bitbake console-gnuradio-image Apr 20 15:30:18 ERROR: Execution of 'git describe --tags' failed: command not foundTA: Apr 20 15:30:24 anyone got a clue why that happens? Apr 20 15:33:04 Next external toolchain problem -- my linker doesn't support --sysroot. Is there any way to avoid the use of --sysroot in OE? From a little poke around it seems like the build depends on it a lot... Apr 20 15:34:24 thhp dont think so Apr 20 15:34:46 woglinde darn... Apr 20 15:35:08 thhp ypu could try 2008 release Apr 20 15:35:29 thhp cannt you build a mips toolchain with current? Apr 20 15:35:45 thhp: LIBTOOL_HAS_SYSROOT = "yes" Apr 20 15:35:50 override it to no Apr 20 15:36:09 JaMa is it well tested? Apr 20 15:36:11 woglinde I /have/ to use a vendor-supplied toolchain Apr 20 15:36:18 ah, it's not only about libtool Apr 20 15:36:24 but linker.. Apr 20 15:37:04 JaMa: sounds promising, but presumably I'd have to override XYZ_HAS_SYSROOT for lots of packages? Apr 20 15:37:21 it's one global flag set in sane-toolchain.inc Apr 20 15:37:36 jama sure Apr 20 15:38:30 woglinde: is there a git tag for the 2008 release? I don't see one Apr 20 15:40:43 thhp try first what jama says Apr 20 15:40:48 args Apr 20 15:40:49 damn Apr 20 15:41:01 woglinde: will do Apr 20 15:41:36 http://cgit.openembedded.org/cgit.cgi/openembedded/log/?h=org.openembedded.stable Apr 20 15:42:04 so much to do to get mpd in meta-oe *sigh* Apr 20 15:42:29 mlip you will get through it Apr 20 15:43:09 i am wondering that things work in org.oe.dev, such much flags in mpd that are not respected anymore ;D Apr 20 15:43:57 I find that crda is currently broken - couple of issues 1) some libnl changes that break it, 2) it uses '-t' param for install and apparently we have a bin/install in oe now which doesn't support that Apr 20 15:44:21 any reason oe/bin/install doesn't support -t or any simple way to use system install? Apr 20 15:51:26 woglinde: Just tried overriding LIBTOOL_HAS_SYSROOT, but to no avail. bitbake.conf has "TOOLCHAIN_OPTIONS = " --sysroot=${STAGING_DIR_TARGET}"" which will kill the linker all the same :-( Apr 20 15:51:53 then try the stable branch Apr 20 15:52:01 but its some outdated now Apr 20 15:52:50 woglinde: thanks, I'll take a look at that Apr 20 15:53:24 and externaltoolchain support is a bit diffrent Apr 20 15:54:21 hm I guess nobody tought about sysroot Apr 20 15:54:27 and externaltoolchain Apr 20 15:54:35 woglinde: any tips on how different would be most appreciated ;-) Apr 20 15:54:45 no sorry Apr 20 15:54:54 the wiki page maybee helps Apr 20 15:54:54 woglinde: np Apr 20 15:55:01 becaus its not updated Apr 20 15:55:06 I'll have a dig Apr 20 15:55:12 or use the history on it Apr 20 15:58:26 woglinde, i've fixed the fts vs lfs breakage in uClibc now Apr 20 16:00:06 blindvt cool Apr 20 16:00:18 unfornatly I will not find the time over easter holdiay Apr 20 16:01:11 but systemd and elfutils should really switch to the POSIX equivalent. I do not remember to have seen fts as a requirement in elfutils though so this argument sounds a bit fishy ;) Apr 20 16:01:25 woglinde, take your time Apr 20 16:02:04 blindvt hm I dont think lennart will listen Apr 20 16:24:19 woglinde, i have a bug report i need ya to look at to see if there is anything else i should look at (fixing the xbmc libtool issue) Apr 20 16:24:42 hm Apr 20 16:24:46 have a nice around-Easter-week Apr 20 16:24:48 i filed a bug report on the xbmc tracker, to help with starting the fixes: http://trac.xbmc.org/ticket/11459 Apr 20 16:24:49 later maybee Apr 20 16:25:14 dandel fixes for them isnt that easy Apr 20 16:25:25 they need the 2.4 libtool m4 macros Apr 20 16:25:35 but that may break some dsitros Apr 20 16:25:38 i figured that :/ Apr 20 16:26:16 also, the xbmc requires libtool, autoreconf and automake to be installed to even compile xbmc. Apr 20 16:26:26 dandel its the balance btw shipping a working configure or not Apr 20 16:26:51 but when most guys who runs configure have autofoo installed anyway Apr 20 16:27:15 so I would not ship the the external macros anymore Apr 20 16:27:31 okay have to got now Apr 20 16:27:33 till later maybe Apr 20 16:27:34 e Apr 20 16:41:15 03Martin Jansa  07master * r284ae563ef 10openembedded.git/classes/testlab.bbclass: Apr 20 16:41:15 testlab: drop PE: from package version when looking for package filename Apr 20 16:41:15 * reported by Otavio Salvador Apr 20 16:41:15 Signed-off-by: Martin Jansa Apr 20 16:41:15 Acked-by: Koen Kooi Apr 20 17:45:10 can a recipe inherit from two recipes? Apr 20 17:47:30 kergoth: can a recipe inherit from multiple recipes? Apr 20 17:47:50 Nedlinpopo: yes Apr 20 17:47:54 inherit is for classes, not recipes, and isn't limited, no Apr 20 17:47:55 and also, how is inherit different from require Apr 20 17:48:06 example: http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/mysql/mysql5_5.1.40.inc Apr 20 17:48:41 kergoth: good distinction, I slightly jumped the gun on that ;) Apr 20 17:48:49 inherit is a require that adjusts the path to get classes/.bbclass, and avoids reparsing the same class twice, and supports the EXPORT_FUNCTIONS "magic" Apr 20 17:49:38 hmm, the recipes/image/x11-image.bb has an inherit image Apr 20 17:49:45 yes, it does Apr 20 17:50:02 as it should. image.bbclass does the things you generally want for an image Apr 20 17:50:04 if i wnat to extend that for my image i shoudl just require it right? Apr 20 17:50:13 what do you mean? Apr 20 17:50:35 well, I want to build a custom filesystem with x11-image plush some of my local packages Apr 20 17:50:42 s/plush/plus Apr 20 17:50:42 right.. so do it Apr 20 17:50:54 is that an inherit or a require? Apr 20 17:50:54 make an image recipe that inherits image and do whatever you want Apr 20 17:51:02 you can do absolutely everything in a recipe you can in a class. Apr 20 17:51:11 if you're loading a class, you use inherit, always, period. Apr 20 17:51:21 if you're loading another recipe from yours, then you use require Apr 20 17:51:57 okay, so it makes more sense to require x11-image and then add to IMAGE_INSTALL right? Apr 20 17:52:05 (i think this is a style point) Apr 20 17:55:03 Nedlinpopo: that's common. See http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/images/angstrom-gnome-image-big.bb Apr 20 17:55:41 okay, i think that's the sensible thing to do here Apr 20 17:55:43 thanks Apr 20 17:55:50 np Apr 20 17:59:06 kergoth_: Hi, for some reason all files in image built from scratch have same UID/GID as they had on buildhost, it there some known problem with old oe.dev, fakeroot and bitbake oe/master? Apr 20 18:00:05 kergoth_: ie target: -rw-r--r-- 1 1000 1000 1.0K Apr 16 07:10 /etc/X11/xorg.conf and 1000 is UID and GID of bitbake user on buildhost Apr 20 18:00:45 JaMa|Off: is that true in the tbz2 as well? Apr 20 18:01:38 this is from ubi, but tar.gz has same issue Apr 20 18:02:15 what's ubi? Apr 20 18:02:47 JaMa|Off: yes, you need to either update to newer oe or set FAKEROOT=fakeroot Apr 20 18:04:03 ah, the Unsorted Block Image file system Apr 20 18:04:24 kergoth: seen that commit, but usually I do update both (bitbake + oe) at once... so I'm surprised that I would get only newer bitbake Apr 20 18:04:41 well, don't know what to tell you. it was fixed in oe weeks ago Apr 20 18:04:57 don't know of any issues remaining Apr 20 18:05:07 kergoth: there was only 13 minutes in between so maybe I was so unlucky :/ Apr 20 18:15:24 kergoth: is it used only when building image or already while preparing .ipk files? Apr 20 18:26:26 JaMa|Off: it's used i packaging as well as image creation, i'd suggest starting from scratch Apr 20 18:26:44 ok ;/ Apr 20 18:27:05 will double check if it's really working Apr 20 18:27:11 but now gtg Apr 20 18:38:51 kergoth, I find that crda recipe uses '-t' option for install and when you added bin/install to avoid reliance on coreutils-native it doesn't support that option Apr 20 18:39:09 this can't be the first recipe that uses the '-t' parm of install? Apr 20 18:40:24 what the fucking hell, orcc still dies here even with gcc-4.4 Apr 20 18:50:32 he ka6sox Apr 20 18:50:59 hiya woglinde Apr 20 18:51:01 how goes it Apr 20 18:51:46 fighting with orc Apr 20 18:51:51 and schroedinger Apr 20 18:57:36 what's the trcik to answer yes to a question from a python configure script? Apr 20 18:57:59 Crofton look at the ti recipes Apr 20 18:59:05 hi ensc Apr 20 19:01:24 bah, just patch the script Apr 20 19:01:41 hihi Apr 20 19:02:10 03Henning Heinold  07org.openembedded.dev * rddc04b7af7 10openembedded.git/recipes/orc/ (7 files): Apr 20 19:02:10 orc: DP=-1 for 0.4.13 and some fixups Apr 20 19:02:10 * use DEFAULT_PREFERENCE=-1 for 0.4.13 because its breaks schroedinger Apr 20 19:02:10 and segfaults on debian unstable Apr 20 19:02:10 * remove unpinned versions Apr 20 19:02:10 * switched ot INC_PR Apr 20 19:03:00 Apr 20 19:26:50 what var points into the sysroot? Apr 20 19:28:41 SYSROOT_DESTDIR ? Apr 20 19:29:13 STAGINGDIR_HOST Apr 20 19:32:50 are you thinking of STAGING_DIR_HOST? Apr 20 19:33:02 or STAGING_DIR_TARGET Apr 20 19:33:33 kergoth, did you see my question about bin/install? Apr 20 19:50:21 tharvey: http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/crda/crda_1.1.1.bb Apr 20 19:50:24 tharvey: where's -t? Apr 20 19:51:04 foerster, in crda's makefile Apr 20 19:51:18 oe_runmake Apr 20 19:51:39 tharvey: ah, you said in the recipe which is why I was scratching my head :) Apr 20 19:51:42 I can submit a patch to maillist installing things manually instead of calling crda's makefile as is currently done Apr 20 19:51:48 sorry Apr 20 19:52:20 never seen the '-t' param used for install myself, seems to just switch position of source/dest params really, at least in the way its used by crda Apr 20 19:52:46 yea, from man page, that's all it seems. Apr 20 19:53:12 was just surprised it wasn't encountered elsewhere and wondering if bin/install should have support added for it Apr 20 19:53:49 crda has other issues though - it needs some EXTRA_CFLAGS/EXTRA_LDFLAGS to deal with libnl - not clear why that has changed really Apr 20 19:54:21 well, you'd want to determine how portable -t is before considering adding it to the wrapper, I think Apr 20 19:54:24 also not clear if the change shoudl use libnl or libnl2 - looks like there has been a move to libnl2 for a lot of things Apr 20 19:54:54 kergoth, I can't really answer that - I don't have an opinion either way. I'm happy to install things within the recipe as a fix Apr 20 19:55:18 I'm really surprised though to not see anyone else mention crda being broken - how are people using mac80211 drivers without it? Apr 20 19:55:40 a patch is likely cleaner than duplicating the install bits, unless there are very few things to install Apr 20 19:57:50 well, if a patch re-writes all the install lines that use -t I think it would just be clearer to do that in the recipe Apr 20 20:00:29 it duplicates the command on every line? Apr 20 20:00:33 what idiots Apr 20 20:00:52 never heard of makefile variables, apparently :) Apr 20 20:05:21 03Stanislav Brabec  07master * r035d0d6d05 10openembedded.git/classes/binconfig.bbclass: (log message trimmed) Apr 20 20:05:21 binconfig.bbclass: Search in installed root instead of S: Apr 20 20:05:21 * It is more consistent. Apr 20 20:05:21 * Fixes subtle binconfig breakages: Apr 20 20:05:21 * binconfig were populated with a different name that upstream intended Apr 20 20:05:21 * populated binconfig files that were not intended to be populated Apr 20 20:05:21 * See http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-April/031972.html Apr 20 20:05:22 03Stanislav Brabec  07master * r06b3943add 10openembedded.git/recipes/ncurses/ (ncurses.inc ncurses_5.7.bb ncurses_5.9.bb): Apr 20 20:05:23 ncurses: Increment PR to trigger rebuild and get correct binconfig files. Apr 20 20:05:23 Signed-off-by: Stanislav Brabec Apr 20 20:05:24 03Stanislav Brabec  07master * r38058be0f8 10openembedded.git/recipes/tcltk/ (5 files in 3 dirs): Apr 20 20:05:24 tk: Remove old version 8.4.19. Apr 20 20:05:25 Signed-off-by: Stanislav Brabec Apr 20 20:05:26 03Stanislav Brabec  07master * r051e8b214d 10openembedded.git/recipes/tcltk/tk_8.5.8.bb: Apr 20 20:05:37 03Stanislav Brabec  07master * r28439c85d7 10openembedded.git/recipes/tcltk/ (4 files in 3 dirs): Apr 20 20:05:37 tcl: Remove old version 8.4.19. Apr 20 20:05:37 Signed-off-by: Stanislav Brabec Apr 20 20:05:47 kergoth_, I may be misunderstanding you. crda does not use autoconf, they have a makefile with an 'install' target, and each use of install (3x) they use the -t option Apr 20 20:06:49 so I would need to completely replace those three lines in their makefile with removal of -t and swap source/dest - personally I would rather just write my own install for the 3 objects in the recipe Apr 20 20:10:19 I didn't say anything about autoconf Apr 20 20:10:52 but fair enough, only 3 lines its not worth the hassle Apr 20 21:18:05 is there anyway to parallelize the do_rootfs step? Apr 20 21:19:09 why? Apr 20 21:19:23 its iocaped Apr 20 21:19:31 and its normaly fast Apr 20 21:19:52 hi risca Apr 20 21:20:00 hi woglinde Apr 20 21:23:57 woglinde: it's taking on the order of 10 minutes, and I pay that penalty every time i redeploy Apr 20 21:24:26 and I think maybe opkg-cl is actually cpubound for a lot of the time Apr 20 21:24:29 Nedlinpopo whats the size of the rootfs at your side? Apr 20 21:24:40 hangon, let me check Apr 20 21:24:41 4 gig or what? Apr 20 21:26:32 only 21MB as a tar.bz2 Apr 20 21:26:55 ~35M uncompressed Apr 20 21:28:24 ugh Apr 20 21:28:36 there is clearly something wrong Apr 20 21:28:45 which is smaller than I would've expected Apr 20 21:28:48 that shouldnt last 2 min for all Apr 20 21:28:55 compressions Apr 20 21:30:08 well do_build is fast, but do_rootfs seems to take a while Apr 20 21:30:44 strange Apr 20 21:30:52 Nedlinpopo it should be really faster Apr 20 21:30:58 not 10 min Apr 20 21:31:06 let me make a sd really quick and I'll time it Apr 20 21:34:02 ugh, writing to this SD card is *really* slow. Apr 20 21:34:13 * Nedlinpopo can't wait for the fast ones to arrive in the mail Apr 20 21:37:03 it's timing Apr 20 21:37:04 ... Apr 20 21:37:13 ? Apr 20 21:37:24 what you are talking now about? Apr 20 21:37:35 get a proper class6 or 10 Apr 20 21:37:41 I just did a `time bitbake x11-image` Apr 20 21:37:45 the class10 is on order Apr 20 21:38:09 shoudl arrive this week I think Apr 20 21:38:49 I'm timing how long the bitbake is taking, so as to have a real measurement Apr 20 21:42:44 all the packages are already compiled, this is just 3 steps: do_package_update_index_ipk, do_rootfs, do_ Apr 20 21:43:07 yes Apr 20 21:44:31 yeah, it's taking a long time Apr 20 21:45:05 currently 24% of one core is used by sh -e Apr 20 21:46:13 and it's the process running the do_rootfs temporary shell script Apr 20 21:48:09 yeah, 12 minutes26 seconds Apr 20 21:48:43 of which 6:31 is in userland, and only 2:49 is in sys Apr 20 21:55:15 I really wonder Apr 20 21:56:18 I wonder if there's something i've done that's stupid? Apr 20 21:57:01 cannt say Apr 20 21:57:08 only that it should last this long Apr 20 21:57:19 hm which branch are you using Apr 20 21:57:21 oe.dev? Apr 20 21:57:26 i think si Apr 20 21:57:28 er, so Apr 20 21:57:39 let me double check Apr 20 21:58:35 i'm tracking dev.openembedded.org/HEAD Apr 20 21:59:37 and I pulled yesterday or the day before Apr 20 21:59:41 also bitbake is 1.12 Apr 20 22:01:24 fyi, org.openembedded.dev is just a compatibility symbolic ref pointing at master. master is it Apr 20 22:02:29 isn't dev.[oe].org the central repo you guys use? Apr 20 22:05:30 what do you mean? Apr 20 22:05:45 the *repository* is openembedded on git.openembedded.org, and the main development branch is master Apr 20 22:05:54 that's where most of the community puts it's recipes, etc Apr 20 22:05:59 yes, that's what i mean Apr 20 22:06:12 there is no dev.oe.org :) Apr 20 22:06:22 i just wanted to be sure that git.oe.org was teh correct *repository* Apr 20 22:06:44 * Nedlinpopo got lazy typing openembedded Apr 20 22:07:23 i was referring to the nonexistent dev subdomain, not the shortening for convenience ;) Apr 20 22:07:36 Nedlinpopo: *repository* != *branch* Apr 20 22:07:43 i know Apr 20 22:08:38 i use git a fair bit. but as you also know, the repository i snot really special, and you may ave seen that pixhawk and gumstix clone the oepnembedded repo with some more latency Apr 20 22:09:09 I just wanted to make sure I cloned the most current one. Apr 20 22:09:15 there are plenty of clones that aren't the official upstream, yes. i keep my personal ones on github Apr 20 22:09:30 * JaMa on gitorious :) Apr 20 22:15:53 kergoth: I have a cumbersome question... how can I find the initial IMAGE_BASENAME from within a package if I have built an initramfs image? I get the latter :/ Apr 20 22:17:14 i.e. echoing LABEL=${DISTRO}--${IMAGE_BASENAME}-${MACHINE} Apr 20 22:18:29 ant__: may be just use image full name? Apr 20 22:18:41 that one as in deploy dir Apr 20 22:18:47 it should be somewhere in bb env Apr 20 22:19:15 it is in kernel.bbclass Apr 20 22:19:16 image = bb.data.getVar('INITRAMFS_IMAGE', d, True) Apr 20 22:19:18 like minimal-console-image-eglibc-ipk-dev-snapshot-20110420-efikamx Apr 20 22:20:40 is for kexecboot.cfg ;) Apr 20 22:20:47 I know ;) Apr 20 22:20:57 LABEL was easy to understand :) Apr 20 22:21:03 he he Apr 20 22:21:56 btw, good idea to place boot.cfg to deploy dir too Apr 20 22:23:24 hm.. ${IMAGE_NAME} sounds too easy... Apr 20 22:23:47 if this works I must have been drunk Apr 20 22:23:59 let see Apr 20 23:03:35 JaMa, I'm fighting libnl vs libnl2 issues as you describe in http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-February/029495.html - what 'manual' steps did you do the clean libnl and linbnl2 that bitbake -cclean libnl libnl2 doesn't do? Apr 20 23:04:11 tharvey: remove all shlibs, pkgdata Apr 20 23:04:43 tharvey: and iirc also manual sysroot cleanup Apr 20 23:06:05 hmm... yes, bitbake clean doesn't seem to unstage them from sysroots Apr 20 23:06:43 so if I understand correctly angstrom blacklists now anything that depends on libnl1 to avoid such contention issues? Apr 20 23:07:04 tharvey: no.. see other threads about libnl Apr 20 23:08:38 you mean the set of patches for recipes that were building with static libnl? Apr 20 23:09:29 yes libnl is now libnl-2 and libnl-1 is built only static lib as libnl1 Apr 20 23:09:35 so no need to blacklist anything Apr 20 23:10:49 has this been more or less done on all recipes? I'm still seeing both libnl-1 and libnl-2 as dependencies in release-2011-03 Apr 20 23:11:21 still thrashing about trying to get a runable crda - I cant be the only one using wifi on release-2011-03? Apr 20 23:11:46 I don't use release-2011-03 but iirc those patches were pulled to it already Apr 20 23:12:00 but ie nmap seems broken in both branches Apr 20 23:13:25 ldd shows both libnl.so.1 and libnl.so.2 used by my crda Apr 20 23:14:10 what a mess Apr 20 23:16:57 tharvey: then libnl-1 wasn't fully removed before rebuilding crda, I guess Apr 20 23:17:15 looks like at least psmisc is using libnl from my deps Apr 20 23:18:08 in package-depends.dot I see pstree-static -> libnl - does this mean pstree has been built with static libnl? Apr 20 23:18:33 no static libnl is named libnl1 Apr 20 23:20:38 I'm currently on release-2011.03 - I need to be on 2011.03-maintenance to pick up these changes don't I? Apr 20 23:21:20 ah sorry yes those changes were only in 2011.03-maintenance Apr 20 23:21:36 I didn't notice that you're talking about release not maintenance Apr 20 23:21:40 crap... Apr 20 23:22:00 ok... another 10 hour build cycle heh Apr 20 23:22:04 going to bed Apr 20 23:22:33 I think you can just switch to that branch without rebuild from scratch Apr 20 23:22:46 but it's not bad idea wrt your libnl issues Apr 21 00:06:45 what does populate_packages_prepend do? Apr 21 00:07:23 prepends to populate_packages. Apr 21 00:07:32 ah Apr 21 00:07:34 okay Apr 21 00:07:51 so.. exactly what it says.. Apr 21 00:07:55 :) Apr 21 00:08:55 I wasn't sure if it was a thing that was done, or a thing that was referenced in th ebuild process Apr 21 00:09:12 _prepend and _append aren't part of the variable name Apr 21 00:09:15 they're operations Apr 21 00:09:37 okay Apr 21 00:09:39 lame syntactically, i know, but a bit late to do much about it now Apr 21 00:09:44 heh Apr 21 00:09:51 that ship has indeed sailed Apr 21 00:09:59 do you use opencv? Apr 21 00:10:10 it also behaves differently than .=/=./+=/=+ Apr 21 00:10:20 _append/_prepend are lazy / postponed Apr 21 00:10:25 they get applied at the end of the parse process Apr 21 00:12:44 is pkgconfig a program unto itself Apr 21 00:12:48 ? Apr 21 00:25:48 google is your friend Apr 21 00:32:09 what I'm really asking is if pkgconfig and pkg-config are the same thing. Apr 21 00:32:37 there is a recipe for pkgconfig, but not pkg-config, and pkg-config is used by opencv Apr 21 00:35:49 pkgconfig is pkg-config Apr 21 01:01:16 anybody know who is koen kooi? Apr 21 01:26:15 Nedlinpopo: yes, he's koen kooi. works for TI, one of the main oe developers, maintainer of angstrom, koen on irc. he doesn't hang out in here anymore, but he's on the beagle and angstrom channels. Apr 21 01:26:27 okay Apr 21 01:27:24 i'm having troubl ewith one of his recipes Apr 21 01:27:34 i'll send him an email Apr 21 01:27:57 thanks Apr 21 01:28:40 i'd suggest hitting the oe list first. the question may well be one that someone else could answer. our maintainership is awfully rough around the edges :) Apr 21 01:29:31 okie dokie Apr 21 01:29:41 what's the address? Apr 21 01:30:31 openembedded.org. click on the link to the mailing lists.. Apr 21 01:33:42 oe-users or devel for problems with a recipe? Apr 21 01:35:30 devel Apr 21 01:40:51 references to oe-users need to be removed Apr 21 01:43:41 apparently I'm not allowd to post to the mailing list? so i hav eto subscribe first? Apr 21 01:43:51 s/so/do Apr 21 02:27:02 kergoth: there's no libopencv-dev provider Apr 21 02:31:55 and opencv-dev is installed, and it's what's not creating the links Apr 21 02:33:48 Nedlinpopo: the .so symlinks always go in the -dev packages. if they aren't in this case, it's a bug in the FILES_ variables that the recipe is overriding. Apr 21 02:34:03 but they *never* go in the main package, ever. Apr 21 02:34:14 good to know Apr 21 02:34:29 the .so symlink is only used for development, not to run anything Apr 21 02:35:16 typically i try and write code linking against the main .so when possible Apr 21 02:35:58 not sure what you mean by that. generally the soname includes a suffix, and ld only looks for .so, so you can only link against the symlink when using -l. you'd have to specify the full path to do otherwise Apr 21 02:37:57 that's what i mean, usually I link with -lname, especially for shared objects, since that's kinda the point, that they might change to more system specific version. Apr 21 02:41:28 everyone does. it's standard practice. like i said, its expected that hte symlink is included in the -dev package, if it isn't, its a minor bug in the recipe, that's all Apr 21 02:41:47 yes, I'm trying to fix the recipe now Apr 21 02:42:28 I don't yet fully understand recipes syntax, so it's taking a bit Apr 21 02:43:07 PACKAGES is a space separated list of package names to be emitted, FILES_ is a space separated list of files or directories or globs. that's how the files get split up from the do_install output into individual packages. Apr 21 02:43:16 the default values (as with most variables) are in bitbake.conf Apr 21 02:44:17 okay, so FILES_ will specify what is copied into a sysroot? Apr 21 02:44:54 but the recipe will compile if any of the PACKAGES are specifed? Apr 21 02:45:16 no. Apr 21 02:45:21 i'm talking about packaging, not the sysroot Apr 21 02:45:31 PACKAGES specifies what binary packages will be *output* from a recipe Apr 21 02:45:39 has absolutely nothing to do with compiling anything Apr 21 02:45:54 do_install controls what gets split into binary packages *and* what goes into the sysroot Apr 21 02:45:57 ah, PACKAGES makes .ipgk files Apr 21 02:46:04 as i said, binary packages Apr 21 02:46:08 no, its not limited to ipk files Apr 21 02:46:15 tar, rpm, and deb are also available Apr 21 02:54:16 welcome back Apr 21 02:54:51 heh, i hate computers some days Apr 21 02:55:12 hah! I think I probably hate computers every weekday at least! Apr 21 02:55:15 :P Apr 21 02:58:03 hey kergoth, can you take a quick look at recipes/opencv/opencv_2.2.bb with me Apr 21 02:58:17 I think I found the bug, but I'm not 100% Apr 21 02:59:00 looking at line 40 **** ENDING LOGGING AT Thu Apr 21 02:59:58 2011