**** BEGIN LOGGING AT Mon Aug 30 02:59:57 2010 Aug 30 05:43:34 gm everyone ? Aug 30 05:44:04 s/?/!/ Aug 30 05:44:49 khem, gcc 4.4.4 on ppc/calamari works for me, now fixing the u-boot recipe; that one worked for me since the git tarball was in my downloads dir Aug 30 05:45:09 eFfeM_work: ok good to know Aug 30 05:45:26 I have already committed the uboot src_uri update fix Aug 30 05:47:08 good (actually I want to move the recipe to the master branch; it is now on a custodian branch because at that time i needed something in that branch that was not mainstream yet) Aug 30 05:47:54 and as far as I know I am the quite likely the only user of calamari (at least I have never heard from anyone else who used it, except likewise and I have the system he used) Aug 30 05:47:58 sure my fix was temporary and I dont have calamari hardware to test on anyway Aug 30 05:48:16 yes Aug 30 05:48:18 :) Aug 30 05:48:18 I still need to test on the target will do so later today Aug 30 05:48:25 sure Aug 30 05:48:45 there is a bug in debian renaming Aug 30 05:48:55 btw that branch you asked for is 4 hrs ago is probably from enrico scholz, he did the last commits in it onncurses Aug 30 05:49:02 ah ok, can happen, np Aug 30 05:49:12 if we rename a library package then the depends are lost hmm Aug 30 05:49:23 causes all libstdc++ headaches Aug 30 05:49:52 ouch Aug 30 05:50:27 libstdc++6 does not appear in depends of any ipk Aug 30 05:50:37 even libstdc++-dev Aug 30 05:50:40 missed it Aug 30 05:50:50 oops Aug 30 05:51:14 firstly if we rename a lib ipk then -dev -static -doc and -dbg should be renamed too Aug 30 05:51:21 but that does not happen Aug 30 05:52:01 need to look into debian.bbclass and package.bbclass for fix I guess Aug 30 05:52:26 hm still a gcc issue, let me pastebin Aug 30 05:52:38 k Aug 30 05:53:44 packaged staging issue: http://www.pastebin.ca/1928820 Aug 30 05:54:12 not sure why it wants to do that in the first place I rm-ed tmpdir last saturday and rebuild from scratch Aug 30 05:56:29 pstage is in tmpdir, this is what ls -l *gcc* in that dir gives: Aug 30 05:56:30 -rw-r--r-- 1 frans frans 20507532 2010-08-28 11:38 staging-gcc-cross-initial-ppce500v2-oe-linux-gnuspe_4.4.4-r5.0_x86_64-linux-ppce500v2-linux-gnuspe.ipk Aug 30 05:56:30 -rw-r--r-- 1 frans frans 21501730 2010-08-28 11:38 staging-gcc-cross-intermediate-ppce500v2-oe-linux-gnuspe_4.4.4-r5.0_x86_64-linux-ppce500v2-linux-gnuspe.ipk Aug 30 05:56:30 -rw-r--r-- 1 frans frans 62014252 2010-08-28 11:40 staging-gcc-cross-ppce500v2-oe-linux-gnuspe_4.4.4-r5.0_x86_64-linux-ppce500v2-linux-gnuspe.ipk Aug 30 05:56:41 hmmm i guess if you have staging then -initial -intermediate recipes are not nedded Aug 30 05:57:15 i dont use packaged staging Aug 30 05:57:15 well actually the dir should be there and it should not look at staging at all Aug 30 05:57:53 ah ok, nevermind this is probably my bad Aug 30 05:58:55 i changed local.conf last saturday, but the first build I started here for some reason still wanted to use 4.3.3 that probably removed gcc from sysroots Aug 30 05:59:14 I'll kick off a fresh build again Aug 30 05:59:21 hmm ok Aug 30 05:59:32 yeah mixing gcc can be nasty Aug 30 06:00:36 yeah I changed in local.conf but the system at work probably had something cached, after setting the env vars again things went ok Aug 30 06:02:07 (that is: i got 4.4.4), probably it still had some old stuff around; then when I got 4.4.4 again I got the staging error. Aug 30 06:02:26 btw will try to find some time to set up something to participate in the testing Aug 30 06:03:00 mostly running into the std problem of disk space Aug 30 06:03:16 use rm_work Aug 30 06:06:46 i will Aug 30 06:07:53 probably will try to do console-image for nios2 and slug as we have similar hardware we use at work Aug 30 06:08:19 and probably try to add cairo or so, that'll drag in some X stuff too Aug 30 06:10:28 hi fahad_usman Aug 30 06:10:40 hi eFfeM_work Aug 30 06:10:43 found some new recipes to tackle this week ? Aug 30 06:10:44 good morning Aug 30 06:11:08 actually tom rini gave me a list of recipes to fix Aug 30 06:11:21 ok, good Aug 30 06:11:23 there are a lot of recipes still left Aug 30 06:11:32 i will be fixing them one by one Aug 30 06:14:28 good Aug 30 06:15:12 Hi Aug 30 06:15:20 hi noor Aug 30 06:15:26 Hi eFfeM_work Aug 30 06:15:34 i was looking for u Aug 30 06:15:46 need some help regarding tcltk Aug 30 06:16:12 I was combining tcl and tk target recipes with native Aug 30 06:16:20 well here I am (but you can always ask here, there are people around that are much more knowledgeable than me) Aug 30 06:16:50 but I am getting invalid TCL_SRC_DIR variable value when tk configure is executed Aug 30 06:17:03 OK Aug 30 06:17:28 well you found an intresting one, the native recipe is older than the target one Aug 30 06:17:41 was that with tcl or with tk ? and what version? Aug 30 06:17:58 the command in tk configure that is executed is eval av=val=\$$ac_var Aug 30 06:18:16 8.5.8 Aug 30 06:18:20 i was building tk Aug 30 06:18:45 and I ready have tcl 8.5.8 converted to native using BBCLASSEXTEND Aug 30 06:19:50 I get following value for TCL_SRC_DIR var Aug 30 06:20:08 /home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/include/tcl-8.5.8 Aug 30 06:20:26 which is wrong Aug 30 06:21:09 it should be /home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/include/tcl-8.5.8 Aug 30 06:21:36 sorry the command executed in configure script is Aug 30 06:21:53 eval ac_val=\$$ac_var Aug 30 06:21:56 thre is a ${D} too much in one of your recipes or in the configure script Aug 30 06:22:23 I just added BBCLASSEXTENDED In the current recipe Aug 30 06:22:41 i didnt change anything in the current target recipe Aug 30 06:23:06 can u explain how TCL_SRC_DIR gets the value using eval command Aug 30 06:23:25 yeah but the do_stage in the native recipe has this: sed -i "s+${WORKDIR}+${STAGING_INCDIR}+g" tkConfig.sh Aug 30 06:23:41 did you merge the do_stage into the target recipe ? Aug 30 06:23:50 nops Aug 30 06:24:18 i think that causes the problem, the do_stage in the native recipe does a lot more than the do_install in the regular recipe Aug 30 06:24:43 and this sed cmd changes a path in tkConfig.sh Aug 30 06:25:59 and eval command gets the path from this sh file? Aug 30 06:26:01 a first step would be to put do_stage into the target recipe (probably name it do_install_virtclass_native and set NATIVE_INSTALL_WORKS) Aug 30 06:26:27 after that merge the two do_installs (target and native) as far as possible. Aug 30 06:27:03 it changes a path so it could well affect the setting of dirs. can't check since I do not have those sources handy and currently I'm running another build Aug 30 06:27:23 OK no problem Aug 30 06:27:53 does eval uses tkconfig.sh file to get the paths? Aug 30 06:28:02 but what i sketch seems the most plausible reason Aug 30 06:28:12 i expect it does, but can't tell without the src Aug 30 06:29:43 thanks for giving me some pointers Aug 30 06:30:02 noor and eFfeM_work: it would be do_install_virtclass-native (with hyphen between virtclass and native) Aug 30 06:30:12 yes, right Aug 30 06:30:23 thanks fahad Aug 30 06:30:30 :) np Aug 30 06:31:28 i sent a patch of Makedevs on thursday Aug 30 06:31:35 didn Aug 30 06:31:38 didn Aug 30 06:31:44 didn't get any feedback yet Aug 30 06:32:23 this one? http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=c7fe35ece6a4f87ecda31104795cb754dbf741fe Aug 30 06:32:54 yes, exactly Aug 30 06:32:59 reviewed it, build it, checked if recipes that depend on it build, that was all ok, so pushed it Aug 30 06:33:09 oh, okay Aug 30 06:33:11 thanks Aug 30 06:33:13 np Aug 30 06:52:43 good morning Aug 30 06:52:51 hi mckoan, had a good holiday? Aug 30 06:53:04 eFfeM_work: yes thanks Aug 30 06:53:20 you missed all the fun here :-) Aug 30 06:54:55 eFfeM_work: really? what fun? Aug 30 06:56:04 nothing special, i removed a lot of old recipes, broke a thing or two, some people were unhappy Aug 30 06:56:15 details on the ML Aug 30 06:59:42 eFfeM_work: hmmm... working with OE is becoming every day more and more complicated ;-) Aug 30 07:00:03 hm, it should not be, it should become simpler Aug 30 07:08:07 if you have suggestions to simplify things, I'm all ears Aug 30 07:40:31 03Roger Monk  07org.openembedded.dev * r59945c94bc 10openembedded.git/conf/distro/include/ (2 files): Aug 30 07:40:31 angstrom-codec-engine-* : Add Preferred DMAI versions Aug 30 07:40:31 Signed-off-by: Roger Monk Aug 30 07:40:31 Signed-off-by: Koen Kooi Aug 30 07:40:33 03Roger Monk  07org.openembedded.dev * r8dd16b11eb 10openembedded.git/ (5 files in 5 dirs): (log message trimmed) Aug 30 07:40:33 linux-davinci: Add patches to disable NAND SUBPAGE on da850-omapl138-evm/hawkboard Aug 30 07:40:33 * NAND SUBPAGE seen to give problems with UBIFS Aug 30 07:40:33 * Workaround identified by Caglar Aug 30 07:40:34 * Applied configuration locally here to board files Aug 30 07:40:34 * Add Initial UBI settings to da850-omapl138 machine Aug 30 07:40:35 * Bump PR Aug 30 07:40:35 03Roger Monk  07org.openembedded.dev * rf578f77c07 10openembedded.git/ (4 files in 3 dirs): Aug 30 07:40:36 linux-davinci: Add SATA configuration patches for da850-omapl138/hawkboard Aug 30 07:40:36 * Fix small typo Aug 30 07:40:37 * Tune Clock/Data Recovery Configuration Aug 30 07:40:37 * Bump PR Aug 30 07:41:09 Signed-off-by: Roger Monk Aug 30 07:41:09 Signed-off-by: Koen Kooi Aug 30 07:42:48 03Roger Monk  07org.openembedded.dev * rba990688f2 10openembedded.git/recipes/angstrom/angstrom-gdm-autologin-hack.bb: Aug 30 07:42:48 angstrom: Add (hack) recipe to enable autologin (timed) for gdm Aug 30 07:42:48 * Append autologin to default custom.conf Aug 30 07:42:48 * Longer term need better replacement/conf mechanism Aug 30 07:42:48 Signed-off-by: Roger Monk Aug 30 07:42:48 Signed-off-by: Koen Kooi Aug 30 07:55:16 03Koen Kooi  07org.openembedded.dev * rf4de7c054a 10openembedded.git/conf/machine/ (da850-omapl138-evm.conf hawkboard.conf): Aug 30 07:55:16 hawkboard, da850-omapl138-evm: change UBI parameters after kernel patch Aug 30 07:55:16 * sub-page size needs some looking at Aug 30 08:20:36 morning Aug 30 08:20:44 morning Aug 30 08:21:04 good morning Aug 30 08:42:54 hi Aug 30 08:50:50 is there any way that I can set SYSROOT_PREPROCESS_FUNCS just for native Aug 30 08:51:06 if I am using BBCLASSEXTENDED In the recipe Aug 30 08:51:32 I mean one function work for target and one for native Aug 30 08:51:52 I want 2 functions in SYSROOT_PREPROCESS_FUNCS Aug 30 08:55:46 hi Noor Aug 30 08:56:42 i think you can do this by appending virtclass_native to the variable Aug 30 08:56:56 i think it would work in this case Aug 30 08:57:47 fahad_usman: virtclass-native :-) Aug 30 08:58:03 eFfeM_work: ya Aug 30 08:58:06 :) Aug 30 08:58:09 thanks Aug 30 08:58:41 eFfeM_work: btw, i just tried to build wireshark that depend on makedevs Aug 30 08:58:56 it was built successfully Aug 30 08:59:07 ah ok, didn't build for me, see ml Aug 30 08:59:17 ya, i saw your mail Aug 30 09:00:27 one thing i suspect was the reason why some of the recipes were failing to build on my system (that i was compaining last week as well) Aug 30 09:00:44 was that during the build, i ran out of space Aug 30 09:01:04 and then i freed up some space and run bitbake again Aug 30 09:01:15 some recipes started to fail Aug 30 09:01:23 hi. is there a bitbake recipe to install the kernel headers? Aug 30 09:01:37 linux-libc-headers Aug 30 09:01:45 cheers Aug 30 09:01:56 and then when i deleted the tmp folder, and build again, they build fine Aug 30 09:02:21 nice Aug 30 09:02:30 will try wireshark again tonight Aug 30 09:03:02 ya, it took it a few hours to build Aug 30 09:19:23 good morning Aug 30 09:33:16 hi. what do you guys make of this I am getting when trying to install linux-libc-headers? http://pastebin.com/9uq6J1Qj Aug 30 09:41:58 something is wrong with the data on your feed or the cached info on your system is wrong Aug 30 09:42:23 c't has a long article about embedded linux... wit a section about OE Aug 30 09:43:09 florian: online? or is this paperware Aug 30 09:43:45 eFfeM_work: paper... might show up online some time Aug 30 09:45:37 http://www.heise.de/ct/inhalt/2010/19/164/ Aug 30 09:45:47 can't read it though too small Aug 30 09:46:45 On page 168 you can see a narcissus screenshot :) Aug 30 10:26:28 hi eFfeM_work, i want to merge a native recipe into its non-native counterpart, but the non-native recipe has a newer version than the native one Aug 30 10:27:25 the version of non-native recipe exist the recipes but it does not get selected automatically by bitbake Aug 30 10:29:15 and when i do bitbake ...-native it says nothig provides ...-native Aug 30 10:30:49 try to make a newer native recipe and use that Aug 30 10:30:54 afk for lunch back in 30 mins Aug 30 11:02:21 please eFfeM_work ping me when you are back Aug 30 11:56:26 eFfeM_work: how best should I be cleaning the data on this system? I only built it a couple of days ago... Aug 30 12:55:02 fahad_usman: had to go to a meeting, i suggest to start updating the native recipe to the latest version and if that works start merging it Aug 30 12:55:27 johntramp: no idea what cmd you gave to install things Aug 30 12:56:33 this is something I do not expect if you did an opkg install linux-libc-headers: //usr/lib/opkg/info/hicolor-icon-theme.postinst: line 7: can't create /etc/gtk-2.0/gdk-pixbuf.loaders: nonexistent directory Aug 30 13:01:18 eFfeM_work: okay, thanks Aug 30 13:04:28 eFfeM_work: i built the ipk with bitbake linux-libc-headers and then copied it to the gumstix board and tried to install with opkg install linux-libc-headers-dev_2.6.31-r5.5_armv7a.ipk Aug 30 13:04:50 strange, and you can install other packages ? Aug 30 13:05:09 that same process worked for screen and htop fine Aug 30 13:06:02 i built the distro with bitbake omap3-console-image if that makes any difference Aug 30 13:06:03 well it seems to start with configuring util-linux-ng-readprofiel Aug 30 13:06:16 noidea why it would do that Aug 30 13:06:47 you build the whole image yourself? or did you get it from the feed ? Aug 30 13:07:03 i mean angstrom autobuilder Aug 30 13:07:39 no i built it Aug 30 13:07:42 if it is from the autobuilder, I suggest removing your ipk and just do an opkg install from the feed Aug 30 13:07:43 ah ok Aug 30 13:08:10 then I fear I'm stuck Aug 30 13:08:48 ok, well i appreciate your efforts Aug 30 13:08:54 np Aug 30 13:08:56 yw Aug 30 13:17:11 can anyone tell me how to handle ${sysconfdir} in new style staging? Aug 30 13:22:59 fahad_usman: what do you mean handle? Aug 30 13:24:43 like we replace ${STAGING_BINDIR} with ${D}${bindir} Aug 30 13:25:01 ${D}${sysconfdir} Aug 30 13:26:07 XorA: when i replace it this way, i get errors Aug 30 13:26:29 | install-catalog: "/home/fahad/systemBuilder/oe-git/build/tmp/work/i686-linux/docbook-sgml-dtd-3.1-native-1.0-r1.0/image//home/fahad/systemBuilder/oe-git/build/tmp/sysroots/i686-linux/etc/sgml/sgml-docbook.cat" does not exist or is empty. Aug 30 13:27:07 dont forget you need to install -D directories before installing into them Aug 30 13:27:16 oh Aug 30 13:27:28 i think that would do the trick, thanks Aug 30 14:33:17 anybody having any luck with openjdk-6? Aug 30 14:36:57 hi Aug 30 14:37:17 I developed a program using autotols to handle makefiles generation Aug 30 14:37:40 it work fine, but when I try to use it with OE the configure script hangs Aug 30 14:37:55 The error I get is Aug 30 14:37:56 | checking whether the C++ compiler works... configure: error: cannot run C++ compiled programs. Aug 30 14:37:57 | If you meant to cross compile, use `--host'. Aug 30 14:38:06 so I think it is related to configure.ac Aug 30 14:38:34 I pute AC_CANONICAL_HOST, but is does not seem to work. What am I missing? Aug 30 14:45:04 ottavio: what does the config.log say? Aug 30 14:47:35 here it is, zecke http://pastebin.com/x2Q0mbBd Aug 30 14:48:41 the error is around line 609 Aug 30 14:49:05 I think it compiles for arm but then it tries to execute on the host Aug 30 14:55:21 03Steffen Sledz  07org.openembedded.dev * r6c7e942fae 10openembedded.git/MAINTAINERS: Aug 30 14:55:21 MAINTAINERS: updated my maintainership Aug 30 14:55:21 Signed-off-by: Steffen Sledz Aug 30 15:20:07 morning Aug 30 15:30:24 anyone looked at http://nodejs.org/ yet? Aug 30 15:30:29 seems like an interesting technology Aug 30 15:31:04 interesting indeed Aug 30 15:45:12 does pkg_postinst get run when making an image Aug 30 15:46:39 yes. some postinsts can and are completed at image creation time -- others have to be postponed until first boot Aug 30 15:52:55 hmm cause I was looking at the syslog-ng recipe which has pkg_postinst() { Aug 30 15:52:55 update-rc.d -f syslog remove Aug 30 15:52:55 } Aug 30 15:52:55 but in my image syslog is still linked to the rc directories. is there any extra step i need to do? Aug 30 15:58:52 ron, you aren't pointing the update-rc.d to the image as the rootfs, no? Aug 30 16:06:01 bye Aug 30 17:15:14 has anyone ever made an image without busybox at all? Aug 30 17:18:18 Crofton, made means? Aug 30 17:18:24 i did Aug 30 17:18:33 but they are not in oe Aug 30 17:18:40 heh Aug 30 17:18:48 we should get one on OE Aug 30 17:18:52 x11-sato2-image Aug 30 17:19:07 and ... bisect-image Aug 30 17:20:10 sato2 will go in oe when ready Aug 30 17:20:24 but bisect was only for a bisect Aug 30 18:14:46 03Khem Raj  07org.openembedded.dev * rc9d34f282f 10openembedded.git/conf/machine/efikamx.conf: Aug 30 18:14:46 efikamx.conf: Define entry point and load adress for kernel. Aug 30 18:14:46 Signed-off-by: Khem Raj Aug 30 18:40:31 hi ericben Aug 30 18:54:48 03Denys Dmytriyenko  07org.openembedded.dev * r7ec84b10a3 10openembedded.git/recipes/ti/matrix-tui_svn.bb: Aug 30 18:54:48 matrix-tui: fix DEPENDS and use system LDFLAGS Aug 30 18:54:48 Signed-off-by: Denys Dmytriyenko Aug 30 19:03:10 hi GNUtoo|laptop Aug 30 19:03:47 I'll PM you Aug 30 20:15:01 Is there an easy way to stop bitbake from stripping files? Aug 30 20:15:27 it isn't bitbake doing it. why would you want to, though? Aug 30 20:17:01 I am doing some development Aug 30 20:17:10 so I want to run unstripped binaries Aug 30 20:17:16 yes Aug 30 20:17:20 INHIBIT_DEFAULT_STRIP or so Aug 30 20:17:24 But gdb grocks .debug dirs :) Aug 30 20:17:26 I am working through building things via srctree and how to test Aug 30 20:17:29 installing the -dbg package is insufficient? Aug 30 20:18:02 not sure, still poking around Aug 30 20:18:12 Really should be enough Aug 30 20:18:25 yeah, I know Aug 30 20:18:38 does that overwrite binaries? Aug 30 20:18:50 i want to sue things like oprofile Aug 30 20:18:52 Tartarus: maybe INHIBIT_PACKAGE_STRIP ? Aug 30 20:19:17 yeah Aug 30 20:20:24 Crofton_|work: no, it puts the .debug sections into files in .debug Aug 30 20:20:33 ie /bin/busybox and /bin/.debug/busybox Aug 30 20:20:35 ah Aug 30 20:20:37 And point gdb at the rootfs like normal Aug 30 20:20:42 and it finds the .debug dirs Aug 30 20:20:44 iirc, been a while Aug 30 20:20:54 gdb knows how to find them by default, yeah, as far as i know Aug 30 20:20:57 that would explain why I think oprofile is not giving me detailed data :) Aug 30 20:21:09 but it appears oprofile does not Aug 30 20:23:22 oprofile did give me detailed info on my app and on the lib, might be that I did use an unstripped appandlib Aug 30 20:23:26 app and lib Aug 30 20:23:47 could not get it work with kernel symbols Aug 30 20:23:59 yeah I am failig with the kernel Aug 30 20:24:04 but there is an option to get raw data and then you can relate it to the symbol file yourself Aug 30 20:24:07 that is what I did Aug 30 20:24:14 also the lib I am loking at is from a cmake package Aug 30 20:24:27 and i think cmake stuff pakages "poorly" Aug 30 20:50:08 don't knwo too much about cmake. Aug 30 20:50:17 sry Aug 30 20:50:21 & calling it a day Aug 30 20:50:31 have a nice day/evening/... Aug 30 22:10:25 Speaking of stripping binaries... I noticed that the openssh recipe is the only one that sets EXTRA_OEMAKE = "'STRIP_OPT='" Aug 30 22:11:23 However, when I'm cross-compiling, do_qa_staging still tries stripping the binaries and complains of errors when doing so Aug 30 22:22:07 of already stripped? Aug 30 22:22:19 It's possible the hook in the openssh recipe changed Aug 30 22:22:24 (s/recipe/Makefile) Aug 30 22:22:49 Tartarus, thanks, taking a closer look now Aug 30 23:01:33 <[Rui]> hi, I've an oe doubpt... Aug 30 23:01:57 <[Rui]> I used to build a package with bb -c build -b /path/to/recipe.bb Aug 30 23:02:13 <[Rui]> but now that isn't working, I mean... it doesn't even fetch the sources. Aug 30 23:02:25 <[Rui]> any suggestions on what may be my problem? Aug 30 23:03:11 hi, i got a | cp: cannot stat `/home/amir/oe/tmp/work/overo-angstrom-linux-gnueabi/ti-cgt6x-1_6_1_9-r4/cgt6x_6_1_9/*': No such file or directory Aug 30 23:03:16 error Aug 30 23:03:56 i tried updating with git pull --rebase which tells me that this bb file needs update but it wont do it Aug 30 23:04:35 also git stash git pull --rebase didnt work either Aug 30 23:05:05 the directory above does exist Aug 30 23:06:42 [Rui]: sounds like you'd already built it. it doesn't rebuild what it's already built. do a -c clean first. also, -c build is redundant, that's the default Aug 30 23:07:12 <[Rui]> kergoth: well, no, it was a long time ago. I suspect I need to add something to conf/local.conf but I'm not sure what Aug 30 23:08:12 <[Rui]> kergoth: meanwhile, I've built and rebuilt my shr-unstable buildtree, and didn't preserve it when changing the GNU/Linux I was using to another one, so I must've lost some change I can't recall. Aug 30 23:08:40 <[Rui]> kergoth: I tried adding BBFILES+="local_recipes_root_path/*/*.bb" but it didn't seem to work. Aug 30 23:09:24 <[Rui]> I would really prefer to build with bb -c build -b path.bb because otherwise I will have to wait for it to check if parsing is needed on over 8k recipes (takes too long). Aug 30 23:10:17 <[Rui]> this is the recipe: http://pastebin.com/9383WcXi which worked previously. Aug 30 23:13:47 anyone? Aug 30 23:20:43 BBFILES has nothing to do with -b Aug 30 23:21:04 BBFILES is how it finds recipes when you aren't using -b Aug 30 23:30:57 <[Rui]> kergoth: yes, it's what I thought, but I was getting desperate :| Aug 30 23:31:49 you're going to have to elaborate on "isn't working" Aug 30 23:32:12 <[Rui]> kergoth: well, did you check my recipe? http://pastebin.com/9383WcXi Aug 30 23:32:28 "Unknown paste id" Aug 30 23:32:28 <[Rui]> kergoth: it seems it's not fetching the SRC_URI I specified Aug 30 23:32:38 that still isn't telling me anything. Aug 30 23:32:42 what exactly *is* bitbake doing? Aug 30 23:32:47 is there an error? does it return with no output? what? Aug 30 23:32:49 <[Rui]> oh, I forgot, only lasted 1h, let me remake Aug 30 23:32:58 <[Rui]> kergoth: it's not fetching the source Aug 30 23:32:59 "isn't working" isn't very useful information Aug 30 23:33:04 again, tells me nothing Aug 30 23:33:18 don't summarize. if there's an error, paste it. Aug 30 23:34:03 <[Rui]> hard to say, the only error is an obvious error, no target for install (because there's no source) even though it passed (too quickly because it did nothing) the previous steps.. I'll do a complete paste... Aug 30 23:35:41 <[Rui]> here: http://pastebin.com/UqAvwd6h (recipe, plus bb session, + only error) Aug 30 23:38:19 ah! Aug 30 23:39:16 *that* behavior. I'm pretty sure that's a bug in the unpack task. in the meantime, you can try setting S = "/media/devdisk/svn/omnewrotate" and SRC_URI = "", and it'll build right in the tree, rather than copying it into WORKDIR in tmp/ Aug 30 23:39:22 * kergoth looks into it Aug 30 23:40:06 <[Rui]> I'd rather it copied it, but it can work like that for a workaround! Aug 30 23:41:46 alternatively, you could override the unpack task Aug 30 23:41:54 do_unpack () { Aug 30 23:42:03 cp -a ${SRC_URI} ${WORKDIR}/${PN} Aug 30 23:42:03 } Aug 30 23:42:10 do_unpack[python] = "" Aug 30 23:42:14 or similar Aug 30 23:44:23 * [Rui] will try that... where is unpack defined? Maybe if I could dive into it I could fix the problem Aug 30 23:44:47 base.bbclass Aug 30 23:44:59 base is the main class that's inherited by all recipes Aug 30 23:45:04 okay, i have a fix, hold Aug 30 23:47:41 03Chris Larson  07org.openembedded.dev * r599b03fe99 10openembedded.git/classes/base.bbclass: (log message trimmed) Aug 30 23:47:41 base.bbclass: fix bug with absolute file:// uris Aug 30 23:47:41 The code which used host/path from the url to construct the destination of the Aug 30 23:47:41 copy was using os.path.join. os.path.join treats a "/" in an element as an Aug 30 23:47:41 indication to drop all previous elements, so the full path resulted in a Aug 30 23:47:41 destination exactly the same as the source, resulting in a no-op unpack. Aug 30 23:47:42 Fixed by using oe.path.join instead. Aug 30 23:47:53 os.path.join("/foo", "/bar", "baz") results in /bar/baz, not /foo/bar/baz Aug 30 23:48:06 so it was dropping the entire WORKDIR portion of the destination path :) Aug 30 23:48:18 if you pull, that should fix it for you Aug 30 23:50:38 <[Rui]> oh, so it was a bug there! cool, at least it gets fixed for others too :) Aug 30 23:50:53 yep Aug 30 23:51:07 <[Rui]> I might as well pull and let it chew a new image again during the night... Aug 30 23:51:19 <[Rui]> kergoth: well, thank you very much :) Aug 30 23:51:22 np Aug 30 23:51:33 it was a good excuse to finally fix that, someone mentioned it last week Aug 30 23:51:40 don't think a bug was ever opened Aug 30 23:52:21 <[Rui]> kergoth: I thought it was a problem with my recipe, being a bit ignorant of oe I just assume from start that the problem is mine :) Aug 30 23:52:56 hehe. usually a good assumption when you're new to something, just not in this case Aug 30 23:54:07 <[Rui]> let's hope it's quickly picked up by shr build tree, though. thanks, again :) Aug 31 00:02:53 hmm... looks like testing-next has been branched. But no email sent to list... Aug 31 00:44:48 <[Rui]> kergoth: if you're still there, I've just updated my build tree and the fix is in, however bitbake is spurting out a nasty error (python dump) http://pastebin.com/nf635MHy Aug 31 00:57:02 bah, damnit Aug 31 00:57:30 [Rui]: what version of bitbake is this? Aug 31 00:58:01 <[Rui]> BitBake Build Tool Core version 1.10.0, bitbake version 1.10.0 Aug 31 00:58:02 [Rui]: for now, you can just remove the bb.debug on line 527 of bitbake/lib/bb/utils.py Aug 31 00:58:17 i just ran into this a couple hours ago, actually, haven't had time to get a fix upstream Aug 31 00:58:45 <[Rui]> perhaps the if as well :) Aug 31 00:59:10 well, yeah :) Aug 31 00:59:16 <[Rui]> ok, 7159 packages to build, which means it's time to go to bed! Aug 31 00:59:28 <[Rui]> thanks, once again. Aug 31 00:59:39 what happens is, the environment cleaning has to happen before the cooker class is instatiated, but that cleaning prints a debug message, and its hte cooker that initializes the debug levels Aug 31 00:59:47 just need to rearrange the order of operations Aug 31 00:59:48 np Aug 31 01:00:29 <[Rui]> yeah, I got that far, I guess I could just have commented it out before asking :| Aug 31 01:01:14 really, the debug levels should default to 0, it shouldn't be necessary to initialize them to print debug messages Aug 31 01:05:16 <[Rui]> zZzZzzZz.... good night! :) Aug 31 01:06:24 * kergoth thinks a 1.10.1 may be in order shortly, due to the 3 cherry picked fixes that are now on the 1.10 branch Aug 31 01:33:00 vala-dbus-binding-tool-native failed to compile: http://tinderbox.openembedded.net/packages/722530/ Aug 31 01:33:06 ideas anyone? Aug 31 02:02:52 hi Aug 31 02:03:53 gm **** ENDING LOGGING AT Tue Aug 31 02:59:57 2010