**** BEGIN LOGGING AT Thu Sep 02 02:59:58 2010 Sep 02 04:00:27 03Graham Gower  07org.openembedded.dev * r1d48a1c15d 10openembedded.git/recipes/gpsd/ (gpsd.inc gpsd_2.90.bb gpsd_2.91.bb gpsd_2.94.bb gpsd_2.95.bb): Sep 02 04:00:27 gpsd: unbreak gpsd_2.38.bb and gpsd_2.39.bb Sep 02 04:00:27 The older recipe(s) are needed for programs which have not updated to gpsd's Sep 02 04:00:27 newer api. e.g. prismstumbler. Sep 02 04:00:27 Signed-off-by: Graham Gower Sep 02 04:00:27 Signed-off-by: Khem Raj Sep 02 04:51:50 Hej! Does someone know how to select a specific gcc version on the host? Have 4.3.4 and 4.4.3 installed and would like to use the older. Sep 02 06:11:46 I do have several compiler versions installed, is it selectable which version bitbake uses for -native recipes? Sep 02 06:15:18 vps, you probably need to set the BUILD_CC, BUILD_LD, etc variables. Check bitbake.conf for the full list Sep 02 06:15:56 Right, the bitbake vars, not shell vars Sep 02 06:16:45 maybe both, i dunno Sep 02 06:16:58 I hope, do_configure will honour this setting too. Let's try... Sep 02 06:17:32 how are your multiple toolchains set up? Sep 02 06:18:14 I'm running gentoo, they are "slotted", but one version provides /usr/bin/gcc Sep 02 06:25:31 backreading the log Sep 02 06:25:40 gm Sep 02 06:25:43 ka6sox-work: wrt freenx: np, was interesting Sep 02 06:26:28 grg thanks for cleaning Sep 02 06:26:56 eFfeM_work, but i don't scrub toilets! Sep 02 06:27:09 khem, Crofton, grg, fully agree that we should work to a more stable system and better W Sep 02 06:27:12 better Q Sep 02 06:27:41 eFfeM_work, i really apreciate the work you've been doing RE removal of cruft Sep 02 06:28:04 it makes fixing things a lot easier when you can see the wood from the trees Sep 02 06:28:05 grg, I know, but doing cleaning and fixing in oe is sometimes equally little rewarding Sep 02 06:28:41 yes, and sometimes there is still a little bit too much wood, and in a lot of cases it is deadwood Sep 02 06:29:17 actually I would like it if we were able to at least indicate which are the recipes that are of a decent qualtiy and maintained and which not Sep 02 06:29:24 so you know what you are getting Sep 02 06:37:50 i have a failing build because mbmerlin is providing a garbage x11.pc file, which has overwritten the one from libx11. wtf? Sep 02 07:21:43 morning Sep 02 07:23:00 hi hrw Sep 02 07:28:05 morning Sep 02 07:36:53 okay garnet is now 64bit...and fixed up..so I can sleep Sep 02 07:38:39 nite! Sep 02 07:39:01 btw ka6sox-work are these freenx recipes working properly, didn't really test them yet myself Sep 02 07:39:43 they are close...I'll push changes Sep 02 07:39:54 once it all builds Sep 02 07:47:12 ka6sox-work: great Sep 02 08:07:51 Hi eFfeM_work Sep 02 08:08:00 hi Sep 02 08:08:18 how one can send a patch to delete any recipe Sep 02 08:08:29 or maintainer do it themselves Sep 02 08:09:52 if you want to delete a recipe just do a git rm Sep 02 08:09:58 good morning Sep 02 08:10:05 or a regular rm and use git add or so to add it Sep 02 08:10:20 try also git add -i Sep 02 08:10:28 it will remove from my local repository Sep 02 08:10:44 git rm will add it immediately to the list Sep 02 08:10:46 how can i recommend the community to remove a recipe when it is not needed Sep 02 08:11:16 remove, commit, generate patch, send to list Sep 02 08:11:19 like if I have mergred native with target recipe using BBCLASSEXTEND then there is no need of native recipe Sep 02 08:11:24 it should be removed Sep 02 08:11:31 OK Sep 02 08:11:39 then it should go in the same commit Sep 02 08:11:48 got it. Thanks Sep 02 08:12:05 Hi, ALL Sep 02 08:12:13 I have a question Sep 02 08:13:31 If I run "bitbake -c devshell " and then inside the newly created shell I will run make, should it automatically find the compiler - g++? Sep 02 08:13:55 Or I have to manually enter the full path to it? Sep 02 08:13:59 IgorK: afaik no Sep 02 08:14:05 Thank you for the info Sep 02 08:14:14 * Jay7 prefer to use independent devshell Sep 02 08:14:23 Jay7: no to what? ;-) Sep 02 08:14:38 jay7: independent = ? Sep 02 08:14:39 easy way to test is in devshell do a g++ -v or so to get the version and target Sep 02 08:14:58 likewise: not called by bitbake :) Sep 02 08:15:21 jay7: do-it-yourself devshell? Sep 02 08:16:00 I tried to do that and unfortunately I had to manually put the path to it. Maybe its a problem with particu;ar atom devshell? Sep 02 08:16:08 any idea? Sep 02 08:16:13 http://docs.openembedded.org/usermanual/html/usage_devshell.html Sep 02 08:16:22 devshell addon part Sep 02 08:17:05 that is easier to use for me Sep 02 08:17:12 I can run devshell at any time :) Sep 02 08:17:31 after/before any task I mean Sep 02 08:18:23 thx, learn smth new today...;-) Sep 02 08:18:46 but what about my original question? Sep 02 08:19:14 IgorK: try to run `which g++` Sep 02 08:20:12 or g++ -v as eFfeM_work suggested Sep 02 08:22:19 well, it gives regular "/usr/bin/g++", whereas should be "arm-angstrom-linux-gnueabi-g++" Sep 02 08:22:45 hi all Sep 02 08:22:52 try $CC maybe? Sep 02 08:23:06 afaik devshell sets the variables Sep 02 08:23:08 btw yes Sep 02 08:23:11 "~/OE/openembedded/temp/cross/armv4t/arm-angstrom-linux-gnueabi-g++" Sep 02 08:25:31 we r getting warmer. "echo $CC" gives "arm-angstrom-linux-gnueabi-gcc...", however it does not produce the whole path to it Sep 02 08:25:46 the $PATH is probably set already Sep 02 08:25:49 when merging native and non-native recipes, if there is a function that is defined in the target recipe and over-ridden in native (and native requires target), how can i handle that in the new comined recipe? Sep 02 08:26:47 no!! :surprised Trying to issue "make" gives "file not found" Sep 02 08:28:12 basicaly I tried to compile a sample by hand, i.e. issuing "make" in the devshell and got the "404: File not found" on the compiler name ;-) Sep 02 08:28:53 devshell addon substitute make with oe_runmake Sep 02 08:29:01 not sure about devshell task Sep 02 08:29:07 so I had to manually prepend the compiler name with whole path Sep 02 08:29:48 Jay7: what is the difference? from my point of view there shouldn't be any... Sep 02 08:30:15 besides I just need to compile the program Sep 02 08:31:57 IgorK: oe_runmake is call of some subroutine which setup variables and call make as bitbake does Sep 02 08:32:06 as I understand at least :) Sep 02 08:32:32 I have done no deep investigation here Sep 02 08:32:54 better to ask real hackers :) Sep 02 08:34:45 but from the developer perspective they should be equal, don't u think? Sep 02 08:35:09 it should Sep 02 08:35:22 but I don't know real situation Sep 02 08:36:03 hm, the real situation is that those 2 are not the same - c above Sep 02 08:36:50 but since core oe-devs are not here I will complain in the morning Sep 02 08:37:14 will go to sleep now, thank u Jay7 Sep 02 08:37:26 np :) Sep 02 08:37:42 good night Sep 02 08:38:10 * Jay7 should go to customers now.. Sep 02 09:04:48 Is there a known error in boost_1.41.0.bb? I have a compile error there in CMakeFiles/boost_wave-mt-shared-debug.dir/instantiate_cpp_exprgrammar.cpp.o: In function `__static_initialization_and_destruction_0': Sep 02 09:05:44 I'm compiling for the beagleboard and including Qt. In my rootfs.bb I have added just the qt-libs I need, to keep the size of my rootfs to a minimum. Now I want to try using the SGX, and I can't figure out what to add to my recipe to make that happen. Sep 02 09:28:37 hello, i just posted a questiong, can anyone answer that? Sep 02 09:46:22 hej, is the process of setting up an autobuilder already documented ? (have been on holiday and off oe quite a time) Sep 02 09:46:33 I maybe can provide some builds too Sep 02 09:54:56 03Koen Kooi  07org.openembedded.dev * re2249025ea 10openembedded.git/recipes/gnome/goffice_0.8.9.bb: goffice: add 0.8.9 Sep 02 10:12:42 mlip: there mlip: just finished to setup a oestat server, still have to write the doc, and then to look at buildbot for this Sep 02 10:13:00 mlip: there seems to be a sample config in contrib/buildbot Sep 02 10:13:29 ericben: so you're not even using tinderbox.openembedded.net ? Sep 02 10:14:35 mlip: it's a private server for internal usage, the next goal is a buildbot which will build private projects and report to internal tindebox and build testing branches and report them to oe's tinderbox Sep 02 10:15:25 ericben: i c, I will have a look into buildbot Sep 02 10:15:28 for oestat, I have it running using lighttpd and sakoman managed to get it running with apache after exchanging our config files and studying the logs Sep 02 10:15:58 i have set up an older version of oestat some time ago for internal usage too, but used lighttpd Sep 02 10:16:40 mlip: yes it works fine here also, the hardest part was to put auth on some pages in order not to have customers being able to see other's customers' build log ! Sep 02 10:17:00 lighttpd is quite flexible for this Sep 02 10:17:18 ye it is; do I get this right and the next branch is merged every monday? Sep 02 10:17:53 for testing ? Sep 02 10:18:12 yes Sep 02 10:18:18 I think it's only a tag but I have a lot of backlog in the mails so I can't yet ell you more on this pôint Sep 02 10:18:34 ok meal time see you later Sep 02 10:18:41 thx, bye Sep 02 10:19:19 ok; if you make progress on buildbot keep me informed, I'll work on this tonight Sep 02 10:53:34 a very simple Sep 02 10:53:38 question Sep 02 10:53:45 how do i use printf in a function such as do_install (), in a recipe file Sep 02 10:54:25 all i need to do is print out the destination variable ${D} Sep 02 11:57:47 ciscokid: these are shell scripts so you can just do Sep 02 11:57:50 echo ${D} Sep 02 11:58:08 ah! eFfeM_work thanks Sep 02 11:58:15 yw Sep 02 13:30:26 eFfeM_work: HI Sep 02 13:37:51 Hello Sep 02 13:38:01 hi Sep 02 13:38:16 need little help regarding git Sep 02 13:38:26 if I have created a patch Sep 02 13:38:36 and then want to add something more to it Sep 02 13:38:41 how can I do that Sep 02 13:38:47 rule #1 on irc, just ask; if someone knows the answer they will reply Sep 02 13:38:54 I want to add removal of one recipe in the file Sep 02 13:39:05 OK Sep 02 13:39:10 undo the commit; do a new git add; then recommit Sep 02 13:39:34 Noor: git commit --amend Sep 02 13:39:47 and if that's not the last commit Sep 02 13:39:56 ericben: can you also add /remove files to it ? Sep 02 13:39:58 OK Sep 02 13:40:08 git rebase --interactive HASH_OF_THE_COMMUT_BEFORE Sep 02 13:40:10 thought --amend was only to hcange the commit msg Sep 02 13:40:21 then change the 1st word of the line you want to edit Sep 02 13:40:32 and you will be able to ammend the commit Sep 02 13:41:14 eFfeM_work: unless I'm mistaken yes, you can totally modify the commit (this assume you haven't pushed the commit upstream of course) Sep 02 13:41:27 ah ok, i am not a git wiz Sep 02 13:41:50 thanks ericben Sep 02 13:42:29 just try to push a kernel patch upstream and you will see how ammend & rebase interactive are useful during review ;-) Sep 02 14:08:25 03Steffen Sledz  07org.openembedded.dev * r490c68ac99 10openembedded.git/recipes/rt2x00/files/ (config.patch makefile.patch): Sep 02 14:08:26 rt2x00: move driver version specific patch files to an own directory Sep 02 14:08:26 Signed-off-by: Steffen Sledz Sep 02 14:48:39 pb_, mind unbanning me from the tsc channel in 15 minutes or so, if it doesn't do so automatically? I hate it when i forget to identify ;) Sep 02 14:56:20 kergoth: heh, done. fyi, you can unban yourself: /msg chanserv unban ##tsc Sep 02 14:56:33 ah Sep 02 14:56:37 thanks, thats good to know Sep 02 14:56:43 i'm sure this won't be the last time i'll need it Sep 02 14:56:56 yeah, I have the same problem every time I get disconnected Sep 02 14:56:59 heh Sep 02 14:57:15 even rp got banned recently, I see :-} Sep 02 14:57:18 someday i'll get around to configuring my irc proxy to just auto identify Sep 02 14:57:20 heh :) Sep 02 15:02:41 * RP did indeed Sep 02 15:02:46 It happens every reconnect Sep 02 15:04:51 mickey|: ping Sep 02 15:07:12 RP__: Do you know something about OE's metatoolchain? Our gcc-cross-sdk is using the BUILD_LDFLAGS that end up with a -rpath in the binaries. Sep 02 15:07:50 zecke: It is using the old sdk stuff or the newer nativesdk stuff? Sep 02 15:08:56 hi all Sep 02 15:09:28 i posted a question here a few hours back, but prolly nobody noticed it Sep 02 15:10:21 RP__: no idea, I was looking at gcc-cross-sdk.inc, it passes BUILD_LDFLAGS. Where is the difference regard to rpath? Sep 02 15:10:43 when merging native and non-native recipes, if there is a function that is defined in the target recipe and over-ridden in native (and native requires target), how can i handle that in the new combined recipe? Sep 02 15:11:00 kergoth_: you still like parsers? Sep 02 15:12:10 zecke: pong Sep 02 15:12:53 zecke: Its using the old sdk approach. It was reasons like this I wrote the nativesdk code Sep 02 15:13:14 okay, I will need to modernize my knowledge Sep 02 15:29:47 fahad_usman: you have to write two different functions Sep 02 15:30:02 Hi RP Sep 02 15:31:29 RP: are you familiar with debian renaming class ? there is something fishy when we rename a library say libstdc++ then the stuff depending on it does not get the dependency right Sep 02 15:31:58 e.g. libstdc++-dev should have libstdc++ in its depends but it does not Sep 02 15:32:12 khem: I am familiar with it. I'd suspect the ++ causing problems? Sep 02 15:32:13 if I look at other -dev packages then the dependency is there Sep 02 15:32:37 RP: no idea. I wanted to look at the issue Sep 02 15:32:44 where should I start looking at Sep 02 15:32:59 RP: I see same issue with libgcc too I think Sep 02 15:33:30 khem: package_renaming_hook does the work iirc Sep 02 15:34:04 RP: ok. Sep 02 15:34:49 if I dump -dev ipk of libstdc++ it has no depenedencies Sep 02 15:35:08 iirc, the problem with libstdc++ is that the base package comes from gcc-cross and the -dev comes from target gcc. Sep 02 15:35:09 or some such Sep 02 15:35:24 oh I see Sep 02 15:35:40 one of these days I wanted to look into gcc-runtime thing Sep 02 15:35:48 yeah, I think that would fix it Sep 02 15:35:52 Poky has that working well now Sep 02 15:35:57 nice. Sep 02 15:36:19 did you have libgcc also separated out ? Sep 02 15:39:31 morning khem Sep 02 15:41:24 hey Sep 02 16:23:17 03Chris Larson  07org.openembedded.dev * r900cc29b60 10openembedded.git/ (classes/base.bbclass classes/patch.bbclass lib/oe/unpack.py): (log message trimmed) Sep 02 16:23:17 do_unpack, do_patch: shift some responsibility around, clean things up Sep 02 16:23:17 - Consolidate 'is this file a patch' logic Sep 02 16:23:17 - Move unpack functions from classes into oe.unpack Sep 02 16:23:17 - Move the unpacking message printing into do_unpack Sep 02 16:23:17 - Move the destination directory determination into do_unpack Sep 02 16:23:18 - Use subprocess's ability to pass in PATH and cwd rather than mangling the cmd Sep 02 16:26:37 http://benjamin-meyer.blogspot.com/2010/06/managing-project-user-and-global-git.html is nice Sep 02 16:31:25 hi Sep 02 16:31:44 03Frans Meulenbroeks  07org.openembedded.dev * ra1e06cb16d 10openembedded.git/MAINTAINERS: Sep 02 16:31:44 MAINTAINERS: forgot to add urjtag yesterday Sep 02 16:31:44 Signed-off-by: Frans Meulenbroeks Sep 02 16:46:36 khem, apparently still have an issue with 4.4.4; everything compiles ok, but when I try to execute programs they get a core dump. might be stack related I didn't have gdb on the target but did add some printf's and after a return things seem to go wrong Sep 02 17:01:31 eFfeM: ok Sep 02 17:01:58 eFfeM: I guess you need to use -O2 in general for ppc with gcc 4.4 Sep 02 17:03:14 eFfeM: change conf/distro/include/sane-toolchain-eglibc.inc Sep 02 17:03:31 FULL_OPTIMIZATION = "-fexpensive-optimizations -fomit-frame-pointer -frename-registers -O2" Sep 02 17:03:41 that will fix your problem Sep 02 17:04:31 checking for /scratch/oe/sysroots/java/inetlib.jar... configure: error: can't find inetlib.jar in specified path Sep 02 17:04:48 while building gnumail_1.1.2.bb Sep 02 17:05:00 and inetlib is already built Sep 02 17:22:38 khem might try that later Sep 02 17:48:48 eFfeM, khem: I found a recent entry in the Ubuntu 10.04 update feed about a fix for GCC / PowerPC in combo with -O0 (no optimization). I already spent some time tracking the patch down. One problem is, I clicked through the GCC native update on my 10.04 installation, so I cannot easily find the corresponding Ubuntu changelog item. Sep 02 17:51:25 let's see if I can find the web site where each package is tracked - is that fully launchpad these days? Sep 02 17:52:44 well for now probably i'll stick to 4.3.3 Sep 02 17:58:22 eFfeM: uhm, I think I saw problems with 4.3.3 actually Sep 02 17:59:05 eFfeM: cannot tell for sure though, haven't encountered problems in a while. I remember our "avahi" daemon was failing with an ICE. Sep 02 17:59:18 Sorry, not an ICE, a core dump. Sep 02 17:59:52 Hmm, I think I can test e500v2 avahi here, let me give it a spin. Sep 02 18:00:34 eFfeM: -O2 will work I have tried it here. never had time to debug the -Os problem Sep 02 18:00:40 but it should be fixable Sep 02 18:01:38 hm, ok Sep 02 18:01:53 likewise, interested in your findings Sep 02 18:02:22 eFfeM: what compiler did you hit problems with (I assume you are targetting e500v2?) Sep 02 18:02:39 eFfeM: our gcc-cross 4.4.4 in OE? Sep 02 18:03:01 yes Sep 02 18:03:38 khem, eFfeM, others: http://free-electrons.com/blog/elc-2010-videos/ Sep 02 18:03:46 i had problems compiling, then khem gave me a fix last saturday, that compiled everything fine, but didn't get to testing until today Sep 02 18:03:53 The ELC 2010 videos are here. Looks like a set of darn interesting talks. Sep 02 18:05:03 nice! thanks for sharing that Sep 02 18:17:20 eFfeM likewise gcc 4.5 works well with ppc exxx Sep 02 18:17:29 with -Os Sep 02 18:17:36 I have tested it on qemu Sep 02 18:17:51 so I think your best bet is gcc 4.5 Sep 02 18:18:05 ok, can give that a try too Sep 02 18:18:13 I can debug 4.4 issue but I have no time literally Sep 02 18:18:24 np Sep 02 18:18:36 i greatly appreciate your help & suggestions Sep 02 18:18:39 I have to prepare course material for a training class and 2 presentations in next 2 weeks Sep 02 18:19:37 who deals with java in OE Sep 02 18:19:43 there is an iteresting bit Sep 02 18:19:45 ouch, lots of work. I need to start lecturing next tuesday, but first evening is on tcp-ip. I can almost dream that one Sep 02 18:19:56 * eFfeM has no java experience' Sep 02 18:20:19 eFfeM: the work I mentioned is in addition to day time work where the project is bleeding too Sep 02 18:20:25 so I am deep in shit Sep 02 18:21:00 khem btw mostly prepared to get thigns on the autobuilder, but haven't found a reliable way to automatically update git as rejects connection sometimes Sep 02 18:21:17 check for the return of git ops Sep 02 18:21:23 and if its not 0 then reissue it Sep 02 18:22:00 one thing thats hitting the large autobuilds is the race conditions Sep 02 18:22:06 yeah probably Sep 02 18:22:08 like txt file busy errors Sep 02 18:22:40 still looking at a good mechanism to avoid having to pull the git every time, probably will create a local mirror (autobuilder always starts with a clean slate) Sep 02 18:23:16 native.bbclass: STAGING_DIR_JAVA = "${STAGING_DATADIR_JAVA_NATIVE}" Sep 02 18:23:25 a.bbclass: STAGING_DATADIR_JAVA ?= ${STAGING_DIR_JAVA} Sep 02 18:23:32 STAGING_DATADIR_JAVA_NATIVE ?= ${STAGING_DATADIR_NATIVE}/java Sep 02 18:23:43 thats from java.bbclass Sep 02 18:24:17 so who sets what here I am confused seems like a loop Sep 02 18:27:34 guess the ?= will break it but no idea how it gets evaluated Sep 02 18:27:50 Robert Schuster could help Sep 02 18:28:28 that was the name i was about to pop Sep 02 18:28:34 something are installed in /java and some are in /x86_64-linux/usr/share/java/ Sep 02 18:28:51 I am convinced that its wrong Sep 02 18:32:17 kicked off a bake for 4.5 but will only test it tomorrow morning (machine is at work and off, I am at home) Sep 02 18:34:10 hi Sep 02 18:35:12 khem there is no loop, read carefully, it is the maze of twisty little passages all different Sep 02 18:35:32 3rd one references staging_datadir_native, while 2nd lline sets staging_datadir_java Sep 02 18:36:57 STAGING_DATADIR_JAVA is set to STAGING_DIR_JAVA is set to STAGING_DATADIR_JAVA_NATIVE which is set to ${STAGING_DATADIR_NATIVE}/java Sep 02 18:37:03 so now do you see the loop Sep 02 18:38:29 no, read carefully Sep 02 18:38:39 the first and the last are not the same Sep 02 18:38:49 STAGING_DATADIR_JAVA Sep 02 18:39:00 STAGING_DATADIR_NATIVE Sep 02 18:43:01 ~seen jama Sep 02 18:43:13 jama <~martin@161-24.13.24.78.awnet.cz> was last seen on IRC in channel #openmoko-cdevel, 9d 21h 57m 54s ago, saying: 'tomorrow I have to arrive to work before 7 :/'. Sep 02 18:43:34 jama no online for more than a week ? Sep 02 18:46:18 eFfeM: yes indeed Sep 02 18:51:29 why do we have java dirs in parallel to sysroots Sep 02 18:51:44 wouldnt they fit into the sysroots Sep 02 18:55:06 i've no idea Sep 02 18:55:13 our classes are full of mysteries to me Sep 02 18:56:00 some jars are in this java dir and others are in sysroot/usr/share/java Sep 02 18:56:09 ugh Sep 02 19:00:05 yeah saw your message on the list Sep 02 19:00:29 there seems to be lack of consistency thats all Sep 02 19:00:32 actually things like python and perl also are tricky when it comes to directories Sep 02 19:01:49 khem: here (x86 host with ubuntu server 9.10, armv5 target, angstrom-2008.1 distro) bitbake gnumail fails at libecj-bootstrap-3.6-r0 Sep 02 19:02:14 are some java tools needed on the host ? Sep 02 19:03:26 ericben: do you have something in ASSUME_PROVIDED Sep 02 19:03:38 complete log here : http://pastebin.ca/1931328 Sep 02 19:04:05 khem: in local.conf ? Sep 02 19:04:11 yes Sep 02 19:04:18 nothing Sep 02 19:04:27 ASSUME_PROVIDED is not defined Sep 02 19:04:53 ok Sep 02 19:04:58 then it should not have failed Sep 02 19:08:30 if anyone has a problem with unpacking or patching in general after that patch went in, let me know. given tom's extensive patchall test i'm not too worried, but given it touched core classes, always a certain amount of concern Sep 02 19:08:35 * kergoth_ ponders Sep 02 19:08:53 kergoth_: ok Sep 02 19:11:28 kergoth, started a build of console-image for sheevaplug about 15 minutes ago Sep 02 19:12:18 it is now building all kind of fine things that are really useful for consoleimage like libxcb, libx11 , xproto and lots of other x11 junk Sep 02 19:12:27 yikes Sep 02 19:12:32 that's just not right Sep 02 19:12:41 it is just a console image, nothing special, nothing added Sep 02 19:12:49 3565 tasks Sep 02 19:13:10 heh, fixing that is going to be one of those not fun tasks that involves wading through and modifying a whole lot of recipes Sep 02 19:13:16 maybe a candidate to go on the janitors page Sep 02 19:13:19 i know insane.bbclass drags in some junk, but not sure if it was that much Sep 02 19:13:26 (though it requires more expertise than most of the ones there currently) Sep 02 19:14:07 kergoth for some reason tasks that impact a whole lot of recipes somehow became less appealing to me recently Sep 02 19:14:32 hehe Sep 02 19:14:33 i wonder why Sep 02 19:15:04 I'll wait a while, until autumn, then I can use the heat it generates to warm my house :-) Sep 02 19:15:19 * kergoth_ hasn't yet given up on making invasive class changes where appropriate, let people whine if they want to Sep 02 19:16:21 oh, if it is needed, I still do it, don't think my reputation can become much lower in the eyes of some people Sep 02 19:17:18 you know, the people who always complain but who themselves do not care about following policies Sep 02 19:17:27 we can't afford to let a minority cause the project to stagnate Sep 02 19:17:44 on the flipside, we need a good method to test big changes without impacting others - testing and/or next branches are likely a good route for it Sep 02 19:17:47 agree Sep 02 19:17:59 the problem is, if i email out a class change, everyone is scared to ack it Sep 02 19:18:05 doesn't get tested Sep 02 19:18:05 that is where the testing-next could be helpful Sep 02 19:18:06 hehe Sep 02 19:18:15 yeah, i think testing will be a big help Sep 02 19:18:49 not unlike git and other projects use of a "next" branch for upcoming stuff, but that method only works if people actually test next Sep 02 19:18:58 i have two class patches out for 3 weeks, bumped them today got an ack on mickey for one of them, no feedback on the other one Sep 02 19:20:00 huhu Sep 02 19:20:19 i'm trying to write a note on quality, hope it can be used to create some awareness and discussion, which eventually might lead to some actions (hope :-) ) Sep 02 19:21:00 is there a dedicated opkg channel? :) or where would one cry out on that topic? :> Sep 02 19:21:03 discussion is always good Sep 02 19:22:22 agree Sep 02 19:22:31 btw who is responsible for stagemanager: I get: Sep 02 19:22:53 cp: cannot stat `/home/frans/workspace/gtd/tmp/work/x86_64-linux/stagemanager-native-0.0.1-r13/sysroot-destdir//home/frans/workspace/gtd/tmp/*': No such file or directory Sep 02 19:23:09 guess this is from do_compile or do_install Sep 02 19:23:30 not sure, i'd suspect RP though Sep 02 19:23:49 will check out the log Sep 02 19:28:33 has anyone encountered a case where a postinst hook was failing because of a timing issue? i.e. opkg tries to execvp it, but returns with error code -1 and errno "No such file or directory", but if you after that ls the file it wanted to execute - its actually there Sep 02 19:28:41 muts be some sort of weird timing issue Sep 02 19:30:10 http://www.colbertnation.com/the-colbert-report-videos/351570/august-24-2010/the-word---control-self-delete is great Sep 02 19:32:57 kergoth, btw is there a simple way to get a list of tasks (apart from -g) Sep 02 19:33:27 laughing my ass off with that video Sep 02 19:33:35 there's -n, but its weirdly slow Sep 02 19:33:48 i guess its forking off the worker processes even though they're not doing anything :) Sep 02 19:38:02 -g gives quite a lot of info which is sometimes useful, but I would just want to get the transitive closure of the DEPENDS (and maybe the RDEPENDS) Sep 02 19:42:28 interesting: console image is now building cairo Sep 02 19:42:45 and vala Sep 02 19:42:53 i don't think "interesting" is the word I've have chosen Sep 02 19:42:59 :-) Sep 02 19:43:20 and I want just a small rootfs that will give me a shell and some basic commands Sep 02 19:44:03 somehow it is interesting, I see things pass by that I didn't know so sometimes I look up a package and learn something new Sep 02 19:44:49 maybe micro-image or something would be better Sep 02 19:45:39 yeah I know, actually if I really wanted it to be small I would make my own, actually I do so at work, typically only busybox and a few utils Sep 02 19:56:56 btw in an image tried to add a depends on another image earlier today (the one image file needs to go into the other one), but bitbake 1.10 ended up in an endless loop; does that sound familiar? Is there a solution for this (don't have the log handy) Sep 02 20:00:23 doesn't ring any bells Sep 02 20:05:04 was at work not sure if I still have a log Sep 02 20:05:05 * eFfeM checks Sep 02 20:05:59 what's the best way to handle multiple machines? have an overlay for each machine with each a different local.conf and then set the right variables before calling bitbake or is there an easier way? Sep 02 20:07:33 no it is not in the log, it was a bitbake error, didn't save it will check if it reappears Sep 02 20:08:18 klaas__: I have one local.conf and script that set MACHINE and a few other vars Sep 02 20:08:35 aha so just not set the MACHINE in the local.conf itself Sep 02 20:08:35 then when swithcing I do Sep 02 20:08:37 roger ;) Sep 02 20:08:44 . script Sep 02 20:08:56 aha didn't think of that :) Sep 02 20:09:12 example for beagleboard: Sep 02 20:09:13 export BBPATH=/home/frans/oe/openembedded/conf:/home/frans/oe/openembedded Sep 02 20:09:13 export PATH=/home/frans/oe/bitbake/bin:$PATH Sep 02 20:09:13 export MACHINE="beagleboard" Sep 02 20:09:13 export BB_ENV_EXTRAWHITE="MACHINE" Sep 02 20:09:18 rest is in local.conf Sep 02 20:09:36 actually in local.conf if you want to you can set TMPDIR to e.g. tmp_${MACHINE} if you prefer Sep 02 20:09:54 ofc you can also set DISTRO in the script Sep 02 20:10:04 well both machines will be the same chip so that won't be needed :) Sep 02 20:10:56 might be useful to have something like bitbake -target beagleboard that then reads a specific conf? Sep 02 20:11:13 bitbake doesn't know about machine Sep 02 20:11:17 MACHINE is an OE construct Sep 02 20:11:26 ah hmm bummer then Sep 02 20:11:40 but conf/machine/${MACHINE}.conf is arleady parsed Sep 02 20:11:46 so i don't really see what you're trying to accomplish Sep 02 20:12:40 just trying to overcome the extra setting of the MACHINE when building for another machine but now that I think of it, it all comes down to an extra command anyway :) Sep 02 20:13:03 even if you had some extra flag you'd still had to issue 2 commands Sep 02 20:13:29 being to eager to ask questions i guess Sep 02 20:18:54 So yeah, at the high level Sep 02 20:19:08 You can't really do more than one host os/target os/target hardware per bitbake invocation Sep 02 20:19:24 But if stuff is organized well, bitbaking for a second machine shouldn't take long Sep 02 20:34:09 kergoth wrt dependencies, blues drags in gst-plugins-base which drags in things like virtual/libx1, gnome-vfs, pango, x264, libx4 etc etc Sep 02 20:34:18 s/blues/bluez4/ Sep 02 20:45:30 klaas__: MACHINE=a bitbake ; MACHINE=b bitbake Sep 02 20:45:45 thats how simple it can get but no simpler :) Sep 02 20:46:47 eFfeM: yes thats a thing I saw too Sep 02 20:53:46 khem, back in office Sep 02 20:59:29 ka6sox-work: ok Sep 02 20:59:44 ka6sox-work: thx, I seem to be all set up Sep 02 20:59:56 excellent Sep 02 21:02:41 is it really needed to rebuild when I just move my tmpdir? Sep 02 21:03:01 yes Sep 02 21:03:34 it's not generally OE causing the issue, stuff it builds hardcodes paths, i.e. in #! lines of generated scripts coming out of ./configure Sep 02 21:03:34 ok so just hacking saved_tmpdir will get me in to trouble then later on? Sep 02 21:03:42 saved_tmpdir is there for a reason Sep 02 21:03:46 ah Sep 02 21:03:48 tought so :) Sep 02 21:04:11 k then let's hope we won't be moving our tmpdir any time soon Sep 02 21:07:33 btw kudos to all, oe is great once you get the hang of it Sep 02 21:24:54 03Mickaël CHAZAUX  07org.openembedded.dev * r66bff3a9a8 10openembedded.git/ (6 files in 5 dirs): Sep 02 21:24:54 Add i.MX51 based "Babbage" board support Sep 02 21:24:54 This is based on the patches released by Freescale as part Sep 02 21:24:54 of LTIB. Sep 02 21:24:54 Signed-off-by: Mickaël Chazaux Sep 02 21:24:56 03Florian Boor  07org.openembedded.dev * r261676fed0 10openembedded.git/: Merge branch 'imx' into org.openembedded.dev Sep 02 22:13:38 Hello Sep 02 22:13:51 I have a hard problem here Sep 02 22:14:39 When i was compiling angstrom distro i modified two files,of same,to create another distro,after this,the system does not recognize the compiled packages! Sep 02 22:21:03 any help? Sep 02 22:41:06 fixed Sep 02 23:16:12 03Tom Rini  07org.openembedded.dev * ra5ee147154 10openembedded.git/recipes/perl/ (perl_5.10.1.bb perl_5.8.8.bb): Sep 02 23:16:12 perl: Use CCLD not ${TARGET_SYS}-gcc in do_compile Sep 02 23:16:12 It's possible that CCLD != ${TARGET_SYS}-gcc once everything is Sep 02 23:16:12 evaluated and CCLD is for when you're using CC for LD. Sep 02 23:16:12 Signed-off-by: Tom Rini Sep 02 23:27:32 i don't know in what channel i ask..but anyone knows good window manager for 240x320 screen ? out illume and opie Sep 02 23:30:57 03Florian Boor  07org.openembedded.dev * r1cc2b1cd63 10openembedded.git/conf/machine/simone.conf: simone.conf: Add jffs2 parameters Sep 02 23:52:25 i make a angstrom compilation..if i change the DISTRO the system will need re-build all packages to new DISTRO ? Sep 02 23:57:48 angelox_123, if you change distro, you should start with a fresh tempdir Sep 02 23:58:13 and it will rebuild all yeah? Sep 02 23:58:44 if you (re)move the tmp dir, it will have no choice but the rebuild all Sep 02 23:58:49 if you remove the tempdir the way grg just suggested, how would it not? all output goes into the tmp dir Sep 03 00:03:57 03Khem Raj  07org.openembedded.dev * r697af28264 10openembedded.git/recipes/startup-notification/startup-notification_0.9.bb: Sep 03 00:03:57 startup-notification_0.9.bb: Add evas to DEPENDS Sep 03 00:03:57 Signed-off-by: Khem Raj Sep 03 00:39:21 Should a program written in C++ RDEPEND upon libstdc++ ? Or is there another mechanism that is used to determine this dependency? Sep 03 00:40:24 (i built an image with midori in it, but it wont run as the resultant image has no libstdc++.so.6) Sep 03 00:41:06 yeah libstdc++ dependencies are missing somehow in ipks Sep 03 00:41:24 its probably due to debian renaming Sep 03 00:41:32 otherwise shlibs should have noticed it Sep 03 00:41:55 when building a package that links with libstdc++ Sep 03 00:48:01 * grg googles shlib Sep 03 01:32:31 khem: rdeps are adjusted for the renaming. debian.bbclass sets PKG_ to the name, which gets saved as pkgdata and used in runtime_mapping_rename in package.bbclass, line 289 Sep 03 01:38:39 i make gpe-image but it don't start..only blink gpe-login Sep 03 01:55:41 angelox_123, what happens when you type '/etc/init.g/gpe-dm start' on the console after logging in? Sep 03 01:56:13 login by ssh or by gpe ? Sep 03 01:56:36 you have a graphical login screen? Sep 03 01:56:46 no Sep 03 01:56:53 don't start Sep 03 01:57:03 remembering i compiled with DISTRO - Angstrom Sep 03 01:57:14 login on the serial console, or ssh. It doesn't matter. Sep 03 01:57:40 that should have been: /etc/init.d/gpe-dm start Sep 03 01:58:27 i'll try Sep 03 01:59:05 you may have to play around with your system to get Xorg to work properly. I'm too lazy for that and just use kdrive as my xserver, which works better for me than the angstrom default Sep 03 01:59:41 yes.it uses kdrive Sep 03 01:59:55 it show X cursor of Xserver..but don't start gpe Sep 03 02:00:34 do you know any cool window manager for embedded devices ? Sep 03 02:01:14 no. I know lots of bad ones though Sep 03 02:01:33 good DISTRO ? Sep 03 02:01:46 trial and error? Sep 03 02:01:59 what? Sep 03 02:02:12 i've been looking at fvwm on my mipsel touchscreen device recently. Sep 03 02:02:44 its nice and lightweight, and very configurable Sep 03 02:03:31 do you think it will is beautifully on 240x320 device ? Sep 03 02:03:45 no Sep 03 02:04:02 maybe look at opie? Sep 03 02:04:24 i tried Sep 03 02:04:44 compiled nice..but don't start Sep 03 02:04:49 my device is EZX Sep 03 02:05:28 the gpe start tslib calibration Sep 03 02:05:31 yeah, opie didn't start for me either. qpe is broken. Sep 03 02:05:48 xtscal is still broken too Sep 03 02:06:09 i keep having to copy over a /etc/pointercal file from my old builds from the start of the year Sep 03 02:07:52 khem, i noticed that libiberty is included with binutils, gcc and gdb. Is there any difference between them and how do you choose which one gets installed? I get the following when bitbaking meta-toolchain: http://tinderbox.openembedded.org/public/logs/task/7558530.txt Sep 03 02:59:43 how can I integrate a new toolchain **** ENDING LOGGING AT Fri Sep 03 02:59:57 2010