**** BEGIN LOGGING AT Thu Jul 07 02:59:56 2011 Jul 07 06:45:37 hi all :) Jul 07 07:15:59 how to say that two recipes provides similar things but they cannot coexists ? Jul 07 07:16:13 is this fine for OE ? Jul 07 07:18:48 alx? Jul 07 07:18:50 explain Jul 07 07:19:52 woglinde, qwt 6 is the new release of qwt (API and ABI changed from 5.2). Jul 07 07:20:04 yes Jul 07 07:20:14 so libs differs? Jul 07 07:20:27 to qwt5 Jul 07 07:20:29 hm hm Jul 07 07:20:35 the compilation of that package is troublesome: so I was supposed to write a qwt_6.0.0.bb and a qwt-e_6.0.0.bb Jul 07 07:20:45 either you hack both so they can coexist Jul 07 07:20:58 one for qt4x11 one for qt4e. Jul 07 07:20:58 or you superceed qwt5 Jul 07 07:21:08 i'm not interested in qwt5 Jul 07 07:21:19 than make the recipe Jul 07 07:21:22 and pin them Jul 07 07:21:42 but still trying to put things together ... qwt5 qwt5e qwt6 qwt6e Jul 07 07:21:51 that's a pain in .... $$ (sorry) Jul 07 07:22:09 why? Jul 07 07:22:26 the makefiles are somewath hacky Jul 07 07:22:41 going heavily by sed Jul 07 07:23:11 hm why you need qwt at all?= Jul 07 07:24:57 we wrote an application that uses qwt6 for plotting Jul 07 07:25:01 data on screen Jul 07 07:27:29 watching qwt5 recipe i discover about a libsuffix ... but in the new system it isn't used anymore ... :( Jul 07 07:32:41 woglinde, so I end up with thinking to write the two recipes having them mutually exclusive Jul 07 07:32:58 but now I don't know how to do this mutual exclusion. Jul 07 07:36:25 mutually exclusive on the device? Jul 07 07:36:41 or build time Jul 07 07:48:48 woglinde on the device. Jul 07 07:49:27 when I build an image I want to be sure that none of the two, or just one of the two will be shipped Jul 07 07:50:05 look for CONFLICTS in Jul 07 07:50:23 hm but I guess opkg will overwrite qtw5 anyway Jul 07 08:19:21 woglinde, is it possible to have multiple commands executed in bitbake ? Jul 07 08:19:37 like bitbake -c clean configure devshell ? Jul 07 08:19:50 boring of those recipe parsing time Jul 07 08:19:51 ^^ Jul 07 08:45:45 alx if you have all deps already build Jul 07 08:46:08 you could use bitbake -b /pathtorecipe/recipe.bb -c whatever Jul 07 08:52:51 woglinde, I'm doing this, but I'm doing a whole set of bitbake -c clean qwt; bitbake -c configure qwt; bitbake -c devshell qwt Jul 07 08:53:05 this means 3 recipes parse time * thousand of cycles ^^ Jul 07 08:53:27 the key is to specify the path to the recipe Jul 07 08:53:48 oh ... ok :) Jul 07 08:54:04 what's a problem to run devshell and do make clean; ./configure; make there? Jul 07 08:54:28 ynezz, that I'm too noob to do that :) Jul 07 08:54:36 it's all just scripts Jul 07 08:54:37 well.. testing the do_configure Jul 07 08:54:43 part of a recipe Jul 07 08:54:57 so .. it's different Jul 07 08:55:01 so you can run temp/run.do_configure Jul 07 08:55:28 running that is the same ? Jul 07 08:55:44 if you don't change the recipe, then it should be Jul 07 08:55:52 I use it all the time Jul 07 08:56:02 I change the recipe at every cycle Jul 07 08:56:29 hm, I don't understand why Jul 07 08:56:55 placing a whole bunch of sed in the recipe before real configure Jul 07 08:57:16 you can modify that run.do_configure directly Jul 07 08:57:29 it's just a shell script Jul 07 08:57:55 there's last line in that script, do_configure Jul 07 08:58:02 add your sed kungfu there Jul 07 08:58:19 I for example prefer to patch the stuff Jul 07 08:58:36 you are showming me a whole new world :) Jul 07 08:58:44 but ... it's late :( Jul 07 08:58:50 and use that patch instead of that sed crap Jul 07 08:58:51 all my sed kungfu works now Jul 07 08:58:52 :D Jul 07 08:59:14 it's a maintaince nightmare, it should be forbidden... Jul 07 08:59:45 I used sed for writing correct ${D} on Jul 07 09:00:10 s/maintaince/maintenance/ Jul 07 09:00:16 rotlf Jul 07 09:24:06 ynezz, how to solve a QA issue ? Jul 07 09:24:10 WARNING: QA Issue with qwt: non -dev package contains symlink .so ... Jul 07 09:25:31 alx_: change your FILES_xxx Jul 07 09:25:45 should be *.so.* instead of *.so* Jul 07 09:25:52 (I'm guessing) Jul 07 09:26:16 bluelightning, didn't get. I'm writing a recipe for a library (qwt 6.x) Jul 07 09:26:20 yep Jul 07 09:26:30 so what FILES_ definitions do you have in your recipe? Jul 07 09:28:01 FILES_${PN}-examples = "${bindir}/*" Jul 07 09:28:01 FILES_${PN} += "${datadir}/${QT_DIR_NAME}/mkspecs/features/*" Jul 07 09:28:23 and RPROVIDES_${PN}-dev = "libqwt-dev" Jul 07 09:29:03 hmm Jul 07 09:29:25 (gently inspired from qwt 5.2.x OE recipe) Jul 07 09:29:50 alx_: maybe ${bindir}/* contains some .so symlinks? Jul 07 09:30:24 ${libdir} contains symlinks Jul 07 09:30:28 that is, .so and not .so.* Jul 07 09:30:46 yeah naturally Jul 07 09:30:58 but the standard FILES_${PN} will not include the .so files Jul 07 09:31:00 also .so.* Jul 07 09:31:10 so it must be one of the definitions you have in your recipe Jul 07 09:31:44 bitbake -e ? Jul 07 09:31:47 :D Jul 07 09:32:19 sure, bitbake -e qwt | grep ^FILES_ Jul 07 09:33:19 # USERNAME=alessandro Jul 07 09:33:19 export USERNAME="alessandro" Jul 07 09:33:19 # FILES_qwt=${bindir}/* ${sbindir}/* ${libexecdir}/* ${libdir}/lib*${SOLIBS} ${sysconfdir} ${sharedstatedir} ${localstatedir} ${base_bindir}/* ${base_sbindir}/* ${base_libdir}/*${SOLIBS} ${datadir}/${PN} ${libdir}/${PN}/* ${datadir}/pixmaps ${datadir}/applications ${datadir}/idl ${datadir}/omf ${datadir}/sounds ${libdir}/bonobo/servers ${datadir}/${QT_DIR_NAME}/mkspecs/features/* Jul 07 09:33:39 sorry for big paste :$ thinking it was smaller Jul 07 09:34:58 the culprit is ${libdir}/${PN}/* Jul 07 09:36:33 alx_: right... hmm Jul 07 09:36:37 so how did that get in there? Jul 07 09:37:17 two library and 6 symlinks Jul 07 09:37:31 two .so :) Jul 07 09:37:55 actually it's correct... what was I thinking Jul 07 09:38:05 symlyncs are 3 per library: so -> so.6 -> so.6.0 -> so.6.0.0 (real lib) Jul 07 09:38:26 the way it's supposed to work is FILES_${PN}-dev is supposed to take away the .so files before FILES_${PN} can include them Jul 07 09:38:48 to avoid name pollution I put libs in ${PN} so the previous version (5.2) isn't touched and they may coexist Jul 07 09:39:14 ok Jul 07 09:39:29 you may need to add something to FILES_${PN}-dev Jul 07 09:40:06 like ${libdir}/${PN}/*.so Jul 07 09:41:28 why in dev ? Jul 07 09:41:51 why don't I put .h in dev and keep .so in ${PN} ? Jul 07 09:42:09 try to remove the FILES line Jul 07 09:42:18 seems you have a normal setup Jul 07 09:42:22 and dont need any FILES_ Jul 07 09:43:03 hm oh Jul 07 09:43:09 looking at the inc Jul 07 09:43:09 PACKAGES_prepend = "${PN}-examples " Jul 07 09:43:21 should be better PACKAGES += Jul 07 09:45:46 the FILES is needed because FILES_${PN} += "${datadir}/${QT_DIR_NAME}/mkspecs/features/*" this is needed (at least in -dev) Jul 07 09:48:07 alx_: .so symlinks with no version suffix are not necessary for normal operation and could even be harmful Jul 07 09:48:34 bluelightning, oh ok! Jul 07 09:49:13 indeed, you don't actually need any symlinks at all Jul 07 09:49:20 but the .so files certainly belong in the -dev package Jul 07 09:49:54 isn't lib itself in${PN} and headers and symlinks in ${PN}-dev ? Jul 07 09:50:13 only the .so symlink belongs in -dev Jul 07 09:50:21 the other symlinks, if you wish to have them at all, belong in ${PN} Jul 07 09:50:58 the default FILES and PACKAGES from bitbake.conf will get that all right for you automatically, it's best to just leave it alone. Jul 07 09:51:19 if you need to add stuff then that's fine, but don't override any of those variables entirely Jul 07 09:51:38 and, in particular, try to refrain from changing the order of the PACKAGES Jul 07 09:52:21 pb_, I got the PACKAGES from qwt recipe Jul 07 09:53:10 that's no guarantee that they are correct Jul 07 09:53:34 loads of the recipes in oe.dev, in particular, were written by crazed h4x0rs and contain all manner of lunacy Jul 07 09:54:03 s/crazed/cursed/ Jul 07 09:54:06 ^^ Jul 07 09:54:06 qwt might be fine, I dunno, but I wouldn't count on it Jul 07 09:54:41 I count on qwt (5.2) based that I'm writing qwt (6.0) Jul 07 09:54:42 gm pb Jul 07 09:54:44 btw. Jul 07 09:54:52 hi woglinde Jul 07 09:54:55 what is gm ? :) Jul 07 09:55:05 lol .. I got it late :) Jul 07 09:57:04 btw .... anyone has control over patch.openembedded.org ? Jul 07 09:57:30 patches., you mean? Jul 07 09:57:35 yes Jul 07 09:57:44 what's the matter, is it out of control at the moment? Jul 07 10:00:02 I'm not into that. Just wondering how to recover my account because there is no "Recover password" not "Activate account" and timeout occurred when I was subscribing so now (1) cannot register again because a user already exists and (2) cannot log-in bacause account is not activated :D :D Jul 07 10:01:45 ah, that is a bit sad Jul 07 10:02:02 I'm not quite sure who's responsible for patchwork, khem maybe Jul 07 10:02:07 03Richard Purdie  07master * r54fe91fe96 10bitbake.git/lib/bb/ (cache.py cooker.py): Jul 07 10:02:07 cache.py: Ensure additional .bbappend files are accounted for Jul 07 10:02:07 Currently if a user adds a new .bbappend file to the system, the cache still Jul 07 10:02:07 thinks the cached data is valid. This code fixes that to ensure additions and Jul 07 10:02:07 changed in append application order are accounted for. Jul 07 10:02:07 [YOCTO #1091] Jul 07 10:02:08 Signed-off-by: Richard Purdie Jul 07 10:02:08 03Richard Purdie  07master * rc7a9ef70ba 10bitbake.git/lib/bb/server/process.py: Jul 07 10:02:09 bitbake/process.py: Ensure queued UI events are queued right before we add our own handler Jul 07 10:02:09 Signed-off-by: Richard Purdie Jul 07 10:02:33 gogo rp Jul 07 10:03:07 yeah the first is a bug I definitely noticed Jul 07 10:07:36 hello Jul 07 10:08:25 when I run " Xorg :1 -once -query " ... I m getting glx, dri, dri2 modules doesnot exit... Jul 07 10:08:46 which package in reciepe provides this modules? Jul 07 10:08:59 I m newbei to it.. can anyone help me... Jul 07 10:16:52 rocker what is your machine? Jul 07 10:17:03 btw. you can disable the modules in your xorg.conf Jul 07 10:18:18 ARM mini2440 Jul 07 10:18:39 woglinde, xorg.conf is completle blank... Jul 07 10:18:45 yes Jul 07 10:19:04 I had created x11-image through bitbake Jul 07 10:19:39 create an xort.conf Jul 07 10:19:45 ups xorg.conf Jul 07 10:19:47 Section "Module" Disable "dri" Jul 07 10:19:48 EndSection Jul 07 10:19:55 args Jul 07 10:19:59 Section "Module" Jul 07 10:20:04 Disable "dri" Jul 07 10:20:08 EndSection Jul 07 10:20:15 you can disbale glx too there Jul 07 10:22:06 for the mini only software 3d rendering would work Jul 07 10:23:44 woglinde, its working .... btw what is the use of this modules? Jul 07 10:23:52 3d Jul 07 10:24:02 graphics Jul 07 10:25:15 woglinde, i had compiled Thincclient for mini2440.... and I m getting .... Couldn't find suitable visual class 4, depth 24 Jul 07 10:26:08 what is thincclient? Jul 07 10:26:13 woglinde, this is the links http://wiki.ncl.cs.columbia.edu/wiki/THINCGetCompileInstallRun#Linux_and_Java_Clients Jul 07 10:26:25 hm maybee you need to force the xorg to 16 bit Jul 07 10:26:58 but how? Jul 07 10:29:02 -depth? Jul 07 10:29:51 rocker the wiki page is outdated in some parts Jul 07 10:30:40 rocker you better what to look at freenx or x2go Jul 07 10:30:46 ups want Jul 07 10:30:53 for using xorg remotly Jul 07 10:31:23 woglinde, ya... actually before cross compiling I had to modify some source to work on my host machine..... it precompiled binary is working but if we compile a source it is not working,.... Jul 07 11:09:59 I guess we need ipv4 in our default DISTRO_FEATURES now in oe-core after the recent change to busybox... am producing a patch now Jul 07 11:16:07 heh, yeah, I guess so Jul 07 11:16:29 I suppose that must mean that nobody is testing uclibc at all with the default DISTRO_FEATURES since it also respects that flag Jul 07 11:19:39 yes Jul 07 11:19:55 interesting that the change went in a month ago and nobody noticed Jul 07 11:20:03 yeah Jul 07 11:20:10 it broke ifup at least that I noticed Jul 07 11:20:26 well, I guess few people (if any) are likely to be using the distroless configuration Jul 07 11:20:40 poky doesn't have ipv4 in DISTRO_FEATURES though Jul 07 11:20:49 oh, that is a bit weird then Jul 07 11:20:54 or rather, it doesn't add it (it relies on the defaults) Jul 07 11:20:59 maybe all the poky crew have embraced ipv6 :-) Jul 07 11:21:03 heh Jul 07 11:21:16 would be nice to be able to be that forward-thinking :) Jul 07 11:30:01 hm, I wonder how hard it would be to exhume my mythtv images for epia. I guess they must be quite bit-rotten by now. Jul 07 11:30:15 * pb_ fed up with garbage dvb-t receiver boxes Jul 07 11:33:21 pb_: epia won't manage HD though, will it? Jul 07 11:33:38 no, but I don't have an HD-capable TV either :-} Jul 07 11:33:59 No, me neither, but I'm future proofed :D Jul 07 11:34:43 intressting Jul 07 11:34:53 we dont have a hd-tv too Jul 07 11:34:57 I'm not sure anybody is actually broadcasting HD on DVB-T at the moment anyway Jul 07 11:35:06 iirc, you need satellite for that which I don't have either Jul 07 11:35:06 no in de for sure Jul 07 11:35:13 ups not in .de Jul 07 11:35:14 I have DVB-S and DVB-T Jul 07 11:35:19 we sat cable and iptv Jul 07 11:35:23 +have Jul 07 11:36:04 I think they'll wait for DVB-T2 before rolling out HD in the UK. Jul 07 11:36:32 oh, wikipedia says freeview HD in "all regions by the end of 2012" Jul 07 11:38:35 what resolution are they using for HD on freesat, is it 1080p25? Jul 07 11:39:32 I dont know Jul 07 11:49:56 apparently the bbc is using 1440x1080i50, for reasons that are somewhat obscure to me Jul 07 11:52:28 lol Jul 07 11:53:52 thast like our goverment/tax tv ard and zdf sticked to 720p hdtv while all other used 1080i Jul 07 12:03:05 1 1080i or 2 720p's on the same channel...and 720p looks better. Jul 07 12:03:18 ka6sox-away sure Jul 07 12:03:26 when the material was produced in 720p Jul 07 12:03:34 which was not the case Jul 07 12:03:45 ah Jul 07 12:03:53 well..the TVs scale anyways Jul 07 12:09:25 hi stefan Jul 07 12:09:41 woglinde: hi Jul 07 12:26:58 can anyone help me how to compile glx module using bitbake-oe? Jul 07 12:29:49 rocker mesa should do it Jul 07 12:30:08 bitbake mesa Jul 07 12:49:04 still hate qwt, but seems working :) Jul 07 12:55:25 AIGLX error: dlopen of /usr/lib/dri/swrast_dri.so failed Jul 07 12:55:47 I m getting this error Jul 07 13:07:33 ka6sox-away: yeah, I can't think why they would have chosen 1080i50 rather than 1080p25. AFAIK there are no CRTs (and hence no interlaced displays) that can display 1080 lines at all, so it seems a bit weird to stick with interlacing for the broadcast format. oh well, no doubt they have their reasons. Jul 07 13:08:03 isn't Pal 625lines? Jul 07 13:08:12 yeah Jul 07 13:08:30 or 576 active lines Jul 07 13:15:15 pb_, these "standards" don't really fit the real world of what the displays are capable of Jul 07 13:17:05 yeah, indeed, but (afaik) all modern displays are progressive scan and I can't think of a single instance where giving them interlaced content (and hence forcing them to rely on their own deinterlacing capabilities, which might or might not be any good) would be better than deinterlacing prior to broadcast where you have (presumably) more resources to throw at the problem Jul 07 13:17:44 and, of course, for content which was originally produced in progressive scan, you end up having to interlace it for broadcast only to have the TV deinterlace it again for display. presumably recovering the progressive frame is not very hard in that case, but it still seems a bit silly to go through that process. Jul 07 13:18:55 I guess maybe you get better motion effects for sports on at least some TVs if you broadcast it interlaced, maybe that's it. Jul 07 13:25:15 anyone Jul 07 13:25:17 ? Jul 07 13:25:26 4~no Jul 07 13:29:53 woglinde, lol Jul 07 14:10:54 does anyone have any tips for building java and classpath with the 2011.03-maintenance branch? Jul 07 14:40:22 gchiii I answered you email Jul 07 14:40:57 if you dont tell us whats going wrong we cannt help Jul 07 14:42:35 woglinde: sorry, I'm not entirely sure what is going on; but classpath builds the log shows lots of errors but bitbake goes on like it just succeeded Jul 07 14:43:51 the configure log shows strange results for the java: checking if javac -Xlint:unchecked,cast,divzero,empty,finally,overrides works... yes Jul 07 14:43:51 checking if java works... LOG: [0x00007f3c8be85740] vm_shutdown Jul 07 14:49:04 how come tslib says it's compiled with a different version of event protocol ? Jul 07 14:50:17 gchiii that comes from jamvm Jul 07 14:50:28 oh no from cacao Jul 07 14:50:55 gchiii sorry dont have the time yet do help Jul 07 14:51:02 maybee in 3 hours Jul 07 14:52:33 ok i should be here all day Jul 07 15:07:21 ehm... may someone help in tracking a problem in tslib ? Jul 07 15:07:28 (OE recipe...) Jul 07 15:07:46 any particular problem? Jul 07 15:08:10 running uImage + rootfs Jul 07 15:08:31 on both real device and qemu - i get this tslib: Selected device uses a different version of the event protocol than tslib was compiled for Jul 07 15:09:53 so, that means the evdev protocol that your kernel is using doesn't match the headers that tslib was built against Jul 07 15:10:46 how is it possible if I didn't git pull while building virtual/kernel and rootfs image Jul 07 15:11:09 presumably because your linux-libc-headers is newer than your actual kernel, or something of that kind Jul 07 15:11:19 what versions are you using, specifically? Jul 07 15:11:42 how to check ? Jul 07 15:12:19 ls tmp/work? Jul 07 15:12:36 ook Jul 07 15:13:32 I found there 4 dir... where to check ... native or cross ? Jul 07 15:14:02 I've all-angstrom .. armv4t mini2440 and x86_64 Jul 07 15:15:42 pb_, version of what ? Jul 07 15:20:42 of the kernel Jul 07 15:21:59 linux-mini2440-2.6.31+git-r0 Jul 07 15:22:19 ok, and what linux-libc-headers? Jul 07 15:22:59 find will say ^^ Jul 07 15:23:09 linux-libc-headers-2.6.37-r0 Jul 07 15:23:19 mmm Jul 07 15:23:30 how come they don't match ? Jul 07 15:23:56 that'd be a question for your distribution folks Jul 07 15:24:04 angstrom ? Jul 07 15:24:09 right Jul 07 15:24:43 maybe noone updated mini2440 as it seems not so in use... oe-core don't have any support for it up to now :) Jul 07 15:24:53 suggestion on how to patch this ? Jul 07 15:26:49 get a newer kernel? Jul 07 15:27:02 I'll try to :) Jul 07 15:27:14 or get kergoth to fix his lame compatibility logic :-} Jul 07 15:28:04 or back out your linux-libc-headers to a version <= 2.6.31 Jul 07 15:28:04 * kergoth blames russell and/or doug lowder and runs Jul 07 15:28:20 (or prior to whatever version they changed the EV_VERSION in Jul 07 15:28:21 ) Jul 07 15:28:25 pb_, third and first seem reasonable.. Jul 07 15:28:27 I dunno when that was, offhand Jul 07 15:28:29 how to do third ? Jul 07 15:28:50 iirc current tslib master only checks against a minimum, and will try to use newer if it can Jul 07 15:29:06 if it explodes in your face, though, don't blame me :) Jul 07 15:29:21 on a related note, evfilter looks interesting. i remember thinking about starting a project like that Jul 07 15:30:03 kergoth: yeah, that's his problem. the "minimum" is the EV_VERSION it found at compile time, so it's blowing up because his runtime kernel is older Jul 07 15:30:35 ahh, i see Jul 07 15:30:45 thats unfortunate, it should really go by what its actually compatible with Jul 07 15:31:04 yeah Jul 07 15:32:18 what could explode is I use mainstream kernel ? :D Jul 07 15:34:06 ID:32.1 PSIP:56.4 bitrate=5330000 no-guide Jul 07 15:36:07 bleh...when will they have focus follows eyes Jul 07 15:40:06 I having trouble compiling fire 4.0.1 Jul 07 15:40:39 I'm sorry ... I am having trouble compiling firefox 4.0.1 Jul 07 15:41:08 it complains that there is no make target for -lX11 Jul 07 15:41:37 has any one experienced this problem ... all help is appreciated Jul 07 15:50:02 can anyone help Jul 07 16:05:29 re Jul 07 16:05:47 gchiii please pastebin your local.conf Jul 07 16:06:16 hm wait have to reconnect Jul 07 16:06:40 can anyone help me with compiling firefox 4.0.1 within the openembedded environment Jul 07 16:07:12 error: no rule for -Xll Jul 07 16:07:22 -lX11 Jul 07 16:09:33 woglinde http://pastebin.com/wRP9Qgms Jul 07 16:13:44 woglinde http://pastebin.com/wRP9Qgms Jul 07 16:14:12 http://pastebin.com/zXBtjrB5 Jul 07 16:14:37 http://pastebin.com/0rPkPYc9 Jul 07 16:16:45 hm okay Jul 07 16:16:55 doesnt look this bad Jul 07 16:17:04 but I wonder why you set the java bits yourself Jul 07 16:17:05 hm Jul 07 16:17:30 i set the java stuff, to see if the defaults weren't working, Jul 07 16:18:30 classpath is failing in the tools/asm area of the build Jul 07 16:20:20 hm Jul 07 16:20:49 so now let me see if stefan_schmidt backported the one cacao patch as promised Jul 07 16:22:22 what was your machine armv4? Jul 07 16:23:14 i think so...it is an Atmel at91rm9200 Jul 07 16:23:51 armv4t Jul 07 16:24:03 hm okay Jul 07 16:29:53 gchiii which programm fails? Jul 07 16:29:55 classpath Jul 07 16:31:18 gchiii I will try it here myself Jul 07 16:31:42 1. ERROR in ./external/asm/org/objectweb/asm/MethodVisitor.java (at line 1) Jul 07 16:31:42 /*** Jul 07 16:31:42 ^ Jul 07 16:31:51 hm Jul 07 16:32:24 woglinde, the javac commands trying to build stuff in external/asm/org/objectweb/asm/ Jul 07 16:33:02 I'm not sure if the problem is really with antlr Jul 07 16:33:41 but I was surprised that bitbake didn't recognize that it failed to fully build Jul 07 16:37:27 okay I am now trying a classpath-nativ build Jul 07 16:38:53 testing with qemuarm Jul 07 16:40:30 classpath-native worked fine, but then again it is tweaked to not inlcude the gjdoc stuff which appears to be in this area as well Jul 07 16:40:48 so its classpath itself? Jul 07 16:40:54 yes Jul 07 16:41:07 hm Jul 07 16:41:17 what are you building? Jul 07 16:41:29 bitbake foo I mean Jul 07 16:42:19 bitbake -v classpath Jul 07 16:42:29 hrrr Jul 07 16:42:40 no I mean why you need classpath on device Jul 07 16:43:00 I am nearly sure we dont build it when compiling openjdk Jul 07 16:44:13 openjdk was too slow. jamvm was quicker and required classpath Jul 07 16:44:41 hm okay Jul 07 16:44:45 My Bitbake just sits on do_fetch, but actually doesn't do anything Jul 07 16:45:44 davidlt: its likely fetching, using wget or something, behind the scenes. check the log of hte task, or your ps/top Jul 07 16:46:00 ideally the fetchers would provide progress information, but that's not the case today Jul 07 16:46:06 kergoth_: that's the problem, CPU: 0, logs are empty Jul 07 16:46:40 using little cpu is meaningless, if its I/O bound.. Jul 07 16:47:00 It's heavy enough that you could see that on top Jul 07 16:47:03 gchiii btw. write a recipe for the actual jamvm Jul 07 16:47:10 it should work much better Jul 07 16:47:31 again, check ps. are there anhy child processes of the do_fetch bitbake process? Jul 07 16:47:39 hm oh Jul 07 16:47:39 or is it blocked on something else? Jul 07 16:47:43 no new release Jul 07 16:47:52 I thought he already did Jul 07 16:47:57 woglinde, how do you mean Jul 07 16:48:16 git version of jamvm Jul 07 16:48:41 oh. i didn't think of that Jul 07 16:48:42 gm all Jul 07 16:49:32 http://git.berlios.de/cgi-bin/cgit.cgi/jamvm Jul 07 16:49:38 but thats only for jamvm Jul 07 16:49:49 we need to solve the classpath problem Jul 07 16:50:37 yes. Jul 07 16:51:16 kergoth_: I just see python child processes, but all at zero, no git or anything running Jul 07 16:51:21 weird. Jul 07 16:51:25 wonder what its doing Jul 07 16:51:37 Me to, and logs are empty Jul 07 16:51:50 what process state? Jul 07 16:52:21 davidlt, an extreme measure might be to use strace on the process ids to see what things are doing on a very low level Jul 07 16:52:39 davidlt kill it and start again Jul 07 16:53:11 kergoth_: S, sleeping Jul 07 16:53:40 what was it doing when it entered this state, parsing or running tasks? Jul 07 16:54:30 parsing, I was compiling omap3-console-image and it stopped at kernel Jul 07 16:56:29 i was referring to the up front parse process. compiling and parsing are two completely different aspects of bitbake Jul 07 16:56:45 so the answer is 'it was running tasks' Jul 07 16:58:38 woglinde, are you getting the same problem with classpath? Jul 07 16:59:01 hmm. Jul 07 16:59:15 gchiii I have phenom2 Jul 07 16:59:19 but this not so fast Jul 07 16:59:32 when compiling from scratch Jul 07 16:59:54 ahh yes. i understand Jul 07 17:00:44 kergoth_: it hangds on read(3, ...) according to strace, lsoft shows that as pipe Jul 07 17:01:29 s/lsoft/lsof/ Jul 07 17:02:31 I tried to submit a request for one of the quantum computers from dwave (http://www.dwavesys.com/en/products-services.html) my boss just laughed Jul 07 17:15:06 gchiii: it's single purpose from what I can tell... you'd need a good use for it for $10M :) Jul 07 17:17:14 bluelightning: it's just as well, i couldn't decide which linux distro to run on it anyways Jul 07 17:17:49 gchiii: if it's limited to discrete optimisation I'm not sure you could run Linux on it :) Jul 07 17:18:05 of course I know you were kidding Jul 07 17:18:17 :) Jul 07 17:32:46 where may I find if tegra 2 is supported ? Jul 07 17:32:51 (by oe-core) Jul 07 17:33:50 oe-core won't support it by itself - maybe there is a layer that will, I don't personally know of one Jul 07 17:33:53 its not support by oe-core Jul 07 17:34:05 I will have to start the layer Jul 07 17:34:38 woglinde, I'll love you :) Jul 07 17:35:01 by now I've to understand why inittab respawn S:2345:respawn:/sbin/getty 115200 ttySAC0 too fast :) Jul 07 17:35:29 you dont have a ttySAC0? Jul 07 17:35:35 check if the device was created Jul 07 17:36:57 lol woglinde how to ? it's an image in qemu... it seems not from kernel prints Jul 07 17:37:28 I'd say no Jul 07 17:45:07 how to enable ? Jul 07 17:45:53 does ssh work? Jul 07 17:46:00 you could check it then Jul 07 17:46:15 or use mknod to make it offline Jul 07 17:46:40 no netrwork Jul 07 17:47:22 alx_: you have some patchwork troubles ? Jul 07 17:47:30 khem, Jul 07 17:47:37 yes... no mail to enable my account Jul 07 17:48:37 ok whats your id Jul 07 17:49:10 03Koen Kooi  07org.openembedded.dev * r9895f55921 10openembedded.git/conf/machine/include/ (ti814x.inc ti816x.inc): Jul 07 17:49:10 ti81xx: sync with meta-ti Jul 07 17:49:10 Signed-off-by: Koen Kooi Jul 07 17:49:13 a.sappia Jul 07 17:49:15 03Koen Kooi  07org.openembedded.dev * r2b2145a310 10openembedded.git/recipes/linux/ (10 files in 6 dirs): Jul 07 17:49:15 linux-ti-81xx-psp: import from meta-ti Jul 07 17:49:15 Signed-off-by: Koen Kooi Jul 07 17:49:19 03Koen Kooi  07org.openembedded.dev * r8d76b99953 10openembedded.git/recipes/ti/ti-ipc_1.23.02.27.bb: Jul 07 17:49:19 ti-ipc 1.23.02.27: import from meta-ti Jul 07 17:49:19 Signed-off-by: Koen Kooi Jul 07 17:49:19 03Koen Kooi  07org.openembedded.dev * r913512909c 10openembedded.git/recipes/ti/ (3 files in 2 dirs): Jul 07 17:49:19 ti-syslink 2.00.00.78: import from meta-ti Jul 07 17:49:19 Signed-off-by: Koen Kooi Jul 07 17:49:57 khem, got to go Jul 07 17:52:36 stefan_: ping Jul 07 20:01:41 The sanity checker doesn't like the net/irda/sir-dev.ko kernel module ("QA Issue: kernel-module-actisys-sir rdepends on kernel-module-sir-de\ Jul 07 20:01:48 v") Jul 07 20:04:18 also "QA Issue: kernel-modules rdepends on kernel-module-g-dbgp" Jul 07 20:06:00 khem: ping Jul 07 20:23:14 if I build a native package, such as "foo-native"… how do I reference the foo executable in another recipe? Jul 07 20:23:36 I assume there is some env var with the path Jul 07 20:24:33 yes Jul 07 20:24:43 STAGING_NATIVE Jul 07 20:24:47 woglinde: i think i have figured out why classpath could fail, without make noticing. the wrapper shell script for javac eats up the return code from running javac Jul 07 20:24:58 msm: assuming it was installed to ${D}${bindir} in do_install, it'll be in the PATH Jul 07 20:25:02 no need tos pecify a path at all Jul 07 20:25:02 gchiii yes Jul 07 20:25:12 kergoth yeah Jul 07 20:25:38 gchiii we are knowing that jamvm and cacao fails sometimes Jul 07 20:25:45 ok - it's installed in STAGING_NATIVE? Jul 07 20:25:49 but seems wit newer compilers and systems Jul 07 20:25:55 it crashes more often Jul 07 20:26:00 i need to know more about the stuff it installed.. let me continue Jul 07 20:26:26 I have no idea yet how we can it make more stable in maintaince Jul 07 20:26:49 msm: see conf/bitbake.conf, search for STAGING_BINDIR_NATIVE Jul 07 20:26:57 you'll see all the applicable variables Jul 07 20:29:30 my oe-core build seems to be stuck on NOTE: package linux-yocto-2.6.37+git1+d5d3c6480d61f83503ccef7fbcd765f7aca8b71b_1+ea7bcd9e408ddfaf5ecd0bcc37956998add58e2d-r20: task do_package_write_ipk: Succeeded Jul 07 20:30:00 cbrake: scroll up, check for another task that started and hasn't completed yet Jul 07 20:30:12 just because the last *message* was from as uccessful task doesn't mean it isn't still working on another one Jul 07 20:30:19 kergoth_: yeah, appears to be: NOTE: package alsa-lib-native-1.0.24.1-r0: task do_fetch: Started Jul 07 20:30:52 woglinde, is it of any use to patch ecj.in so that the wrapper script returns the error code after the 5 tries of javac? Jul 07 20:31:38 gchiii not really Jul 07 20:37:57 Hiya Jul 07 20:38:15 I am trying to cross-compile 2.6.39 with armel-linux-gnueabi-gcc v4.5 and when booting, I get this jibberish: http://pastebin.com/n0qJmzch is that a compilation error? 2.6.36 compiles just fin Jul 07 20:39:00 dFence: which board Jul 07 20:39:17 khem: mini6410 (samsung s3c6410) Jul 07 20:40:00 well its not compile error Jul 07 20:40:14 probably something is wrong in your serial setup may be Jul 07 20:40:59 khem: checked kernel-settings, unless there's been a change in the boot-parameter for console it checks out Jul 07 20:43:11 yay, it works: NOTE: package core-image-minimal-1.0-r0: task do_rootfs: Succeeded Jul 07 20:46:21 with oe-core, is it possible to get a list of licenses in a image? Jul 07 20:46:35 I found build/tmp/deploy/images/licenses Jul 07 20:46:48 but it would be nice to have a simple list Jul 07 20:50:12 mickeyl: pong Jul 07 20:50:30 mickeyl: heya! How is the family? Get some sleep sometimes? :) Jul 07 20:58:05 mickeyl: going to sleep now. ping me on irc or jabber tomorrow. Jul 07 20:58:07 night all Jul 07 21:00:23 woglinde: it appears that javac doesn't like "-bootclasspath '' " but does fine with "-bootclasspath :: " Jul 07 21:01:51 cbrake: could we get the licenses in classic oe Jul 07 21:03:17 khem: pidge@#yocto is updating the functionality to collect license per image Jul 07 21:03:45 khem: right now all licenses are collected in a common directory, but that does not tell us what licenses apply to a particular image Jul 07 21:04:00 yeah that would be useful Jul 07 21:04:13 some sort of bill of material Jul 07 21:04:37 yeah licenses are under / dir Jul 07 21:04:58 deploy/licenses// Jul 07 21:06:24 cbrake: are you using angstrom Jul 07 21:06:49 khem: the deploy/licenses/common-licenses/ lists the licenses -- that is kind of a nice feature Jul 07 21:07:11 khem: I'm just doing oe-core builds right now--trying to figure out the license tools Jul 07 21:07:17 khem: will be moving to Angstrom shortly Jul 07 21:08:43 hi all Jul 07 21:08:54 khem: ping Jul 07 21:08:55 hi woglinde khem pb_ :) Jul 07 21:09:32 khem: I did my homeworks, I'd have a recipe to submit to your attn. Jul 07 21:09:40 ant__: sure ok Jul 07 21:09:47 best by mail Jul 07 21:11:28 khem, did you see my login @ patchwork ? Jul 07 21:12:11 whats your login Jul 07 21:12:17 my login is a.sappia Jul 07 21:15:28 alx_: yes Jul 07 21:15:32 so what do u want Jul 07 21:15:45 did u forget passworf Jul 07 21:15:50 well there was a timeout Jul 07 21:16:01 when i was doing singup Jul 07 21:16:06 signup Jul 07 21:16:29 and the account is inactive Jul 07 21:16:41 i know my password Jul 07 21:20:16 khem: sent Jul 07 21:20:45 alx_: ok retry now Jul 07 21:20:53 ant__: will look Jul 07 21:21:13 khem, thanks Jul 07 21:21:20 thx, now klcc can be installed and removed Jul 07 21:22:02 ant__: ok Jul 07 21:34:31 khem: both initramfs images in the layer do build fine now. I'll try now to add the missing bits to kernel.bbclass Jul 07 22:14:52 I'm trying to update a kernel from 2.6.31 (latest on repo from mini2440) to 2.6.37... I used the same defconfig... but ... on boot the new kernel has no ttySAC0 :( Jul 07 22:15:02 I've finished ideas :( Jul 07 22:15:09 config seems ok :( Jul 07 22:19:44 whats the best way to build a package multiple times with a different configuration? is there an example of this in action? Jul 07 22:21:46 khem: kernel.bbclass + xz/lzma stuff added. I'll send patch to ML Jul 07 22:22:23 (let me see it boots before :p) Jul 07 22:22:29 alx_: seems a kernel issue to me Jul 07 22:22:34 ant__: nice Jul 07 22:22:52 msm: usually thats not so well intended Jul 07 22:23:38 khem: well im trying to build multiple u-boot images Jul 07 22:23:49 msm: for same machine ? Jul 07 22:23:53 so it is quite literally compiling u-boot 2-3 times to generate these image Jul 07 22:24:03 for a specific boot media Jul 07 22:24:22 sdcard, spiflash, nor, nand Jul 07 22:24:30 depending upon what the board has Jul 07 22:24:38 msm: one way I think of is to have separate recipes per media you are building for Jul 07 22:24:52 ah Jul 07 22:25:00 so depens on machine Jul 07 22:28:03 msm: you can change do_compile basically Jul 07 22:28:24 I think such a feature would be acceptable for machine meta repos Jul 07 22:28:33 khem, I'll be away since 5AM today for a week. My problem is the difference between linux-libc-headers and kernel interfaces. I wrote a recipe for same version kernel (2.6.37) for mini just copying the old version to new. The new recipe is in an overlay (no BBLAYERS), with the defconfig in the linux-mini2440_2.6.37+git subdirectory (in the overlay there is the same structure as on openembedded core). I don't know is it is : (1) Jul 07 22:28:33 a wrong recipe or (2) a kernel issue or (3) something else. My work here is to see if the mini2440 has enough power for the program we developed: the check is to plot on the monitor 12 track realtime (1Khz sampling rate). Jul 07 22:29:49 So far I'm fighting with oe-classic and it's broken dependencies (solved), qwt6 for qt4e (who knows if it works)... Jul 07 22:30:12 Well, hang on msm :) Jul 07 22:30:16 alx_: try to isolate Jul 07 22:30:19 There's a multi-uboot.inc around somewhere I think Jul 07 22:30:36 Might only be in one of ulf's branches tho :( Jul 07 22:31:01 alx_: does the old root file system work with new kernel ? Jul 07 22:31:06 And there's similarly multi-kernel.inc Jul 07 22:31:39 khem, didn't checked. Jul 07 22:32:04 khem, let's talk again next week. Jul 07 22:32:23 my flight is coming. still to do my luggage ... Jul 07 22:32:40 see you all soon Jul 07 22:35:15 khem: strangely do_populate_sysroot() { Jul 07 22:35:15 <------>: Jul 07 22:35:15 } works and fails w/out ":" Jul 07 22:35:51 Tartarus: i'll have a look at these recipes Jul 07 23:20:16 alx_ Jul 08 00:08:57 good nite Jul 08 02:21:40 I'm trying to set up the bitbake & oe toolchain and I'm having issues with it ignoring local.conf, thoughts? Jul 08 02:22:56 incorrect BBPATH Jul 08 02:23:02 thats how all conf/ files are located Jul 08 02:24:18 I've tried putting local.conf in next to bitbake.conf and it still doesn't seem to register Jul 08 02:24:24 and it definitely sees bitbake.conf Jul 08 02:24:41 it complains if I change BBPATH to not include the parent of conf/bitbake.conf Jul 08 02:35:03 foobraz: you should *not* put local.conf by bitbake.conf Jul 08 02:35:12 also, bitbake.conf is in two locations, both bitbake and OE Jul 08 02:35:16 you want the oe one, not the bitbake one Jul 08 02:45:29 alright, good to know. Will that affect the local.conf issue though? Jul 08 02:49:04 do a bitbake -e | grep BBPATH= Jul 08 02:49:07 check sanity Jul 08 02:51:33 well huh... that gives me no output at all Jul 08 02:53:12 alright, I switched it to the openembedded bitbake.conf and now it's just kind of barfing errors Jul 08 02:53:19 * foobraz is googling vigorously **** ENDING LOGGING AT Fri Jul 08 02:59:58 2011