**** BEGIN LOGGING AT Fri Jun 26 02:59:57 2009 Jun 26 03:43:43 Hello. I'm trying to convince helloworld-image to work properly, is this the right place? Or should I bugger off to #angstrom? Jun 26 03:44:16 Specifically, I'm not sure how to make it include a kernel anywhere. Jun 26 06:09:39 gm Jun 26 06:48:08 Hi! .... why does DEPENDS="cmake-native" not install cmake prior to do_configure ? Jun 26 06:57:16 hmm .. .does bitbake skip dependencies interactive mode ? Jun 26 06:57:25 hmm .. .does bitbake skip dependencies in interactive mode ? Jun 26 09:12:38 mickey|linuxtag: good morning Jun 26 09:21:42 03Sebastian Krzyszkowiak  07shr/import * rc28471b562 10openembedded.git/recipes/freesmartphone/opimd-utils_svn.bb: opimd-utils: add icons to package Jun 26 09:21:52 Is there support for localisation packages separately in OE ? Jun 26 09:22:20 oespirit: can you be more specific? Jun 26 09:22:50 pb_: I have lots of packages with only po files, which I need to write recipes for. Jun 26 09:28:04 pb_ : The 'compiling' of .po and installation into locale pakcages is taken care of , in the autotool-ed packages, but in this case, the source isnt autotool-ed. I only have debian info and po folder Jun 26 09:28:22 morning pb_ Jun 26 09:31:01 oespirit: so, you just have a bunch of .po files and no build infrastructure to go with them? there's no support for that as such. I guess you could write a new class to handle it though. Jun 26 09:31:24 pb_: will do that. thanks :) Jun 26 09:33:15 morning Jun 26 10:26:42 Anyone here using software opengl? I have trouble usign it. (undefined symbol: _glapi_tls_Context)) **** BEGIN LOGGING AT Fri Jun 26 10:29:06 2009 Jun 26 10:53:05 * * OE Bug 5256 has been created by lukas.gorris(AT)gmx.de Jun 26 10:53:07 * * i2c-tools fetch fails Jun 26 10:53:08 * * http://bugs.openembedded.org/show_bug.cgi?id=5256 Jun 26 12:25:26 how to use overlaying for machine conf files too? Jun 26 12:47:41 just set your BBPATH appropriately. Jun 26 12:48:48 how do i use a custom toolchain for a single custom .bb file when building e.g. a console-image? Jun 26 12:54:54 03Michael Smith  07org.openembedded.dev * rceb3562c1e 10openembedded.git/recipes/tcpdump/ (tcpdump_3.9.7.bb tcpdump_4.0.0.bb): Jun 26 12:54:54 tcpdump: fix mix of tabs and spaces Jun 26 12:54:54 Signed-off-by: Michael Smith Jun 26 12:54:59 03Michael Smith  07org.openembedded.dev * raf34699297 10openembedded.git/recipes/tcpdump/tcpdump_4.0.0.bb: Jun 26 12:54:59 tcpdump 4.0.0-r2: remove extra ${sbindir}/tcpdump.4.0.0 Jun 26 12:54:59 Signed-off-by: Michael Smith Jun 26 12:56:09 03Michael Smith  07org.openembedded.dev * r9bc5c3798e 10openembedded.git/recipes/iputils/iputils_20071127.bb: Jun 26 12:56:09 iputils 20071127: add docbook dependency to fix build Jun 26 12:56:09 Looks like the newer snapshots don't include manpages, so Jun 26 12:56:09 docbook-utils-native is required to build them. Jun 26 12:56:09 Signed-off-by: Michael Smith Jun 26 13:19:36 pb_: for .conf? :) Jun 26 13:21:17 booxter: yes, that's what BBPATH is for. Jun 26 13:21:36 pb_: can you show me what is the right way to configure it? Jun 26 13:22:07 pb_: mine is BBPATH=/home/booxter/projects/epam/epm-tbox:/home/booxter/projects/epam/epm-tbox/build:/home/booxter/projects/epam/epm-tbox/openembedded:/home/booxter/projects/epam/epm-tbox/overlay Jun 26 13:23:11 booxter: those directories don't mean much to me, but the syntax is correct. BBPATH is simply searched from left to right, and bitbake will use the first file it finds. Jun 26 13:23:30 it works just like $PATH in the shell Jun 26 13:24:37 hello all Jun 26 13:24:51 please how can i fix this error on arm architecture ? Jun 26 13:24:51 Alignment trap: rtnetd (26878) PC=0x00008a6c Instr=0xe18400b6 Address=0x00012011 FSR 0x813 Jun 26 13:25:14 i have #pragma pack(1) Jun 26 13:28:44 pb_: it works for .bb files not .conf Jun 26 13:28:49 overlays too Jun 26 13:28:54 or am I wrong? Jun 26 13:35:05 * * OE Bug 5261 has been created by chrisfriedt(AT)gmail.com Jun 26 13:35:07 * * guile-native_1.8.2 fails to build Jun 26 13:35:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5261 Jun 26 13:45:50 morning Jun 26 14:08:22 I want to customize recipes/psplash/files/psplash-default, where should I have to add my custom modification? Jun 26 14:09:49 basically, with a specific machine I need PARAMS="--angle 90" Jun 26 15:06:23 for example machine topas910 has its own setting PARAMS=" -a 270" in recipes/psplash/files/psplash-default Jun 26 15:07:00 but I don't understand how bitbake dig into this directory Jun 26 15:07:50 if I create recipes/psplash/files/mymachine would be enough? Jun 26 15:12:26 mckoan: see ${FILESPATHPKG} Jun 26 15:13:00 or maybe ${FILESPATH} actually Jun 26 15:13:15 pb_: thx Jun 26 15:17:47 03Tom Rini  07org.openembedded.dev * r83f41716ab 10openembedded.git/recipes/eglibc/ (4 files): Jun 26 15:17:47 eglibc: Make sure that scsi headers are available when needed Jun 26 15:17:47 This is fixing bug #4262 for eglibc (original did glibc and linux-libc-headers). Jun 26 15:17:47 See b8bb2a62b3916470c752bd79d31322b4358fc676 for the rest of this. Jun 26 15:26:42 03Michael Smith  07org.openembedded.dev * r84efca2e2a 10openembedded.git/ (4 files in 3 dirs): Jun 26 15:26:42 Add fping 2.4b2_to, plus patch from SourceForge. Jun 26 15:26:42 Signed-off-by: Michael Smith Jun 26 15:36:26 how to make angstrom to autologin ? Jun 26 15:36:40 as user root Jun 26 15:37:26 lol Jun 26 15:37:58 "motion" depemds on mysql AND pgsql Jun 26 15:38:08 now thats lightweight :> Jun 26 15:38:10 sx: best ask #angstrom about that Jun 26 15:38:47 anyone else noticed that startx doesn't seem to work for xfbdev at least? it wants /usr/bin/X, which doesn't exist, but /etc/X11/Xserver does. hmm Jun 26 15:39:01 * kergoth should really fire up an old zaurus and throw a OpenEmbedded built distro on it Jun 26 15:39:06 pb_, ok, asked there Jun 26 15:39:21 kergoth: I guess not many people are using startx. I think most folks have X started automatically at boot time. Jun 26 15:39:32 * kergoth nods, gotcha Jun 26 15:40:03 obviously that's no excuse for it being broken, but it probably is true that nobody else has noticed :-} Jun 26 15:40:16 if startx exists, and an x startup fails, i can see someone trying to use it to manually start it, to see what's failing, however. easy enough fix I'm sure, though Jun 26 15:40:18 * kergoth nods Jun 26 15:40:28 * kergoth adds to the list Jun 26 15:40:49 the whole x startup script situation is a bit of a nightmare at the moment, hard coded paths all over the place Jun 26 15:41:25 I tried building a gpe-image for micro the other day but it loses pretty badly because a whole lot of scripts want to look for things in /usr which doesn't exist. Jun 26 15:41:34 ick Jun 26 15:42:08 and, in most cases, the paths seem to be hard-wired for no good reason. for example, loads of them do "exec /usr/bin/Xfbdev" or whatever, rather than letting the shell find Xfbdev on $PATH. Jun 26 15:42:21 weird Jun 26 15:42:34 exec doesn't need a full path, that seems odd Jun 26 15:42:36 okay, so XSERVER + virtual/xserver is all a machine has to do to get the right bits built and used if they use task-x11, right? Jun 26 15:42:41 * kergoth is out of the loop on this stuff Jun 26 15:42:52 I think so, yeah Jun 26 15:43:07 not that I'm exactly in the loop either, but perhaps less out of it than you Jun 26 15:43:08 heh Jun 26 15:43:12 hehe Jun 26 15:43:49 i should actually use this sheevaplug at some point, not that itd help me test this, lacking a display, but still, having a device to run an OpenEmbedded built distro so i can freshen up on the structure would probably be good. vmware doesn't like me very much Jun 26 15:43:54 hrmph Jun 26 15:45:02 I use the 'plug -- no far only a few problems Jun 26 15:45:20 one of them is that the .29 kernel in OE tree is crap. you need the .30 Jun 26 15:45:29 kergoth: you can use qemu for testing, I've had fairly good luck with that. Jun 26 15:49:10 hmm, that's a good idea Jun 26 15:55:24 the i386 system qemu can run xfbdev (and maybe xvesa). arm system qemu can run xfbdev too, but I've never had the patience to fiddle with all the kernel bits to make that work properly. Jun 26 15:58:07 does anyone know how to make angstrom to autologin in console as root? Jun 26 15:59:25 I couldn't get an answer in #angstrom channel Jun 26 16:09:12 * kergoth fires off a micro/qemux86 micro-image Jun 26 16:15:04 pb_: pinch the kernel config from poky? ;-) Jun 26 16:34:38 RP: ever seen: line 967, in execute_runqueue_internal - KeyError: 5898 (or whatever)? Jun 26 16:34:58 kergoth: hmm, yes Jun 26 16:35:00 I've seen it a few times now over the past couple months, no rhyme or reason to it Jun 26 16:35:04 quite odd Jun 26 16:35:11 * kergoth hasn't had time to really investigate though Jun 26 16:35:16 kergoth: I'm trying to remember what that means :/ Jun 26 16:36:16 Basically it means there is a bug somewhere in runqueue or taskdata and something was removed from the taskgraph when it shouldn't have been Jun 26 16:36:26 or a reference is corrupted somehow Jun 26 16:36:57 I know there were a load of zero reference bugs in there that looked a bit like it but I think I caught all of them Jun 26 16:37:33 (if not foo: checks which removed the first element of arrays when they shouldn't have done) Jun 26 16:43:08 interesting, I'm getting this every time i run here. /me tries wiping the cache, just for the hell of it Jun 26 16:46:30 kergoth: Its some compnation of the things you happen to be doing. It can be influenced by the order of the .bb files on disk for example Jun 26 16:46:49 Although I think I sorted some of the inputs to avoid that and make things more reproducible Jun 26 16:47:26 hrmph Jun 26 16:48:10 * kergoth cranks up the debugging Jun 26 16:49:16 kergoth: I was a bit scared when the order of files on the disk was influencing a crash... Jun 26 16:49:26 yes, that's scary indeed Jun 26 16:49:31 debugging that would be a royal pain in the ass Jun 26 16:50:04 i should cp -a this project to someplace else and see if it still happens, and if so, keep it there for safekeeping, as a reproducer Jun 26 16:53:50 kergoth: yes, those bugs are usually hard to track without a testcase Jun 26 16:56:34 oh, i forgot, i already had an OpenEmbedded tree that reproduced it, a quick 'locate breaks' found metadata.breaksbitbake. /me sees if it still does so.. Jun 26 17:08:49 wow, this circular dep is taking it a loooong time to identify :) Jun 26 17:14:04 who is in charge of the openmoko distro config in org.OE.dev? Since they have their own repo, is there actually anyone in charge of it? Jun 26 17:14:41 I'd like to drop guile-native-1.8.2 -> bug 5261, but openmoko has it set as preferred version Jun 26 17:14:45 !oebug 5261 Jun 26 17:14:48 * * Bug 5261, Status: CONFIRMED, Created: 2009-06-26 13:33 Jun 26 17:14:49 * * chrisfriedt(AT)gmail.com: guile-native_1.8.2 fails to build Jun 26 17:14:50 * * http://bugs.openembedded.org/show_bug.cgi?id=5261 Jun 26 17:15:32 Laibsch: I think the openmoko bits are merged into upstream oe periodically, but that might be on hold due to mickey's ongoing boycott. Jun 26 17:15:40 try asking Ainulindale, he might know the situation Jun 26 17:16:04 I wonder which way the merge is happening? OM->OE or other way round? Jun 26 17:16:10 ping Ainulindale Jun 26 17:17:39 pb_: how do you like the bug tracker these days? Any better? Jun 26 17:18:04 * Laibsch really should do a poll on "what keeps you from being more active in the tracker" Jun 26 17:18:18 If there are structural problems, I'd like to fix them Jun 26 17:19:00 * kergoth has a tendency to forget it exists, for some reason Jun 26 17:23:27 Hehe Jun 26 17:23:40 I kind of had the feeling kergoth was a forgetful type ;-) Jun 26 17:23:46 Yet we love him Jun 26 17:39:42 kergoth, pb_: I'm increasingly frustrated with builds breaking four or five times everytime because SRC_URI is no longer valid. Keeping that uptodate is just silly work. Jun 26 17:39:42 I wonder if something like http://www.sidebranch.com/witpa/ could be easily integrated into do_fetch as a fallback? Jun 26 17:39:49 http://article.gmane.org/gmane.comp.handhelds.openembedded/17776 has more info on witpa Jun 26 17:41:10 that seems promising, and would be easy to make use of Jun 26 17:41:35 * kergoth thinks, wonder what the best way of integrating it would be Jun 26 17:41:56 probably tweak do_fetch to call it out before it calls the bb.fetch code Jun 26 17:42:13 although.. Jun 26 17:42:26 Laibsch: yeah, the bug tracker does seem better nowadays. main obstacle to me using it over the last few weeks has just been time shortage. Jun 26 17:43:00 kergoth: I think you'd want to go on using SRC_URI as the primary uri, and only use the witpa thing if the original uri was dead. Jun 26 17:43:14 pb_: cool. Leaves me wondering why it still seems to be used so little Jun 26 17:43:15 aye, that was the "although.." Jun 26 17:43:15 hehe Jun 26 17:43:49 Yes, my idea was also to use the orig SRC_URI and fall back to witpa in case of failure Jun 26 17:43:53 itd be easy to call out witpa in a ${@} block to add items from it to MIRRORS, I'm thinking. then add witpa-native to the do_fetch depends Jun 26 17:44:16 In fact, I've been working a little on a witpa-native recipe these Jun 26 17:44:17 day Jun 26 17:44:18 s Jun 26 17:44:26 would have to ensure that the ${@} expansion wouldn't fail if witpa isn't yet around, of course, just in case MIRRORS gets expanded before do_fetch is run Jun 26 17:44:34 right Jun 26 17:44:36 Unfortunately, I could not get it completed Jun 26 17:45:04 ${@"^http://.*/*\t%s" % url for url in os.popen....} Jun 26 17:45:05 or something Jun 26 17:45:12 * kergoth ♥ generator expressions Jun 26 17:46:12 RP: okay, have a safe copy of this build area + metadata that causes one of those keyerrors. i'll just leave it there for now and investigate at some point in the future, unless you want it Jun 26 18:23:36 rp: what should we do about poky bugs that are fixed in .dev? Jun 26 18:23:43 !oebug 4907 Jun 26 18:23:45 * * Bug 4907, Status: UNCONFIRMED, Created: 2008-12-05 19:44 Jun 26 18:23:45 * * kgilmer(AT)gmail.com: classpath-initial-0.93 incorrect $S Jun 26 18:23:47 * * http://bugs.openembedded.org/show_bug.cgi?id=4907 Jun 26 18:23:49 !oebug 4908 Jun 26 18:23:50 * * Bug 4908, Status: UNCONFIRMED, Created: 2008-12-05 20:16 Jun 26 18:23:51 * * kgilmer(AT)gmail.com: classpath-native 0.97.2 invalid $S Jun 26 18:23:53 * * http://bugs.openembedded.org/show_bug.cgi?id=4908 Jun 26 18:24:04 They were fixed with the introduction of BPN Jun 26 18:24:35 My vote goes to close them as invalid, but I wanted to give you the chance to fix it in poky Jun 26 18:26:15 closing invalid, not a bug in .dev Jun 26 18:27:05 * * OE Bug 4907 has been RESOLVED (INVALID) by Jun 26 18:27:07 * * classpath-initial-0.93 incorrect $S Jun 26 18:27:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4907 Jun 26 18:29:05 * * OE Bug 4908 has been RESOLVED (INVALID) by Jun 26 18:29:07 * * classpath-native 0.97.2 invalid $S Jun 26 18:29:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4908 Jun 26 18:34:39 pb_: http://paste.debian.net/40353/ is my witpa-native recipe Jun 26 18:35:17 It currently fails in do_populate_staging: http://tinderbox.openembedded.net/packages/97166/ Jun 26 18:35:41 To be honest, I have no idea why. Can you give me a hint? Jun 26 18:35:56 | make: *** No rule to make target `install'. Stop. Jun 26 18:36:02 Can we use some kind of default? Jun 26 18:36:18 * Laibsch thinks of autotools_stage_all Jun 26 18:36:23 Let's try that Jun 26 18:40:25 hi.. Jun 26 18:40:50 somebody can give some documentation about how to create a build enviromnent with oe? Jun 26 18:41:05 * * OE Bug 5256 has been RESOLVED (INVALID) by lukas.gorris(AT)gmx.de Jun 26 18:41:07 * * i2c-tools fetch fails Jun 26 18:41:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5256 Jun 26 18:41:14 im trying to compile a "ocaml compiler" Jun 26 18:41:20 but this need ncurses.. Jun 26 18:41:36 arm-angstrom-linux-gnueabi/bin/ld: cannot find -lcurses Jun 26 18:42:26 tegila: do you have libcurses in staging Jun 26 18:43:10 i have beagleboard-demo-image compiled, staged and packaged.. Jun 26 18:43:19 working on my beagleboard Jun 26 18:44:13 khem: how can i told to compiler this? Jun 26 18:44:19 i cant find libncurses.a Jun 26 18:44:48 tegila: its looking for libcurses see missing 'n' Jun 26 18:45:23 khem: thank for the staging idea.. Jun 26 18:45:30 i found libncurses at: staging/armv7a-angstrom-linux-gnueabi/usr/lib/ Jun 26 18:45:43 tegila: yes thats right place Jun 26 18:55:47 hmm, i should make the mrproper task actually useful. make it remove just the downloaded sources of the recipe it's running against, not the whole schebang Jun 26 18:56:15 * kergoth adds to his list Jun 26 19:32:39 Laibsch: Yes ? Jun 26 19:33:33 I was wondering who to talk to about openmoko related things in .dev Jun 26 19:33:53 !oebug 5261 Jun 26 19:33:54 * * Bug 5261, Status: CONFIRMED, Created: 2009-06-26 13:33 Jun 26 19:33:56 * * chrisfriedt(AT)gmail.com: guile-native_1.8.2 fails to build Jun 26 19:33:56 * * http://bugs.openembedded.org/show_bug.cgi?id=5261 Jun 26 19:34:12 guile-native 1.8.6 seems to build fine for everyone reporting to tinderbox Jun 26 19:34:25 I'd like to drop 1.8.2 instead of spending effort to fix it Jun 26 19:34:34 OM uses it as preferred version in .dev Jun 26 19:34:52 Can you tell me more if there is any actual work being done for OM in .dev? Jun 26 19:35:09 Is there any synchronization going on? Jun 26 19:35:14 Laibsch: nothing in dev Jun 26 19:35:21 Laibsch: all was done in FSO MS5.5 Jun 26 19:35:30 Soon to be included/merged Jun 26 19:38:50 can we drop the OM distro config bits, then? Is it OK to drop guile-native 1.8.2? Jun 26 19:40:56 ~seen mario-goulart Jun 26 19:40:57 mario-goulart was last seen on IRC in channel #oe, 2d 23h 1m 55s ago, saying: ':-)'. Jun 26 19:42:27 mmm...where is mario-goulart ? is he still working on the mips rootfs? Jun 26 19:46:23 03Stanislav Brabec  07org.openembedded.dev * r2fe85b7a84 10openembedded.git/ (63 files in 30 dirs): Jun 26 19:46:23 many files: Introduction of virtual/libusb0 provided by libusb or libusb-compat: Jun 26 19:46:23 * for more see http://lists.linuxtogo.org/pipermail/openembedded-devel/2009-June/012217.html Jun 26 19:56:22 pb__: you've been messing with micro images right, do we have anything to do the first boot postint executions for a ONLINE_PACKAGE_MANAGEMENT="none" filesystem? Jun 26 19:56:35 pb__: currently it seems to rely on the configure script the runtime opkg installs, and uses opkg to do it Jun 26 20:00:20 kergoth: yeah, that needs a bit of attention. none of the packages I care about seem to need that, so I haven't been too worried, but in the general case it is an issue. Jun 26 20:01:08 kergoth: easiest thing would probably be to use qemu somehow, or of course re-engineer the packages to handle the postinst some other way. Jun 26 20:01:14 okay, that's what i figured, will have to give that some thought Jun 26 20:01:22 the whole way postinsts are done at the moment is a bit of a mess, it really needs overhauling Jun 26 20:01:34 i think with none the postinsts dont even go into an image, since theyr'e in with the opkg metadata Jun 26 20:01:44 this paradigm of checking for $D and doing offline things is, well, the worst. Jun 26 20:02:10 aye. well, lacking qemu bits at the time, it was the only way to do any of it at all at build time Jun 26 20:02:17 but nowadays we should leverage what we have Jun 26 20:02:19 yeah, agreed Jun 26 20:03:01 also, for packages that just have boilerplate-type postinsts, update-rc.d and that kind of thing, we could just arrange for the classes in question to make the right things happen transparently, no need to actually run the target postinst at all. Jun 26 20:03:23 i.e. teach update-rc.d to emit a special pseudo-postinst to be run on the host, using host tools, during image construction. Jun 26 20:03:36 that's a good point, that would be useful Jun 26 20:04:49 that'd also make it slightly easier to do post-postinst processing for systems without online package management. in the micro case, for example, once all the packages have been installed and update-alternatives has made its choices, we know that those will never change and hence there is an opportunity to remove some of the redundant symlinks and/or binaries Jun 26 20:05:35 that'd be cool, go through and remove any update-alternatives type indirections, collapse em all Jun 26 20:05:46 right Jun 26 20:06:09 the only tricky case is things like busybox where you have multi-function binaries. Jun 26 20:06:27 obviously you don't want to rename busybox to /sbin/insmod, say, just because it happens to be the provider of insmod. Jun 26 20:06:51 well, if we automated the naming of upd-a binaries to foo.${PN}, it could compare the basenames, if they dont fit that, it assumes its a different $0 for a reason Jun 26 20:29:58 yeah, that'd work Jun 26 20:48:02 Ainulindale: can we drop the OM distro config bits, then? Is it OK to drop guile-native 1.8.2? Jun 26 20:59:09 otavio, did you have news on mario-goulart and his work? Jun 26 20:59:15 otavio, oops hi Jun 26 22:03:04 Laibsch: he...wrt zaurusd_svn...well, today is broken..patch won't apply... Jun 26 22:03:42 I'll have a look at it in 30mins Jun 26 22:03:44 brb Jun 26 22:04:01 ant__: you are loosing me 90% of the time when initiating contact with me Jun 26 22:04:09 don't know what you are talking about Jun 26 22:04:32 except that I was wondering if zaurusd should not be included from collie machine config Jun 26 22:24:11 03Stanislav Brabec  07org.openembedded.dev * rf933c2275a 10openembedded.git/ (20 files in 9 dirs): libusb: Update to latest versions of libusb1 and libusb-compat, removed old versions. Jun 26 22:25:22 Laibsch: have you recently built zaurusd ? Jun 26 22:27:04 no Jun 26 22:27:10 did I say I did? Jun 26 22:27:39 tinderbox logs start from 2009-06-18... Jun 26 22:27:43 just asking Jun 26 22:27:48 sorry to bother you Jun 26 22:27:51 was last time Jun 26 22:27:52 end Jun 26 22:45:51 re tinderbox, see announcement to the list Jun 26 22:45:53 http://oe-tinderbox-old.leggewie.org/ Jun 26 22:46:45 zaurusd seems to build fine for lumag **** ENDING LOGGING AT Sat Jun 27 02:59:56 2009