**** BEGIN LOGGING AT Mon Nov 22 02:59:58 2010 Nov 22 04:42:03 while creating the oe ecosystem on your computer why is it necessary to put your files in a directory which is created with mkdir -p Nov 22 04:42:46 mkdir -p isn't necessary, it's just faster than creating a chain of subdirectories individually Nov 22 04:43:18 ksinkar, mkdir -p foo/bar will create foo as well as foo/bar. whereas mkdir foo/bar will fail if foo doesn't yet exist Nov 22 05:15:31 k Nov 22 06:04:08 Hi Nov 22 06:04:56 i want submit kernel , u-boot and x-load patches to open embedded for my new board Nov 22 06:05:11 to which mailing list i have to send these patches Nov 22 06:05:14 nani_: ok, follow the patch submission guidelines on wiki Nov 22 06:05:27 there is a whole document with all the info you need Nov 22 06:05:42 Please guide me properly Nov 22 06:06:31 follow this http://wiki.openembedded.org/index.php/How_to_submit_a_patch_to_OpenEmbedded Nov 22 06:07:15 grg: howdy Nov 22 06:07:23 khem, hi Nov 22 06:07:39 grg: did u had a chance to try release brach Nov 22 06:07:56 khem, i haven't done a world build yet Nov 22 06:08:02 grg: can you also try minimal-uclibc DISTRO with mipsel targets ? Nov 22 06:08:08 but did x11-image and mico-image Nov 22 06:08:15 grg: dont worry about world now a days Nov 22 06:08:17 uclibc fails somewhere... Nov 22 06:08:28 ok micro-uclibc would be fine too Nov 22 06:08:39 or rather, busybox fails in some header somewhere (i forget the details) Nov 22 06:08:55 grg: hmmm, if you find the error lemme know Nov 22 06:09:01 we shoukd fix it Nov 22 06:09:35 it had to do with off64_t not being found in a busybox file. when i traced it through to the headers, everything looked fine Nov 22 06:09:53 i'll do the build again and send the error to the mailing list Nov 22 06:10:05 grg: thats because its expecting largefile feature from DISTRO_FEATURES I think Nov 22 06:10:21 grg: see minimal-uclibc for DISTRO_FEATURES Nov 22 06:10:25 ok Nov 22 06:11:14 khem, also, i hit a race building gnutls-native. I'm looping on it, trying to reproduce. but haven't reproduced in >1hr now Nov 22 06:11:31 hmmm Nov 22 06:11:58 there's a problem with dependencies in one of the Makefiles in the guile subdir Nov 22 06:12:35 ok, does disabling PARALLEL build helps ? Nov 22 06:12:39 failure looks like this: http://tinderbox.openembedded.net/public/logs/task/7589267.txt Nov 22 06:12:48 khem, yeah, it would (haven't tested) Nov 22 06:13:03 (btw, that's not my failure, that's someone else's that google found) Nov 22 06:13:24 it is definitely a makefile bug Nov 22 06:13:41 i just have no auto-fu, so can't fix it Nov 22 06:14:21 khem, how do you handle a asdf.c dependeny on qwer.h in a Makefile.am ? Nov 22 06:15:08 asdf.c: qwer.h Nov 22 06:15:22 hmm that simple, eh? Nov 22 06:17:08 yes but depends how the rules are set Nov 22 06:17:35 sometimes there are group rules Nov 22 06:18:56 there it was given like send patches to openembedded-devel@lists.openembedded.org but can i send kernel patch to this list. Nov 22 06:19:44 yes as long as they pertain to your kernel recipe Nov 22 06:21:54 grg: well I told you wrong it should be asdf.o: asdf.c qwert.h Nov 22 06:23:00 khem, well i'm not sure of all the possible failure modes. I'll send something to the gnutls list and see what they come up with Nov 22 06:23:32 it looks to be a general problem that would have a general solution Nov 22 06:23:56 my failure which occurs on a different file, http://tinderbox.openembedded.net/public/logs/task/10751151.txt Nov 22 06:24:00 means i have modified the recipe to support the patch. i can send only the recipe patch file or i can send the patch file of kernel also Nov 22 06:26:13 nani_: look at the mailing list archive for some sample of patch submission Nov 22 06:26:22 if you are not getting it Nov 22 06:28:06 grg: unterminated ifdef hmmm is enum.h a generated file ? Nov 22 06:28:15 khem, yes Nov 22 06:28:27 grg: ok hmmm Nov 22 06:28:32 but there are multiple generated .h files and multiple files that include them Nov 22 06:28:48 grg: then look for rules for errors.c Nov 22 06:28:56 and add enums.h as I suggested Nov 22 06:29:13 you may find that a group of files need this Nov 22 06:29:15 i think i have a fix, but as i cant seem to reproduce the failure i'd rather get confirmation of the fix from a gnutls dev on their mailing list Nov 22 06:29:27 usually they will be grouped together already Nov 22 06:29:35 yes Nov 22 06:29:40 03Martin Jansa  07master * rf2bb9f98e3 10openembedded.git/recipes/openmoko-3rdparty/mokojeweled_git.bb: Nov 22 06:29:40 mokojeweled: bump PR to rebuild after last EFL upgrade Nov 22 06:29:40 Signed-off-by: Martin Jansa Nov 22 06:29:45 yes sending to gnutls is a good idea Nov 22 06:37:30 /home Nov 22 06:57:05 gm Nov 22 07:42:01 How to reduce disk space required for OpenEmbedded build process? There is no effect from INHERIT += "rm_work" in the config. Does something changed in "org.openembedded.dev" branch with this option? Nov 22 07:42:35 I use bitbake-1.8.18. Nov 22 08:03:29 plyatov: INHERIT += "rm_work" should lead to reduced space as most of the work dirs is removed (actually all but the temp dir) Nov 22 08:05:07 there is no effect from this option at all. Nothing was removed in a process of building. I have ~22GB TMPDIR now! Nov 22 08:06:17 ka6sox, the idea was to send your vote only to graeme Nov 22 08:06:18 does this tested by somebody? Nov 22 08:06:41 plyatov: I use it often, but it only applies to things that you build afterwards,it will not prune your existing work dir Nov 22 08:06:56 I understand Nov 22 08:07:54 enable the option in your local.conf; do a bitbake -cclean helloworld; bitbake helloworld, then do a find tmp/work/*/helloworld* Nov 22 08:08:02 hmmm....email still baffles me sometimes. Nov 22 08:09:03 :-) Nov 22 08:15:04 good morning Nov 22 08:15:49 hi mckoan Nov 22 08:16:29 hi Nov 22 08:39:36 eFfeM_work: The INHERIT += "rm_work" option has effect only if I build a concrete recipe, but it useless if I build image for my machine. Nov 22 08:40:31 mornin Nov 22 08:49:22 morning Nov 22 08:50:11 i have a little problem in a build: * check_data_file_clashes: Package udev wants to install file /home/pc/oe/build/tmp-angstrom_2008_1/rootfs/x11-image/etc/udev/rules.d/50-hostap_cs.rules Nov 22 08:50:15 But that file is already provided by package * udev-hostap-cs-vcc-workaround Nov 22 08:50:18 * opkg_install_cmd: Cannot install package task-base-extended. Nov 22 08:52:41 * Nov 22 08:52:41  Nov 22 08:59:46 i have this error 3 times Nov 22 08:59:58 are there some tricks to prevent this? Nov 22 09:03:22 cazze: which machine? I built akita-angstrom+x11-image with no errors yesterday Nov 22 09:04:08 h2200 Nov 22 09:05:59 good morning Nov 22 09:07:15 mckoan: i think i miss something in my machine file Nov 22 09:07:59 eFfeM_work: beside the lzma issue, build broke again for bluetz-dtl1-workaround http://pastebin.ca/1997864 Nov 22 09:08:40 mckoan: grrr, no, i have preferred machine udev-compat141 in it Nov 22 09:10:22 cazze: task-base-extended failed too on my build Nov 22 09:10:37 with the same error? Nov 22 09:10:58 no, ^^^ Nov 22 09:11:36 in my case after building bluez-dtl1-workaround the issue disappear Nov 22 09:12:32 then there is thestrange conflict wrt device_table Nov 22 09:12:52 ant_work: i have the device_table error also :-) Nov 22 09:15:10 could be related to http://bugs.openembedded.net/show_bug.cgi?id=3593 ? Nov 22 09:15:34 ant_work: wrt the dtl1-workaround, perhaps ask a bluez or task wiz, this is for pcmcia, I don't think I ever build a pcmcia error Nov 22 09:15:57 eFfeM_work: what rings a bell, last week the issue was not present Nov 22 09:16:17 but then again it is an RRECOMMENDS so if dtl1 workaround is not there it should not complain, it is a recommends, not a depends Nov 22 09:16:21 ant_work: git bisect ? Nov 22 09:16:22 and I can't see obviously related commits :/ Nov 22 09:19:51 me neither Nov 22 09:23:14 mckoan: but the ankita does have the udev-hostap-cs-vcc-workaround, that's why you build without problems :-) Nov 22 09:24:46 ant_work: not sure bu maybe bitbake -v -v -v -D -D -D can shed any light too Nov 22 09:24:49 cazze: I build for zaurus c7x0 which differs from akita only very marginally Nov 22 09:25:20 mckoan: doens notNOT have Nov 22 09:25:35 eFfeM_work: I'm a bit upset by those race issue plaguing my builds Nov 22 09:25:37 sorry for the garbage, i have a bad connection Nov 22 09:25:54 first I thought it was Gentoo but the races are repeated on Ubuntu Nov 22 09:25:55 mckoan: does not have udev-hostap-cs-vcc Nov 22 09:26:13 ant_work: I can imagine, they should not happen Nov 22 09:29:23 eFfeM_work: though, I pulled from bitbake master Nov 22 09:29:56 but rewinding back seems no cure Nov 22 09:30:18 i'm not sure that this is a bitbake issue, could also be in the oe classes Nov 22 09:30:35 right Nov 22 09:31:01 (for the listeners, we ar talking about paralel builds with multiple bbthreads) Nov 22 09:31:24 * ant_work kicks the Logi°°°° keyboards Nov 22 09:36:57 cazze: no udev-hostap-cs-vcc Nov 22 09:37:16 mckoan: that's what i tried to type :-) Nov 22 09:37:24 LOL Nov 22 09:37:42 mckoan: thats the point that failes for my x11-image Nov 22 09:40:44 good morning Nov 22 09:40:49 morning Nov 22 10:17:22 Hello all! Nov 22 10:17:41 Does somebody knows, why INHERIT += "rm_work" option has effect only if I build a concrete recipe, but no effect if I build an image for my machine? Nov 22 10:19:16 plyatov: where did you declare it? Nov 22 10:19:24 rm_work is after build? Nov 22 10:19:41 mckoan: In a local.conf Nov 22 10:19:42 plyatov: bitbake image -crm_workall? Nov 22 10:19:58 guessing Nov 22 10:19:58 Hi Nov 22 10:20:26 I try to build OE on OSX. Nov 22 10:20:40 hrw: I just type "bitbake gms-image" Nov 22 10:21:52 Afer,I use command "bitbake console-image" Nov 22 10:21:56 hrw: and when many packages builded as dependencies for "gms-image", they does not cleaned. Nov 22 10:22:24 it say "Please install following missing utilities: chrpath" Nov 22 10:22:40 How to resolve this problem.? Nov 22 10:23:02 install it? Nov 22 10:23:11 HS8JCV: just install this chrpath and that's it. Nov 22 10:23:27 How to install.? I use OSX Nov 22 10:23:43 I can't find package or sourcecode. Nov 22 10:23:43 hi Nov 22 10:24:03 HS8JCV: better build under linux (e.g. in a VM) Nov 22 10:24:06 ericben: hi Nov 22 10:24:12 anyone faced this : FATAL: Task /home/ebenard/oetest/openembedded/recipes/images/minimal-image.bb (do_rm_work) has circular dependency on /home/ebenard/oetest/openembedded/recipes/images/minimal-image.bb (do_rm_work_all) Nov 22 10:24:16 ? Nov 22 10:24:32 hm. no. very strange Nov 22 10:25:27 eFfeM_work : Use linux better.Right? Nov 22 10:25:28 HS8JCV: check out macports or homebrew. perhaps they have chrpath covered. Nov 22 10:25:44 HS8JCV: but I agree with eFfeM_work :) Nov 22 10:26:30 OK,I will resize partition for linux. Nov 22 10:26:42 Which linux distro? Nov 22 10:26:43 HS8JCV: most of the people use linux as host, using things like osX or mingw will cause you lots of grief and there is not much knowledge/experience with it, so hardly any advise/support available Nov 22 10:27:13 HS8JCV: I use ubuntu 10.04 @ work (opensuse @ home), guess ubuntu is most widely used Nov 22 10:29:36 thank you for advantage.My MacPro use i7 620m. Do you know my macpro must be install Ubuntu 64bit or 32bit.? Nov 22 10:31:40 eFfeM_work : My CPU is i7 core. I must be install ubuntu 64bit or 32 bit.? Nov 22 10:32:17 HS8JCV: you can use either, but I would probably go for 64 bit Nov 22 10:34:02 eFfeM_work : thank you again. Nov 22 10:34:12 HS8JCV: yw Nov 22 10:42:46 little question, if i bitbake base-image and later i want to extend that to console-image without rebuilding everything, how should i proceed? how can i rebuild the image for my device? Nov 22 10:43:59 just bitbake console-image :) Nov 22 10:44:35 ok, and will it take the packages with it between the base-image and console-image bitbake? Nov 22 10:44:52 yes, if you didn't remove tmpdir only the mising packages will be built Nov 22 10:45:02 ant_work: great, thx Nov 22 10:45:55 ant_work: i'll do a step by step image then :-) Nov 22 10:46:32 ant_work: you are on gentoo also i think? Nov 22 10:46:35 you'll end up with two images in your deploy dir Nov 22 10:46:46 yes, I do Nov 22 10:47:02 well, I *was* building on Gentoo until saturday... Nov 22 10:47:07 ant_work: i have set my source dir to /usr/portage/distfiles Nov 22 10:47:39 but sometimes i have to change the group permissions again on the /usr/portage/distfile dir again Nov 22 10:47:47 I see, never run eclean distfiles then Nov 22 10:47:57 it stops giving the permission to my user who is in the portage group Nov 22 10:48:45 is that a know problem? Nov 22 10:49:03 eFfeM_work: strange, I cleanend the build & tmp dir and now I don't have anymore the circular dep error ... Nov 22 10:49:10 hmm Nov 22 10:49:12 yes, read about 'userpriv' Nov 22 10:50:41 cazze: sorry, 'userfetch' Nov 22 10:51:31 hm.. no, it's not what youìre looking for Nov 22 10:52:21 morning Nov 22 10:52:49 hey XorA Nov 22 10:52:56 ant_work: where do i have to read about userfetch? :-) Nov 22 10:53:00 whats cooking this morning? Nov 22 10:53:07 read Gentoo FEATURES Nov 22 10:53:26 XorA: we have some easter-eggs boiling ;) Nov 22 10:53:33 ant_work: ok, thx Nov 22 10:54:38 XorA: fwiw it seems you never experience race issues on first build...magic Fedora? Nov 22 10:55:10 ant_work: I only have BB=2 -j 2 Nov 22 10:55:52 I see, BB=4 -j5 on XeonQuad Nov 22 10:56:04 still uptime reveals sub-optimal load Nov 22 10:56:55 I have kernel compiles going in other windows, so OE is just background task Nov 22 10:57:27 there are pases during glibc build where load is extremely low Nov 22 10:57:45 probably during locale generation Nov 22 10:57:53 *phases Nov 22 10:59:17 or when the packages are disabling parallel make Nov 22 11:02:12 locale is still single threaded Nov 22 11:02:50 and glibc packaging takes an insane amount of time and is AFAIK single threaded Nov 22 11:04:49 morning Nov 22 11:07:23 How to configure my machine to use "opkg" instead of "ipkg" (for org.openembedded.branch)? Nov 22 11:08:05 because ipkg-utils-native does not fetchable anymore... Nov 22 11:09:11 plyatov: rm_work is doing his work but late enough.. Nov 22 11:09:19 https://spreadsheets.google.com/a/jay-tech.ru/ccc?key=0AhY8lmfkCabTdGZRSGZ0eXg5cWhkV2k1c0Jvd0lvVnc&authkey=CLeK150N&gid=4 Nov 22 11:09:33 here are my 'bechmarks' according to disk space usage Nov 22 11:10:08 minimum disk space needed to build one image is 20Gb at this time Nov 22 11:11:01 Jay7: so, bitbake will do rm_work after image creation? Nov 22 11:11:19 before creation Nov 22 11:11:35 Jay7: Thanks! Nov 22 11:11:59 second small peak on graph is image creation imho Nov 22 11:12:35 khem: ping Nov 22 11:12:37 03Stefan Schmidt  07master * r7abb690089 10openembedded.git/conf/machine/bug20.conf: Nov 22 11:12:37 bug20.conf: Use beagleboard config for x-load Nov 22 11:12:37 We don't provide a x-load config for the bug20 in OE yet. Still omap3.inc adds Nov 22 11:12:37 it to the image depends which makes all images fail for this machine. Avoid this Nov 22 11:12:37 by building the beagleboard x-load config until we build our bug20 x-load with Nov 22 11:12:37 OE. Nov 22 12:20:27 pb_ any opinion on my update-rc.d patch (as you are the maintainer): see this (original msg seems not archived): http://thread.gmane.org/gmane.comp.handhelds.openembedded/39792 Nov 22 12:21:32 mickeyl, cbrake: any news about my ssh key? :) Nov 22 12:22:01 eFfeM_work: I think the answer to Koen's original question is "because the OE convention is to keep local patchsets separate from the upstream source files". Nov 22 12:22:28 eFfeM_work: as to whether your patch would make sense to apply to the "upstream" version, my guess is probably yes although I haven't reviewed it in detail. Nov 22 12:22:52 there is no particular urgency to that though and I would be happy to just leave it as it is for now. Nov 22 12:24:06 pb_ there is an urgency, the recipe as it is now does not fetch any more as handhelds.org seems gone. my patch is not really a patch, it just has replaced the cvs fetch with an inlined/local copy of update-rc.d Nov 22 12:24:48 eFfeM_work: sorry, I meant that there is no urgency to apply the local patch "add-verbose.patch" to the upstream sources. Nov 22 12:24:49 currently dev head does not build for me with an emptied downloads folder (distro minimal) as update-rc.d fails Nov 22 12:25:01 your change to make a local copy of the upstream version is fine. Nov 22 12:25:18 ok, can you add an ack (or may I consider this to be the ack) Nov 22 12:25:30 you can take this as an ack, go ahead and check it in Nov 22 12:25:34 ok Nov 22 12:26:14 will do so in a short while Nov 22 12:27:04 btw is bernhard reuter-fischer here? or does he not frequent irc? Nov 22 12:29:19 pb_ will do the patch tonight, it is not on patchwork and gmail is a pain to fetch patches from, but I have a copy at home (where I made the patch) Nov 22 12:29:25 ok, cool Nov 22 12:29:34 apparentlly patchworks has not picked up patches after nov 19 :-( Nov 22 12:30:15 eFfeM_work: I think it too... strange Nov 22 12:46:03 guess someone needs to kick it, maybe ka6sox-work can do so when he wakes up Nov 22 13:01:28 we have regular problems with infrastructure Nov 22 13:01:32 that is bad Nov 22 13:12:28 Greetings! I'm writing an article-thingie about OE and I read in a manual from 2006 that OE contained machine configurations for around 60 target machines, how high is this number atm? Nov 22 13:13:25 bitbake@jama ~/dev/conf/machine $ ls -1 *.conf | wc -l Nov 22 13:13:25 297 Nov 22 13:14:51 Thanks! Nov 22 13:16:19 * Jay7 write last time 'about 300 machines' :) Nov 22 13:16:38 hehe :) Nov 22 13:18:38 azariah4: might be of interested: http://files.meetup.com/1590495/openembedded.pdf Nov 22 13:19:03 azariah4: a presentation about OE from Khem (done sometime this year) Nov 22 13:20:24 rschus: Cool! Thanks! Nov 22 13:34:30 rschus: do you know when this presentation was given ? Nov 22 13:35:35 rschus: place a link to presentation somewhere on wiki please Nov 22 13:35:42 we should have it in single place Nov 22 13:35:56 ericben: the file says it was created on march 19th Nov 22 13:36:00 looks quite old Nov 22 13:36:00 this year Nov 22 13:36:43 this is because I try to keep the list on presentations about OE here : http://wiki.openembedded.org/index.php/Presentations Nov 22 13:40:07 this presentation is now mirrored on this page Nov 22 13:40:15 It was a couple of months I worked with oe now and have a small problem. I'm adding some extra files to my local lighttpd recipe. domainname.se.crt and domainname.se.pem. The lighttpd doesn't have the "files" directory which I was expecting. Instead it has it's lighttpd dir where the config file, index.html etc are located. I put my 2 files in there as well and added them to SRC_URI. I also added two "install"-lines. The error I'm getting when builing is ".... Nov 22 13:40:15 ....i486-angstrom-linux/lighttpd-1.4.18-r7.1/lighttpd/domainname.se.crt': Not a directory". My recipe: http://pastebin.com/wtzRgyg0 Nov 22 13:42:35 Jay7: just wanted to add khem's talk. someone was faster than me Nov 22 13:42:50 rschus: :) Nov 22 13:43:14 each time I find a presentation I put it here + a local mirror on the wiki Nov 22 14:04:36 I can't see why it works with lighttpd.conf but not with my custom files. They are all in the same directory and installed the same way Nov 22 14:05:44 Is it ignoring them because of an "unknown" file extension or something? Nov 22 14:09:24 I guess the question is: why are the files not being copied to 'work' Nov 22 14:09:51 jonvox: did you try cleaning and rebuilding? Nov 22 14:12:02 That made it! Nov 22 14:12:04 * jonvox hides Nov 22 14:13:13 cleaning will make it unpack the files again, thus making them available in ${WORKDIR} Nov 22 14:16:56 Ok, thanks a lot Nov 22 14:17:21 np Nov 22 14:28:54 khem, ping Nov 22 14:31:01 khem, pristine oe master doesn't build this local.conf (see private msg). Umong the problems is fux0red up largefile handling in oe / util-linux Nov 22 14:32:03 khem, (why on _earth_ does micro mess around with util-linux ?!) Nov 22 14:51:18 Is there an alternate to HandheldCVS? This site seems down and a few mandatory .bb files rely on it: "cvs [checkout aborted]: connect to anoncvs.handhelds.org(128.31.0.25):2401 failed: Connection timed out" . resulting in a permanent error Nov 22 14:57:31 Jordinar: cvs-archives was uploaded to mirror, but people was not decided yet how to deal with cvs recipes Nov 22 14:59:15 So what's the best choice now? looking for a mirror in that case or relax until handhelds.org is up again? Nov 22 14:59:55 their site seems to be down quite often .... if you enter "handhelds.org" in google, it offers you automatically "handhelds.org down" as autocompletion Nov 22 15:01:43 Jordinar: you can add : INHERIT += "own-mirrors" and SOURCE_MIRROR_URL = "http://sources.openembedded.org/" to your local.conf, this should workaround the handhled's cvs problem Nov 22 15:01:47 blindvt`: are you building uclibc_git (NPTL)? Nov 22 15:01:50 Jordinar: it's not clear when it will be back up... the admin has been out of general contact for some time Nov 22 15:02:35 ericben: thanks for your mail, looking forward to the patch Nov 22 15:02:56 ericben: Does own-mirrors catch that tho? Nov 22 15:03:11 Tartarus: this seems unless I made a mistake here Nov 22 15:03:22 I had it try and fetch from CVS still even tho I have an own mirror setup (and have and have been populating it since before hh.org went away) Nov 22 15:03:22 Tartarus: 'morning there Nov 22 15:03:27 morning ant Nov 22 15:03:29 own mirrors adds a PREMIRRORS for all sort of sources Nov 22 15:03:33 including cvs Nov 22 15:03:36 yeah Nov 22 15:03:42 But is it really catching this case :) Nov 22 15:04:14 well, my build host has fetched update-rc so until I put the file here and I don't remember that it seems that own mirror did the job Nov 22 15:04:22 I had a failure locally Nov 22 15:04:30 And I have apps.update-rc.d_anoncvs.handhelds.org_r0_7_.tar.gz in my mirror Nov 22 15:04:51 Well, here's hoping it was a transient error then, heh Nov 22 15:05:37 Tartarus: ah ok interesting. formt he date it seems the file was automatically fetched here Nov 22 15:05:57 I'm removing it for the next build Nov 22 15:06:58 I chose machines using cpu I have on my board (armv4, v5, v6, v7) for testing the release and it seems several machines' default kernels fail to compile ... Nov 22 15:07:59 I'm not sure I'll be able to fix these errors as I don't have boards to test the kernels Nov 22 15:09:17 Tartarus, about the Gentoo's perl not using threads, maye one of your commit forcing the use of /usr/bin/env perl is the cause... Nov 22 15:12:59 hm, anyone an idea on this: You asked to pull from the remote 'git://git.busybox.net/uClibc++', but did not specify Nov 22 15:12:59 a branch. Because this is not the default configured remote Nov 22 15:13:11 I have already added ;branch=master to the SRC_URI Nov 22 15:16:44 JaMa|Wrk: have you built perl with 'ithreads' use flag? Nov 22 15:17:28 eFfeM_work, that's a git message, isn't it? (as opposed to BB) Nov 22 15:17:50 HopsNBarley: yes, but oe generates the cmd line, btw this is not on oe head Nov 22 15:18:18 git isn't sure where to merge what you are fetching. Nov 22 15:18:55 ant_work: it shouldn't since /usr/bin/env perl should be returning our perl Nov 22 15:18:59 guess so, but no idea how to tell it that I just want a specific rev Nov 22 15:19:13 That said, there's always git bisect and git bisect run :) Nov 22 15:19:20 Tartarus: something is peeking at host perl Nov 22 15:19:37 and requiring threads support Nov 22 15:20:23 since let say a week Nov 22 15:21:02 ant_work: try to rebuild it with ithreads Nov 22 15:21:54 eFfeM_work, i believe you can supply tag=commit_id Nov 22 15:22:04 ant_work: or try to bother Roman Khimov about this Nov 22 15:22:12 HopsNBarley: will try Nov 22 15:30:35 ericben: this override doesn't work ... it seems that these 2 lines in local.conf don't influence anything Nov 22 15:36:36 is there a sort system like the use flags in gentoo for the recipes? Nov 22 15:36:53 the recipe that fails, has: SRC_URI = "${HANDHELDS_CVS};module=apps/update-rc.d;tag=r0_7 ... Do I need to change there in case of hh.org down? Nov 22 15:47:18 ant_work: no Nov 22 15:47:51 no "This Perl not built to support threads" errors? Nov 22 15:48:21 haven't seen it in my builds Nov 22 15:48:58 bibake master fwiw Nov 22 15:49:52 kergoth_: you around? Nov 22 15:50:07 1.10 here Nov 22 15:51:11 hello all iam trying to use oe with pre-built-chain , iam having some doubts can any body help me ..? Nov 22 15:51:45 JaMa|Wrk: 5.12.1 Nov 22 15:52:13 well, perl-5.12.2-r1, last stable Nov 22 15:52:24 for x86 Nov 22 15:52:38 [ebuild R ] dev-lang/perl-5.12.2-r3 USE="gdbm -berkdb -build -debug -doc -ithreads" 0 kB Nov 22 15:52:57 hm.on server I have dev-lang/perl-5.12.2-r2 Nov 22 15:53:04 at home should be the same Nov 22 15:53:06 ~amd64 Nov 22 15:53:06 i heard amd64 is See http://www.debian.org/ports/amd64/ and the #debian-amd64 IRC channel, or at http://www.jukie.net/~bart/debian/amd64/ Nov 22 15:54:31 JaMa|Wrk: I dont' have any ~x86 Nov 22 15:54:51 so is supposed to be 'stable' and it was until last week Nov 22 15:55:10 ..now I'm puzzled... Nov 22 15:56:21 ant_work: if you can say when I should get this error I can try it (i don't build many perl recipes in my images) Nov 22 15:57:08 http://fr.pastebin.ca/1997517 Nov 22 15:58:59 what's on /oe/build/tmp/sysroots/i686-linux/usr/bin/automake line 139/144? Nov 22 15:59:21 I can check later Nov 22 15:59:46 do you have it _after_ if ($perl_threads) ? Nov 22 16:00:04 sorry, I'm on different host Nov 22 16:00:14 ok Nov 22 16:00:47 and probably I already cleaned tmpdir... Nov 22 16:01:03 hello all i would like to know what should BBFILES contain in local.conf Nov 22 16:01:29 JaMa|Wrk: np, I can reproduce it (I did) Nov 22 16:09:12 blindvt`: your best bet with uclibc is minimal-uclibc distro, I dont make changes to micro so much Nov 22 16:09:33 blindvt`: it might be missing largefile feature in DISTRO_FEATURES Nov 22 16:10:03 blindvt`: and util-linux could have been pulled via some other package you have check the dep graphs Nov 22 16:10:20 there is lot of fat that gets pulled in into image thats is not needed Nov 22 16:10:31 stefan_schmidt: hi, whats up Nov 22 16:13:04 can any body say why am i getting this error " bitbake icedtea Nov 22 16:13:04 ERROR: IO Error: file 'conf/bitbake.conf' not found Nov 22 16:17:00 khem: just wanted to know if you want the git ref for commits that should go into the release branch or a patch on the ml. Nov 22 16:17:13 khem: I just sent the patch, so the question is outdated :) Nov 22 16:17:25 as i'm trying to update my embedded sytem by burning a new jffs2 filesystem on NAND, i'd like to know if there's a way to crypt the image file so that the user won't be able to open it or update the device with its own custom filesystem ? Nov 22 16:22:42 useing oe i can get the cross compiled exe is it not can any body confirm me this ..? Nov 22 16:23:29 can any body confirm me this plz...? Nov 22 16:32:40 we should force HH.org problem resolution Nov 22 16:33:17 kergoth: https://github.com/foerster/bitbake/compare/parallel-parsing...cache-progress Only problem is that we need to get the UI and server to run in different processes so the UI could actually handle the events. Nov 22 16:34:29 Could patch it into parse_next, waiting for the cache to signal that it's done loading, but seems hacky. Would be nice if the UI and server were separated early on. Nov 22 16:36:03 agreed, I disliked having to do the self.started crap, rudimentary state machine really, had to to hand control back to the UI Nov 22 16:36:15 does unpickling it that way slow it down, or does it take a similar amount of time? Nov 22 16:36:39 in my runs, it was oddly faster by .4 seconds Nov 22 16:36:57 hah Nov 22 16:36:59 nice Nov 22 16:37:26 right near 8.2 seconds each time, either way. Nov 22 16:37:47 i was confused because cooker.serve is never called with _none_ server. Nov 22 16:38:07 * foerster shakes head Nov 22 16:39:24 hmm, wonder if we could use what you've done there to more incrementally update it with new information, avoid having to hold the full .depends_cache dict in ram at all Nov 22 16:39:43 before the parallel parsing changes, that dictionary took up like 130 megs of ram Nov 22 16:39:51 iirc, anyway Nov 22 16:39:53 wow Nov 22 16:40:01 should check again Nov 22 16:40:22 i accidentally did a "p self.depends_cache" in the debugger ;) Nov 22 16:40:26 https://github.com/kergoth/bitbake/commit/cf5c7261aad184f7f17cb0511fe814ae1f716262 Nov 22 16:40:28 hehe Nov 22 16:42:55 nicely done, by the way, anything we can do to make it *feel* less slow is a good thing. the user needs to know we aren't just hung doing nothing :) Nov 22 16:43:24 thanks. Just need to figure out what to do with the separation of processes. Nov 22 16:43:43 Can't we just put the server in a new process in bin/bitbake? Nov 22 16:45:27 Funny thing - I used to wonder what the hell it was doing sometimes at the beginning. Once in a while it would seem like it sat there for quit a while befor saying "parsing.." Nov 22 17:01:09 wondering why some of my builds show in tinderbox as 'started' even though they completed on my end Nov 22 17:01:43 Jordinar: Tartarus : http://pastebin.com/fQQYRkft Nov 22 17:03:27 tharvey: +1 Nov 22 17:03:43 in http://wiki.openembedded.org/index.php/Testing should we add sep entries to the table for the testting_ branch vs release-2010.12? Nov 22 17:04:24 tharvey: I see no point Nov 22 17:04:36 now release-2010.12 is current testing branch :) Nov 22 17:04:46 oh... didn't realize that Nov 22 17:05:07 technically it isn't so Nov 22 17:05:13 but logically yes Nov 22 17:05:18 the tested configs that cbrake posts when he tags - they seem automated, where do they come from? Nov 22 17:05:51 from this table Nov 22 17:05:53 just want to make sure that once we move on to another branch that the details of what was tested and results for release-2010.12 are easy to find Nov 22 17:06:11 ie in a wiki page that discusses release-2010.12 Nov 22 17:07:24 ah, now I see Nov 22 17:07:28 this may be good idea Nov 22 17:07:49 after release we can copy this table to release page Nov 22 17:07:53 x86 angstrom-2008.1 x11-image failed on release-2010.12 in vlc/x264_r2245.bb http://tinderbox.openembedded.net/public/logs/task/10610828.txt Nov 22 17:08:09 and start working on some [semi]automated way of reporting Nov 22 17:08:09 ya, we should make it really easy for non devs to find info on tagged release Nov 22 17:08:10 it fails even on arm Nov 22 17:09:27 http://tinderbox.openembedded.net/public/logs/task/10303075.txt Nov 22 17:09:42 have not seen this.. Nov 22 17:10:00 I'll run testbuilder this night again then Nov 22 17:10:37 I think I successfully built x86 x11-image last test round, but didn't get it into tinderbox :( Nov 22 17:13:03 oops sorry that was x86 console-image I was building not x11-image, suprised that x264-r2245 is even built for that Nov 22 17:23:57 I think we should have a sep table for the release, if not a sep page since there will be one at some point Nov 22 17:24:26 ericben: ok, thanks Nov 22 17:28:45 hmm tinderbox need oestats_stop to be called for it to think a build has completed it looks like, not clear what calls that and when Nov 22 17:36:18 03Frans Meulenbroeks  07org.openembedded.dev * re4721e70af 10openembedded.git/recipes/php/ (php-native_5.3.2.bb php.inc php_5.2.13.bb): (log message trimmed) Nov 22 17:36:18 php: fixed path Nov 22 17:36:18 with the arrival of a new release the old php tarball moved to the Nov 22 17:36:18 museum site, Fixed SRC_URI and moved it from the inc file to the bb file Nov 22 17:36:18 so when adding the latest version there is no SRC_URI clash in the .inc Nov 22 17:36:18 file. Nov 22 17:36:19 Also introduced INC_PR Nov 22 17:39:48 03Frans Meulenbroeks  07org.openembedded.dev * r24acf53f55 10openembedded.git/recipes/update-rc.d/ (3 files in 3 dirs): (log message trimmed) Nov 22 17:39:48 update-rc.d: make update-rc.d local Nov 22 17:39:48 the recipe did not build any more as handhelds.org is gone. Nov 22 17:39:48 Since this one is fairly crucial as lots of recipes depend on it Nov 22 17:39:48 I decided to put the update-rc.d file in update-rc.d/ instead of Nov 22 17:39:48 fetching it. Nov 22 17:39:49 It has been decided (after discussion on irc with Philip Balister, the maintainer of Nov 22 17:46:21 I try to build base-image and I see a lot of unneded stuff builded like gnome-vfs, multimedia codecs and so on. Why this stuff included as a dependencies for this image? Nov 22 17:46:40 Build Configuration: BB_VERSION = "1.8.18" METADATA_BRANCH = "org.openembedded.dev" METADATA_REVISION = "f2bb9f9" TARGET_ARCH = "arm" TARGET_OS = "linux-gnueabi" MACHINE = "gms" DISTRO = "angstrom" DISTRO_VERSION = "2010.7-test-20101122" TARGET_FPU = "soft" Nov 22 17:47:24 plyatov_, do a bitbake -g base-image and look at pn-depends.dot for your answer Nov 22 17:48:32 there are a lot of seemingly simple things that require all kinds of support. One such that I run into often is gstreamer (has a million plugins and will build them all) and bluez (bluetooth - will build GUI components for various plugins even if you don't have display) Nov 22 17:48:43 tharvey, re oe-stats, it's very strange Nov 22 17:48:55 When I tried debugging that last week, it was calling _stop when it should Nov 22 17:52:07 Tartarus, hmmm... and still not showing as completed on tinderbox? when should it call? wanted to look at whatever logs are present for that to see if there was an error pushing stats - seems likely that there is no retry on failure Nov 22 17:52:24 So, lemme pull up the code Nov 22 17:52:43 When I was trying to debug Nov 22 17:52:56 BuildCompleted events happened and triggered calling oestats_stop Nov 22 17:53:04 (in both image failed and worked cases, iirc) Nov 22 17:53:30 Could be that yes, it needs to re-try on error Nov 22 17:53:35 Didn't dig too hard at the time Nov 22 17:56:57 but what triggers BuildCompleted event? a grep for BuildCompleted only shows tinderbox.bbclass checkign for that string - not clear where its created Nov 22 17:59:30 bitbake itself Nov 22 18:01:48 ah... so there would be no log then Nov 22 18:32:33 tharvey: my uclibc-git build broke on x264 Nov 22 18:33:25 ffmpeg should fail too Nov 22 18:34:12 (issue is missing log2f) Nov 22 18:34:14 a recent break? I vaguely recall building successfully last weeks testing branch Nov 22 18:34:23 ka6sox you there? Nov 22 18:35:03 tharvey: last week Nov 22 18:36:40 ant_work, ya we need to get that fixed Nov 22 18:36:54 * tharvey back after a reboot Nov 22 18:51:51 tharvey: khem knows about the issue. it seems stubs are missing Nov 22 18:52:28 so is the process to patch master and then get someone with commit to cherry-pick that patch into release? Nov 22 18:52:30 ericben|away: saw your email on premirrors, if I have more than one are they space separated ? or is there another separator Nov 22 18:52:42 or are we supposed to post a patch to each branch in the maillist? Nov 22 18:52:45 bbl Nov 22 18:56:01 03Frans Meulenbroeks  07org.openembedded.dev * r3ec3c40e42 10openembedded.git/recipes/php/php_5.2.13.bb: Nov 22 18:56:01 php_5.2.13.bb: fixed pathnames in pear.conf Nov 22 18:56:01 pear.conf had some hostnames in it. Nov 22 18:56:01 Added a sed line to remove STAGING_DIR_NATIVE from the file. Nov 22 18:56:01 Signed-off-by: Frans Meulenbroeks Nov 22 19:03:02 oops this commit message could have been a little bit better hostnames should have been host pathnames Nov 22 19:08:45 * kergoth reviews parallel parsing code to try to find Tartarus's issue Nov 22 19:14:15 hm get a md5sum failure on libsamplerate Nov 22 19:17:19 Tartarus: hmm, you know offhand which method is used to select one vs the other of those uboots? preferred_version, default_pref, compatible_*, ..? Nov 22 19:18:32 not off hand Nov 22 19:18:45 * kergoth pokes around Nov 22 19:30:58 hrm Nov 22 19:38:01 03Eric Bénard  07org.openembedded.dev * r8d1542c4bf 10openembedded.git/classes/ (amend.bbclass base.bbclass): (log message trimmed) Nov 22 19:38:01 handle PACKAGE_ARCH change in an amend.inc Nov 22 19:38:01 * setting PACKAGE_ARCH = "${MACHINE_ARCH}" in an amend.inc actually Nov 22 19:38:01 leads to a package with the correct machine arch but the Nov 22 19:38:01 work/stage/stamp dirs are not correct because MULTIMACH_ARCH is not Nov 22 19:38:02 updated after amend.inc is parsed. Nov 22 19:38:03 * this patch : Nov 22 19:40:05 eFfeM: for the mirrors, it's one per line Nov 22 19:43:25 ah ok Nov 22 19:57:54 hmm, I have a recipe that is called libdsm-api_git.bb and for some unexplainable reason the ipk that is generated is called libdsm-api-v2-1.0.0_0.0.0+gitr3+9720e5386d465f4f093cda861d5b9b7332503cda-r2.5_armv5te.ipk instead of libdsm-api_0.0.0+gitr3+9720e5386d465f4f093cda861d5b9b7332503cda-r2.5_armv5te.ipk Nov 22 19:58:05 this does not make sense to me Nov 22 19:58:08 where does that name come from? Nov 22 19:58:16 I tried setting PN = "libdsm-api" Nov 22 19:58:20 but that did not help Nov 22 20:05:06 Jin^eLD: look at PR Nov 22 20:05:15 PR="r2" Nov 22 20:05:47 http://pastebin.mozilla.org/862408 Nov 22 20:05:50 this is the recipe Nov 22 20:06:01 I am puzzled... Nov 22 20:06:49 it is PV Nov 22 20:06:50 0001-uclibc-moved-to-git-head.patch Nov 22 20:07:01 the library that gets generated is indeed called ./libdsm-api-v2.so.1.0.0 Nov 22 20:07:06 oops Nov 22 20:07:07 PV = "0.0.0+gitr${SRCPV}" Nov 22 20:07:30 but that does not explain libdsm-api-v2-1.0.0_0.0.0+gitr3+97.... Nov 22 20:07:50 name is something like ${PN}${PV} Nov 22 20:08:01 not exactly Nov 22 20:08:11 well, "v2-1.0.0" is not in PV and not in PN Nov 22 20:08:53 it generated this ipk libdsm-api-v2-1.0.0_0.0.0+gitr3+9720e5386d465f4f093cda861d5b9b7332503cda-r2.5_armv5te.ipk Nov 22 20:09:05 I understand that the PV stuff is after the first underscore Nov 22 20:09:27 it is a lib, those have name mangling Nov 22 20:09:33 hm Nov 22 20:09:40 look at another lib Nov 22 20:09:56 khem btw just mailed the final version of uclibc++ (git) Nov 22 20:11:50 it seems the previous version, before this checkout, did not mangle like that, so now I end up with a different package name and an unupgradable system Nov 22 20:12:38 I should be able to solve it via PROVIDES or RREPLACES or something Nov 22 20:16:57 Downloading http://192.168.0.22:50500/content/serve/ipk/armv5te/libdsm-api-v2-1.0.0_0.0.0+gitr3+9720e5386d465f4f093cda861d5b9b7332503cda-r2.5_armv5te.ipk. Nov 22 20:16:58 Segmentation fault Nov 22 20:17:00 oops Nov 22 20:17:03 sorry wrong window Nov 22 20:33:30 03Steve Sakoman  07org.openembedded.dev * re42fed4d39 10openembedded.git/recipes/gnome/ (2 files in 2 dirs): Nov 22 20:33:30 system-tools-backends: Fix broken angstrom distro patch Nov 22 20:33:30 Signed-off-by: Koen Kooi Nov 22 20:33:42 03Steve Sakoman  07org.openembedded.dev * r5960e445d6 10openembedded.git/recipes/linux/linux.inc: Nov 22 20:33:42 linux.inc: Fix logic error in CORTEXA8FIXUP test Nov 22 20:33:42 Signed-off-by: Koen Kooi Nov 22 20:35:02 hello Nov 22 20:36:07 I am a complete noob when it comes to OE. I'm working on a project for my embedded class. I think I have the partition portion, at least putting linux on the first partition down Nov 22 20:36:22 but now I need to put an OE image into the second partition, Nov 22 20:36:57 my question.. is there a nice step by step way of building my own OE image instead of using a prebuilt one? Nov 22 20:42:52 Beagle4: check the wiki Nov 22 21:01:08 Has anyone tried building a kernel with the SDK produced by meta-toolchain recently? Nov 22 21:01:35 It fails with a bus error during the first second of the build (in oldconfig it seems) Nov 22 21:01:55 Confirmed on two hosts running two versions of Ubuntu Nov 22 21:07:42 Maybe a stupid question: Is there an easy way in OE to get a generated list of the versions of the packages built into an image? Nov 22 21:10:44 TMPDIR/deploy/images/-testlab/installed-packages.txt Nov 22 21:21:28 Hmmm... Dont seem to have any generation of TMPDIR/deploy/images/-testlab/installed-packages.txt, maybe something is missing. Nov 22 21:21:49 INHERIT += "testlab" Nov 22 21:22:15 testlab another Angstrom goodie :-) Nov 22 21:24:02 * kergoth fixes a critical bug in parallel-parsing, anyone using it please pull Nov 22 21:26:11 i could use a 2nd ack for uclibc++ Nov 22 21:28:23 Just found my testlab files :-), Nice! Nov 22 21:29:46 Anybody know if there is an easy way of getting a generated list of the Open Source Licenses used in all packages built into an image? Nov 22 21:40:30 effeM: trade you one for libffi :-) Nov 22 21:48:30 deal, peeking at the libffi patch now Nov 22 21:48:35 and seeing if i can build it Nov 22 21:50:04 review is ok, if it builds ok, you have an ack Nov 22 21:56:38 builds fine too, ack sent Nov 22 22:05:44 effeM: I havent built you uclibc++ as I dont have a uclibc build, but my review of the recipe it looks good Nov 22 22:06:01 effeM: due to the low impact and obviously low number of users Ill give an Ack Nov 22 22:06:08 ok ty Nov 22 22:06:32 actually you do not need an uclibc build, you can also use uclibc++ with eglibc or glibc Nov 22 22:08:08 effeM: ah didnt realise, thought it was tied tightly Nov 22 22:08:23 * XorA thinks some relaxation is calling, cya guys tomorrow Nov 22 22:11:11 cya Nov 22 22:11:35 khem will do the uclibc++ license change in a separate patch, easier .... Nov 22 22:11:55 Hi. is there a fix for : cvs [checkout aborted]: connect to anoncvs.handhelds.org(128.31.0.23):2401 failed: Connection refused thx Nov 22 22:12:16 "Handhelds.org is undergoing renovations - The site should be active on the week of December 1st, 2010!" Nov 22 22:12:49 eFfeM: ok Nov 22 22:13:11 push of both patches is on its way Nov 22 22:14:01 03Frans Meulenbroeks  07org.openembedded.dev * r458075f8b0 10openembedded.git/recipes/uclibc++/ (4 files in 3 dirs): (log message trimmed) Nov 22 22:14:02 uclibc++: moved to git head Nov 22 22:14:02 Removed the 0.1.9 version as it is quite old and does not compile Nov 22 22:14:02 renamed files dir to uclibc++ Nov 22 22:14:02 added patches to get things compiling. Nov 22 22:14:02 Patches are gracefully picked from openwrt Nov 22 22:14:02 (https://dev.openwrt.org/browser/packages/libs/uclibc%2B%2B/patches?rev=24017) Nov 22 22:14:09 03Frans Meulenbroeks  07org.openembedded.dev * re27787857e 10openembedded.git/recipes/uclibc++/uclibc++_git.bb: Nov 22 22:14:09 uclibc++: fix LICENSE Nov 22 22:14:09 Signed-off-by: Frans Meulenbroeks Nov 22 22:24:43 khem, sane-toolchain.inc still has a comment that says: "# Prefer glibc 2.6 and uclibc 0.9.30, these have had the most testing." :) Nov 22 22:42:12 grg: heh Nov 22 22:42:21 grg: will correct them Nov 22 22:42:22 khem, morning Nov 22 22:43:51 khem, you mentioned largefile DISTRO_FEATURES yesterday regarding my busybox/uclibc failure... does this mean the ublibc distro(s) should have that added to DISTRO_FEATURES ? Nov 22 22:44:23 micro-uclibc distro that is. minimal-uclibc.conf looks like it has it already Nov 22 22:45:54 grg: actually the recipes should be tuned to disable large file support if the feature is not set Nov 22 22:46:05 but some places the detection happens stupidly Nov 22 22:46:12 ok ill have a go at that instead Nov 22 22:46:28 probably need to tweak the busybox config Nov 22 22:46:33 so to paper over the whole issue adding largefile to DISTRO_FEATURE Nov 22 22:46:37 would solve issue Nov 22 22:46:47 gym time Nov 22 22:47:03 enjoy Nov 22 22:47:16 grg: Unknown package 'xserver-kdrive-fbdev'. Nov 22 22:47:26 I get this when building x11-image Nov 22 22:47:30 for minimal Nov 22 22:47:35 hmm Nov 22 22:47:46 opkg then bails out Nov 22 22:47:58 | Collected errors: Nov 22 22:47:58 | * opkg_install_cmd: Cannot install package xserver-kdrive-fbdev. Nov 22 22:47:58 | * resolve_conffiles: Existing conffile /scratch/oe/rootfs/x11-image/etc/device_table is different from the conffile in the new package. The new conffile will be placed at /scratch/oe/rootfs/x11-image/etc/device_table-opkg. Nov 22 22:48:01 i am building with xserver-xorg in my local.conf Nov 22 22:48:42 probably also need opkg to say "this error is the fatal one" too Nov 22 22:50:07 i'd love to see the device_table thing fixed, but its got its tentacles dug in deeply Nov 22 22:51:58 device_table seems harmless Nov 22 22:52:33 I've seen this with bug about locale-base-fr-fr Nov 22 22:52:36 khem, xserver-kdrive-common.inc probably needs PROVIDES += "${PACKAGES}" . I seem to recall doing that at some stage for kdrive Nov 22 22:53:02 Jay7, everyone will get the non-fatal device_table error Nov 22 22:53:03 what build to start today? Nov 22 22:54:17 Jay7, it should be fixed so that a static /dev can be used again Nov 22 22:56:34 03Martin Jansa  07master * r237192e0a8 10openembedded.git/recipes/eglibc/ (eglibc-package.bbclass eglibc.inc): Nov 22 22:56:34 eglibc-package: move eglibc-dev to the end of PACKAGES var Nov 22 22:56:34 * currently it picks too much stuff to eglibc-dev package Nov 22 22:56:34 * update FILES_eglibc-extra-nss to include only .so files Nov 22 22:56:34 Acked-by: Khem Raj Nov 22 22:56:35 Tested-by: Graham Gower Nov 22 22:56:36 Signed-off-by: Martin Jansa Nov 22 22:56:36 03Martin Jansa  07master * rd28cdb2d2e 10openembedded.git/ (8 files in 4 dirs): Nov 22 22:56:37 xorg: new versions 2010-11-22 libx11, rendercheck, libxxf86misc Nov 22 22:56:37 Signed-off-by: Martin Jansa Nov 22 23:03:10 obi, Crofton|road, khem: what about cherry-picking gtk staging fix to release branch? Nov 22 23:03:56 calling it a day, cay Nov 22 23:04:11 JaMa: about perl threads: Nov 22 23:04:33 | This Perl not built to support threads Nov 22 23:04:33 | Compilation failed in require at /oe/build/tmp/sysroots/i686-linux/usr/bin/automake line 139. Nov 22 23:04:34 oh.. it is not commited in master yet even Nov 22 23:04:39 Jay7: sure, once it's in master ;) Nov 22 23:04:49 obi: what do you need to push? Nov 22 23:04:52 if ($perl_threads) Nov 22 23:04:53 { Nov 22 23:04:53 require threads; Nov 22 23:04:59 ah.. your key is broken Nov 22 23:05:18 JaMa: #139 require threads Nov 22 23:05:31 to be exact: my key is ok. the git-ssh-thingy on the server is broken ;) Nov 22 23:05:54 obi: ok, as you say :) Nov 22 23:05:55 apparently it doesn't activate new keys Nov 22 23:06:02 well, I can push it Nov 22 23:06:10 we have khem's ack Nov 22 23:06:19 and 3 tested-by Nov 22 23:06:36 great :) Nov 22 23:06:38 what about license? Nov 22 23:06:56 can be changed later, i'm not going to investigate it Nov 22 23:09:07 the thing is, that the LGPL/GPL usually says: "can be distributed under any version if no version is specified OR can be distributed unter specified and later versions if stated in the source code OR ELSE must be distributed under the specified version" Nov 22 23:09:24 obi: can you send patch directly to me? Nov 22 23:09:43 to be sure about signed-off lines and etc Nov 22 23:10:14 ok Nov 22 23:11:00 I'll look into sources about license Nov 22 23:12:11 would it be ok for you to pull from a git tree instead? Nov 22 23:13:09 obi: git-send-email you mean? Nov 22 23:13:26 or pull from your tree? Nov 22 23:13:30 pull from my tree Nov 22 23:13:53 well, say me url Nov 22 23:14:08 JaMa: bug found... http://bugs.gentoo.org/show_bug.cgi?format=multiple&id=343439 Nov 22 23:14:14 su - root Nov 22 23:14:24 rm -rf Nov 22 23:14:25 ;) Nov 22 23:14:32 gtk+-1.2 contains GPLv2 in COPYING file Nov 22 23:14:41 ant8468 Nov 22 23:14:46 LGPLv2 Nov 22 23:15:21 wireless keyboard went crazy.... Nov 22 23:15:27 UFO's around? Nov 22 23:15:30 source code contains header of LGPLv2+ Nov 22 23:15:38 obi: yes, LGPLv2 Nov 22 23:15:54 git://git.opendreambox.org/git/obi/openembedded.git for-upstream Nov 22 23:15:56 "version 2 of the License, or (at your option) any later version." Nov 22 23:16:08 so, I'll add LGPLv2+ as LICENSE Nov 22 23:16:16 did you check every file? ;) Nov 22 23:16:23 hehe.. Nov 22 23:16:30 i think nobody cared about GPLv1 yet Nov 22 23:16:45 we should have special man for this :) Nov 22 23:16:47 even though GPLv2 also allows switching to v1 Nov 22 23:16:56 if no version is specified in the source Nov 22 23:18:14 which means that some of the recipes marked as GPLv2+ are probably GPLv1+ Nov 22 23:18:39 so what to place there? :) Nov 22 23:19:09 obi: btw, did you know easy way to cherry-pick commit from remote tree? :) Nov 22 23:19:23 "git pull git://git.opendreambox.org/git/obi/openembedded.git for-upstream" Nov 22 23:19:26 that should work Nov 22 23:19:30 or should I clone tree and then use git-format-patch? Nov 22 23:19:31 http://git.opendreambox.org/?p=obi/openembedded.git;a=summary Nov 22 23:19:56 i rebased it on master 2 minutes ago Nov 22 23:20:08 6 minutes :) Nov 22 23:21:05 obi: just downloaded a patch from web-interface :) Nov 22 23:21:13 http://git.opendreambox.org/?p=obi/openembedded.git;a=patch;h=3830fcca86e7bd5ea9e5d3ee3db1e0931070049f Nov 22 23:21:58 you can push the two other patches as well Nov 22 23:22:24 the patch for swfdev-gnome contains a build fix Nov 22 23:22:39 should be included in the release Nov 22 23:23:05 and the other one has two acks Nov 22 23:23:10 yes, I see Nov 22 23:23:11 ok Nov 22 23:27:12 hm.. git am fails.. Nov 22 23:27:21 Jay7: who is our per expert? Roman? Nov 22 23:27:27 ant__: afaik, yes Nov 22 23:27:41 see ML history for perl commits :) Nov 22 23:28:11 ant__: post to ML and CC Roman Nov 22 23:29:05 I see a big problem here...automake-1.11 reads Automake/Config.pm for "our $perl_threads = 1;" Nov 22 23:29:17 it is "0" on my build host Nov 22 23:29:36 so the host perl (Gentoo one) is running... Nov 22 23:31:42 but reading config from sysroot :/ Nov 22 23:31:51 our $libdir = '/oe/build/tmp/sysroots/i686-linux/usr/share/automake-1.11'; Nov 22 23:34:52 03Andreas Oberritter  07master * r9fe39e4a31 10openembedded.git/recipes/gtk+/gtk+-1.2_1.2.10.bb: (log message trimmed) Nov 22 23:34:52 gtk+-1.2: convert to new style staging (fixes xmms build) Nov 22 23:34:52 * gtk+-1.2 didn't install its pkg-config file, but gtk-config Nov 22 23:34:52 depended on it and went crazy without it. xmms uses gtk-config. Nov 22 23:34:52 Signed-off-by: Andreas Oberritter Nov 22 23:34:52 Tested-by: Yury Bushmelev Nov 22 23:34:53 Tested-by: Philip Balister Nov 22 23:34:59 03Andreas Oberritter  07master * r0f0c9c2999 10openembedded.git/recipes/swfdec/swfdec-gnome_2.26.0.bb: Nov 22 23:34:59 swfdec-gnome: use the pkg-config file installed by swfdec Nov 22 23:34:59 * swfdec installs swfdec-0.9.pc, not swfdec.pc Nov 22 23:34:59 Signed-off-by: Andreas Oberritter Nov 22 23:34:59 Acked-by: Khem Raj Nov 22 23:35:02 03Andreas Oberritter  07master * r8f9e8f1a39 10openembedded.git/recipes/swfdec/ (9 files in 3 dirs): Nov 22 23:35:02 swfdec: remove unused old recipes Nov 22 23:35:02 Signed-off-by: Andreas Oberritter Nov 22 23:35:02 Acked-by: Khem Raj Nov 22 23:35:02 Acked-by: Frans Meulenbroeks Nov 22 23:35:09 obi: pushed Nov 22 23:35:12 thanks Nov 22 23:35:17 but gtk+ w/o LICENSE Nov 22 23:35:51 ant__: raise this in ML Nov 22 23:36:33 khem: how about cherry-pick commits above to release branch? Nov 22 23:37:38 I see other archs set # ac_cv_path_PERL=${ac_cv_path_PERL=no} Nov 22 23:47:42 I've started build of angstrom-gnome-image on release branch with commits above picked Nov 22 23:54:29 * Jay7 -> sleep Nov 22 23:55:42 Jay7: yes I will Nov 22 23:56:22 khem: I'll note you tomorrow about build status Nov 22 23:56:29 Jay7: msg sent Nov 22 23:56:31 all 3 commits cherry-picked here Nov 22 23:57:06 ant__: I hope it will be fixed quickly :) Nov 22 23:57:41 probably most distros don't see the problem Nov 22 23:57:56 Jay7: ok Nov 22 23:58:05 host-distro whose perl is compiled with threads Nov 22 23:58:29 ant__: btw, my ArchLinux have ithreads too Nov 22 23:58:37 khem: I guess this could be an automake issue ... Nov 22 23:59:32 btw, this build is accounting # of process spawned :) Nov 23 00:02:22 well... sleep! Nov 23 00:02:35 gn Nov 23 00:04:37 Jay7: I have picked few more patches Nov 23 00:04:42 git pull --rebase your tree Nov 23 00:25:37 ant__: around ? Nov 23 00:25:46 ant__: just replied to your message Nov 23 00:25:47 yep, Nov 23 00:25:52 can you remove Nov 23 00:26:04 do_configure_prepend Nov 23 00:26:09 and see if that helps Nov 23 00:26:30 btw forcing $perl_threads=0 let me finish the build Nov 23 00:26:45 in the Config.pm of sysroot Nov 23 00:26:45 sure you have found the cause of problem Nov 23 00:27:06 and a band-aid Nov 23 00:27:09 ant__: thats fine but while its using perl from sysroot then it should use everything from there Nov 23 00:27:16 ofc Nov 23 00:27:26 why does it use some stuff from sysroot and some for build box Nov 23 00:27:33 did u investigate that ? Nov 23 00:27:43 he, I suppose on first build you're using host perl Nov 23 00:35:12 khem: yes, fully removing do_configure_prepend allows e.g. libx11 to configure Nov 23 00:35:27 with perl-native Nov 23 00:35:47 that seems the culprit Nov 23 00:45:19 you have to fix for the non-native automake, though Nov 23 00:45:29 zZZ now Nov 23 00:45:32 good night Nov 23 00:46:38 do we have a place to provide a cache of sources once release is complete? seems like a release branch is useless without pinning sources Nov 23 00:48:07 tharvey: we do Nov 23 00:48:57 how is it being populated? for what machines/images? Nov 23 00:56:23 its source...and manually Nov 23 01:01:11 tharvey: hmm, I think OE has a sourcemirror but thats a backup Nov 23 01:01:20 it does not have everything there Nov 23 01:01:38 it still depends on web Nov 23 01:03:29 what si the process to get things stored there? I'm all for making sure the machines/images I test get their sources posted if I have access to Nov 23 01:04:19 I see this as pretty important... one of my release-2012-12 builds failed on clean DL_DIR b/c of some temporary issue with a specific source host Nov 23 01:05:35 you put the source somewhere I can access Nov 23 01:05:41 and I can put them there Nov 23 01:05:51 also put the .md5 files Nov 23 01:06:01 now I gotta drive home Nov 23 01:06:05 bbiab Nov 23 01:39:08 strace fails to build with micro-uclibc, because its looking for struct statfs64. The kernel still has support for statfs64, but libc doesn't. Strace can't get the definition from libc, but needs it to decode the syscall. http://tinderbox.openembedded.org/public/logs/task/10863125.txt Nov 23 01:40:21 khem, should i just #ifdef out strace's statfs64 code and assume those syscalls will never be hit, or is there a better solution here? **** ENDING LOGGING AT Tue Nov 23 02:59:57 2010