**** BEGIN LOGGING AT Mon Apr 25 02:59:57 2011 Apr 25 10:12:48 morning Apr 25 13:43:30 HELLO WORLD! Apr 25 13:44:09 Does anyone know which recipe python's contextlib.py comes from? I need this in my image! Apr 25 13:44:59 I really prefer not to just copy that from my build host's directory. How do i do this properly? Apr 25 13:54:20 JavaRocky: looks like you need the package python-misc Apr 25 13:54:38 YO foerster!!! Apr 25 13:55:03 foerster: did you know that from memory or was there a method you used to find this out? Apr 25 13:55:33 method Apr 25 13:56:07 can i learn your kung-fu? Apr 25 13:56:14 cd tmp/work/python-2.6.6-ml12.2/packages-split Apr 25 13:56:18 find . -name context* Apr 25 13:56:36 /python (missed a slash) Apr 25 13:56:50 if you have rm_work enabled, this stuff won't be there Apr 25 13:57:52 phoa Apr 25 13:58:52 foerster: i can see the file is there, but i dont understand how you made the connection to python-misc! Apr 25 13:59:23 the python recipe provides all packages in packages-split Apr 25 13:59:30 so, the leading dir name is the package name you need Apr 25 14:00:09 there is also a search on angstrom site which you can use to find files in packages Apr 25 14:01:04 i get it now.. ahhh. thank you so much Apr 25 14:01:09 np Apr 25 14:01:23 ive been banging my head for hours Apr 25 14:01:54 been there, done that. Apr 25 14:02:18 so painful Apr 25 14:02:25 or maybe not, I misremebered Apr 25 14:02:55 XorA|gone: cant find what you are refering to Apr 25 14:03:10 see above :-D Apr 25 14:03:15 ahh i c Apr 25 14:03:34 im using network boot to test my images.. does anyone else do this? or is that old hat? is qemu the way to go? Apr 25 14:04:40 bad thing about network boot is, my build machine is not setup to allow internet connectivity to my overo embedded device.. bummer.. Apr 25 14:55:18 03Søren Holm  07master * raaf30cc3e1 10openembedded.git/recipes/wvstreams/wvstreams_4.6.bb: Apr 25 14:55:18 wvstreams: update version from 4.6.0 to 4.6.1 Apr 25 14:55:18 This patch is build tested using `angstrom-2010.x` and `minimal` for `MACHINE = "beagleboard"`. The build still fails when uClibc is used. Apr 25 14:55:18 Signed-off-by: Søren Holm Apr 25 14:55:18 Signed-off-by: Paul Menzel Apr 25 16:33:11 Hey all. Apr 25 16:34:58 anyone seeing checksum failures for ncurses_5.7? Apr 25 16:49:39 03Tom Rini  07master * r7abb8142f9 10openembedded.git/recipes/gstreamer/gstreamer.inc: Apr 25 16:49:39 gstreamer: Add bison-native and flex-native to DEPENDS Apr 25 16:49:39 Signed-off-by: Tom Rini Apr 25 17:13:57 bah Apr 25 17:14:07 patch.bbclass needs a minver/maxver to go with minrev/maxrev & co Apr 25 17:14:54 ahhh, nevermind, it looks like the patch disappeared and Kooen pulled it into the tree itself. *blech* Apr 25 17:14:55 bluez 4.91 fixes the sbc_math thing Apr 25 17:15:31 @Tartarus how would this help? Apr 25 17:15:38 other problem :) Apr 25 17:15:58 I think you just need to do a -c clean ncurses ncurses-native Apr 25 17:16:00 and things should be fine Apr 25 17:16:32 nah, the patch pulls down and fails the checksum. clean tree, new build, nothing in the local mirror. Apr 25 17:17:08 Kooen pushed a change into master that includes the needed patch as a file, instead of grabbing it from the remote website. Apr 25 17:17:50 @Tartarus, what would minrev/maxrev do? you're thinking that any patch in the range would be acceptable? Apr 25 17:18:40 well, what the counterparts do today is let you list a patch in an inc file Apr 25 17:18:49 and say but only use it when we're on srcrev whatever Apr 25 17:19:00 in this case I don't want to move the patch to 5 different bb files Apr 25 17:19:11 nor have to move the whole SRC_URI into the. .91 recipe Apr 25 17:19:22 ahhh. Apr 25 17:22:37 In trying to write a recipe for Nginx, I've run into this package-qa error: "No GNU_HASH in the elf binary" I think this means I need to pass a flag to the linker, but it appears LDFLAGS is being ignored. Apr 25 17:23:02 pingswept: thats the root cause of the problem to start with Apr 25 17:23:13 pingswept: you need to add the option to compiler name Apr 25 17:23:35 hey khem. Apr 25 17:24:06 khem: I think I get that. I'll give it a try. Apr 25 17:25:03 pingswept: TARGET_CC_ARCH += "${LDFLAGS}" Apr 25 17:25:10 darknighte: howdy Apr 25 17:25:26 khem: Strangely, that doesn't work. Apr 25 17:26:02 pingswept: hmm, you might need to tinker with makefiles Apr 25 17:26:22 pingswept: can you paste the compile logs somewhere Apr 25 17:26:26 I think the problem is that I need to pass the linker option through the Nginx configure script. Apr 25 17:27:00 error: C compiler arm-angstrom-linux-gnueabi-gcc --hash-style=gnu is not found Apr 25 17:27:04 Hang on a second. Apr 25 17:27:20 pingswept: thats a linker option Apr 25 17:27:33 so compiler driver will barf rightly so Apr 25 17:27:48 Here's the recipe: http://www.pastie.org/1832076 Apr 25 17:27:58 khem: I have a question on bind. Apr 25 17:27:59 you have to add -Wl, in front of it to make sure gcc driver bypasses processing it and feeds it to linking step Apr 25 17:28:02 As you say, this generates barf. Apr 25 17:28:15 darknighte: shoot Apr 25 17:29:09 pingswept: you should inherit autotools Apr 25 17:29:14 in the recipe Apr 25 17:29:23 then remove those functions totally Apr 25 17:29:43 khem: do you recall the patch that Ryan submitted? Apr 25 17:29:45 you dont need to define do_configure and do_install Apr 25 17:29:53 darknighte: eglibc ? Apr 25 17:30:09 actually, this one was to bind, I believe. Apr 25 17:30:16 let me get a commit #. Apr 25 17:30:17 hang on. Apr 25 17:31:17 khem: it was adding a DISTRO feature for ipv6 to bind. Apr 25 17:32:49 pingswept: try something like this http://www.pastie.org/1832093 Apr 25 17:33:00 darknighte: ah I remember Apr 25 17:34:21 khem: it looks like Tom's commit: 4b632377f15b965904a1ba4316608f235f5c74ae rolls that back and mentions that you know why. Apr 25 17:34:49 khem: Nginx doesn't actually use autotools. Apr 25 17:34:57 But I think your other suggestion cracked it. Apr 25 17:35:08 khem: I just was wondering if that was correct and what the new mechanism is for it? Apr 25 17:35:22 khem: With -Wl,--hash-style=gnu, it works. Apr 25 17:36:35 darknighte: read http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg15346.html Apr 25 17:36:42 pingswept: ok Apr 25 17:36:58 Have any other feedback on this recipe? Apr 25 17:37:03 here's the final version: http://www.pastie.org/1832098 Apr 25 17:37:19 If it looks good, I can send a patch to the mailing list. Apr 25 17:37:24 pingswept: however the way you pass this option is not going to work in all cases please note that not all arches support GNU hash Apr 25 17:37:37 Ah, I see. Apr 25 17:37:45 khem: thanks. Ryan didn't mention that to me and I missed it on the list. Apr 25 17:37:52 you should use ${LDFLAGS} Apr 25 17:38:06 cool, one less patch to maintain. Apr 25 17:38:11 :) Apr 25 17:38:32 * khem waves at denix Apr 25 17:38:41 khem: How does ${LDFLAGS} work? It doesn't appear in bitbake -e. Apr 25 17:39:30 huh Apr 25 17:39:49 Is it TARGET_LDFLAGS? Apr 25 17:40:45 Actually, TARGET_LDFLAGS already contains the --hash-style option. Apr 25 17:41:22 pingswept: it should b exported in your environment Apr 25 17:41:40 TARGET_LDFLAGS, but it seems that LDFLAGS is not. Apr 25 17:42:13 s/TARGET_LDFLAGS/TARGET_LDFLAGS is exported/ Apr 25 17:42:17 pingswept: if TARGET_LDFLAGS has it then so should LDFLAGS for target recipe Apr 25 17:43:41 You mean that the contents of TARGET_LDFLAGS show up in LDFLAGS in the context of the recipe? Apr 25 17:50:58 khem: Thanks for the help. I'm still stuck on how to make LDFLAGS appear correctly, but I need to eat some lunch. Back later. Apr 25 18:06:13 ka6sox-work: around ? Apr 25 18:06:23 someone rebooted garnet Apr 25 18:06:32 and my temp is gone Apr 25 18:34:19 * Tartarus does work in a branch for once, send-email's, and switches back to master before he breaks something again Apr 25 18:34:22 * Tartarus pats self on back Apr 25 18:41:21 rofl Apr 25 18:41:52 * mwester anticipates a marked decline in effective productivity for Tartarus today. ;) Apr 25 18:55:18 JaMa: with a D_P_shr = "1", can I get an acked-by? Apr 25 18:55:28 sure Apr 25 18:57:57 mwester, your build yesterday pointed out to me 2 deficiencies in the autobuilder infra...I'm working now to fix that. Apr 25 19:00:28 looks like ncurses is recently broken. Apr 25 19:00:35 *still* Apr 25 19:00:42 * darknighte sighs heavily Apr 25 19:01:19 Is there a way to stick -M -C -C into ~/.gitconfig so it always does that for format-patch? Apr 25 19:02:35 anyone have a copy of "ncurses-5.7-20110115.patch.gz" Apr 25 19:04:13 apparently, the 5.7 tree has been pruned on invisible-island.net Apr 25 19:05:01 the 2011-03-maintenance still has a reference to that tree. Apr 25 19:12:43 Hi .... where can I get pwclient ? Apr 25 19:24:05 sgh: https://patchwork.kernel.org/pwclient/ Apr 25 19:24:40 sgh: and also look at https://patchwork.kernel.org/help/pwclient/ Apr 25 19:24:49 khem: thanks ...... where do I get a username/password. Apr 25 19:25:00 sgh: you need to register Apr 25 19:25:15 sgh: which patchwork site are you interested in Apr 25 19:25:25 khem: duh ... I see the link ... thanks again. Apr 25 19:26:56 Tartarus: under [format] Apr 25 19:27:18 ok Apr 25 19:27:20 what exactly? Apr 25 19:28:25 Tartarus: these options no Apr 25 19:28:28 :( Apr 25 19:28:47 That's what I thought :( Apr 25 19:29:16 you can set stuff like attach subjectprefix etc Apr 25 19:30:55 heh Apr 25 19:31:17 So, now to figure out why sometimes libusb1 says that no, clock_gettime doesn't come from -lrt and sometimes it does Apr 25 19:31:56 (which in turn makes bluez say that no, -lusb doesn't provide usb_get_busses and then use its own version of it and hit static vs non static decl bugs, one of which we hacked around, most of which we don't) Apr 25 19:38:51 Tartarus: lrt should have it Apr 25 19:38:58 Yes Apr 25 19:39:05 And if I go and -c clean it's right Apr 25 19:39:07 Tartarus: I wonder if there are two sources providing Apr 25 19:39:12 Or if I start from nothing and do it, it works Apr 25 19:39:21 But if I just say "build a big thing", it doesn't Apr 25 19:39:23 when does it fail? Apr 25 19:39:28 hmm Apr 25 19:39:28 And it's been that way a long long time it seems Apr 25 19:39:36 some race ? Apr 25 19:39:51 whats the test to determine it ? Apr 25 19:39:52 Not sure at all Apr 25 19:39:57 openembedded/recipes/bluez/bluez4/fix-dfutool-usb-declaration-mismatch.patch has been around since 2009 Apr 25 19:40:08 And is a symptom Apr 25 19:40:40 (since the real test failed back then too, almost certainly, with failed to link because clock_gettime needed by -lusb, rather than usb_get_busses was missing) Apr 25 19:41:27 I think looking at the configure test for it will give more clues Apr 25 19:41:41 * khem needs lunch Apr 25 19:41:54 heh Apr 25 19:42:59 khem: It's a pretty trivial test, getting it to fail will be the helpful part :( Apr 25 19:43:34 and maybe the hard part Apr 25 19:43:38 * Tartarus turns down threads Apr 25 19:44:05 khem, your build seemed to finish...but its missing rm_work Apr 25 19:58:07 btw Apr 25 19:58:20 I've asked already but that was on wheekend Apr 25 19:58:48 I'm trying to create recipe for gpredict: http://pastebin.com/CAZ92v5L Apr 25 19:58:59 but do_configure fails with http://pastebin.com/4uwPcEQM Apr 25 19:59:20 I'm not so experienced to fix that :) Apr 25 20:45:44 I've made 3 recipes using "inherit distutils" for Python packages. Mysteriously, they all want to install /usr/lib/python2.6/site-packages/site.py. Apr 25 20:45:58 This causes a conflict. Any ideas how to resolve it? Apr 25 20:46:26 site.py doesn't appear in any of the source tarballs. Apr 25 20:52:44 I see that recipes/iotop/iotop_0.2.1.bb has a do_install_append step that removes site.py. That seems like a workaround to the same problem I'm having. Apr 25 20:58:04 pingswept: i've seen that issue too Apr 25 20:59:02 kergoth: Should I just remove the site.py files like iotop does? Seems hackish, but I suspect it will work. Apr 25 20:59:42 i'm sure it will, better than nothing, but we need a real general solution, clearly Apr 25 21:00:24 I think the problem probably occurs whenever we have one Python package depending on another. Apr 25 21:00:51 Or at least, it doesn't occur for some solitary Python packages with no dependencies. Apr 25 21:01:19 Actually, that can't quite be right. Apr 25 21:01:32 kergoth: I would wager it is probably like the dir node for mkinfo. Each one adds it self to a manifest for the system, if it doesn't exist it generates it. Probably inhibit all, until and actual image build and generate for all. Apr 25 21:01:41 I have two packages that are not dependent on each other that conflict. Apr 25 21:02:05 I bet it's something in the setup.py files. Apr 25 21:02:08 ReaperOfSouls: yeah, or we need some sort of update- script to run in the postinsts Apr 25 21:03:50 kergoth: BTW did you see the message I sent about remote layering on the oe-core list. I wondered if you had any thoughts there. Apr 25 21:04:19 Jay7: this needs updates to configury Apr 25 21:04:23 haven't had a chance to review the code yet, will take a look Apr 25 21:04:50 Alot should feel familar. :) Apr 25 21:05:01 ka6sox-away: garnet got rebooted Apr 25 21:05:26 ka6sox-away: and it lost the new mapping for tmpdir that you did last time Apr 25 21:05:37 khem, yes, I added more RAM to it. Apr 25 21:05:46 it was swapping too much. Apr 25 21:05:52 ka6sox-away: ok then can you make this mounting in fstabs ? Apr 25 21:05:54 did you remap it? Apr 25 21:05:59 so it gets remounted Apr 25 21:06:02 no I did not Apr 25 21:06:23 okay its pointed to the right place now it seems. Apr 25 21:06:32 I will set it up right. Apr 25 21:06:33 actually I modified my script to use /scratch Apr 25 21:06:39 ah, okahy Apr 25 21:06:49 and you added rm_work too? Apr 25 21:07:11 /dev/xvda4 is /scratch Apr 25 21:07:19 not mapped into my tmpdir Apr 25 21:07:31 yes rm_work is on Apr 25 21:07:44 okay for this build do you want me to jsut leave it as youhave it? Apr 25 21:07:56 or stop and have me set it up correctly? Apr 25 21:08:30 I can stop Apr 25 21:08:42 ok its off Apr 25 21:10:51 khem: what should I do with that recipe? do I need a patch against configure.ac? Apr 25 21:11:51 khem its there. Apr 25 21:11:53 start over now. Apr 25 21:17:52 anyone have some pointers on how to walk backwards from a dependency to see what is causing it to be brought in? Apr 25 21:20:15 ka6sox-away: where did original contents of build go ? Apr 25 21:20:20 I have an old version of fakeroot-native being pulled in that doesn't exist on the mirror anymore. Apr 25 21:20:26 ka6sox: hey Apr 25 21:20:42 darknighte: usually bitbake -g is your friend Apr 25 21:22:00 khem: I must confess that i used the interactive shell almost exclusively and it being broken has made me relearn a few things. Apr 25 21:23:33 khem, buried...still on the old disk. Apr 25 21:23:38 do you need them? Apr 25 21:23:57 ka6sox-away: nah its ok I can recreate stuff Apr 25 21:26:52 anyone else notice that berlios.de appears to be down? Apr 25 21:30:36 down for me too. Apr 25 21:41:43 03Patrik Dahlström  07master * r03d5dcf53f 10openembedded.git/recipes/mozilla/firefox_3.6.8.bb: (log message trimmed) Apr 25 21:41:43 firefox: add correct CFLAGS and LIBS of libIDL-2.0 Apr 25 21:41:43 * autoconf does not fill in any CFLAGS or LIBS for libIDL-2.0. This Apr 25 21:41:43 patch sets the appropriate flags in autoconf.mk.in for Apr 25 21:41:43 HOST_LIBIDL_CFLAGS and HOST_LIBIDL_LIBS. Apr 25 21:41:44 The error message is the following [1][2]. Apr 25 21:41:45 […] Apr 25 21:46:43 03Paul Menzel  07master * rfd4a829460 10openembedded.git/recipes/ffmpeg/files/omapfbplay-errorhandling.patch: (log message trimmed) Apr 25 21:46:43 omapfbplay: move patch to new location after switch to Libav Apr 25 21:46:43 This patch fixes up commit 692d43ee [1] Apr 25 21:46:43 commit 692d43eeb6596df52b5984ae1bd14739f9ec9e0f Apr 25 21:46:43 Author: Koen Kooi Apr 25 21:46:44 Date: Mon Mar 28 15:47:23 2011 +0200 Apr 25 21:46:45 libav: add proper recipe for git version Apr 25 21:48:56 Jay7: yes I think it needs some upgrades to configure stuff Apr 25 21:49:07 Jay7: what version was it build for ? Apr 25 21:49:13 it should say at the top of configure Apr 25 21:49:14 file Apr 25 21:49:21 moment Apr 25 21:50:19 khem: did you mean that: # Generated by GNU Autoconf 2.65. Apr 25 21:50:38 yes Apr 25 21:50:51 but it has been overwritten in this case Apr 25 21:50:59 I was asking for original one Apr 25 21:51:08 without running do_configure Apr 25 21:51:10 ah, one from sources archive? Apr 25 21:51:11 it should show you Apr 25 21:52:14 # Generated by GNU Autoconf 2.67. Apr 25 21:52:30 it's after do_patch Apr 25 21:52:35 (no patches here atm) Apr 25 21:52:49 hmm so its ahead of us Apr 25 21:53:11 we use 2.65 Apr 25 21:53:23 there might be some macro incompatibilities Apr 25 21:53:46 while regenerating with newer versions is easier Apr 25 21:53:49 I was sure that running autoreconf should be enough.. Apr 25 21:53:52 this case is reverse Apr 25 21:54:41 you might have to tailor make the do_configure a bit here Apr 25 21:54:49 since you dont need to run autoreconf Apr 25 21:55:16 but may be write a do_configure which just runs oe_runconf Apr 25 21:55:37 well.. let's try Apr 25 21:55:38 you might be troubled by libtools stuff but give it a chance Apr 25 21:56:25 if maintainers wrote a sane confugure.ac we have enough reins to build without regenerating configure Apr 25 21:59:32 | configure: error: Your intltool is too old. You need intltool 0.21 or later. Apr 25 21:59:43 | checking for intltool >= 0.21... /var/tmp/oe/minimal/mipsel/work/mipsel-oe-linux/gpredict-1.3-r0/gpredict-1.3/configure: line 3906: intltool-update: command not found Apr 25 21:59:45 Jay7: yeah Apr 25 21:59:46 even this Apr 25 22:00:10 I think this would need major surgery Apr 25 22:01:04 automess.. Apr 25 22:05:50 khem: you got any suggestions on how to prune the useful information down from bitbake -g? Apr 25 22:07:49 darknighte: I think mwester had some perl thingy Apr 25 22:07:56 but I usually vi it and look around Apr 25 22:08:23 theoretically you can open the dot files in some dotty program Apr 25 22:10:27 Jay7: can you bitbake intltool-native Apr 25 22:10:57 Jay7: and get rid of do_configure override as you did last tim Apr 25 22:11:01 let's try again :) Apr 25 22:11:20 i.e. remove do_configure from recipe? Apr 25 22:11:24 yes Apr 25 22:12:11 just add DEPENDS = "intltool-native" Apr 25 22:12:19 to your recipe that should do it Apr 25 22:12:42 trying Apr 25 22:13:49 | No package 'goocanvas' found Apr 25 22:13:55 well, thats better Apr 25 22:14:16 khem: thanks, configure is passed that point Apr 25 22:14:22 that gooey stuff is out of autoconf Apr 25 22:14:26 now I should find what is goocanvas :) Apr 25 22:15:07 oh.. seems we have no recipe for it.. Apr 25 22:15:13 canvas widget for GTK+ Apr 25 22:16:21 http://developer.gnome.org/goocanvas/unstable/GooCanvas.html Apr 25 22:16:46 http://live.gnome.org/GooCanvas Apr 25 22:17:43 Jay7: I think you need to add recipe for goocanvas too Apr 25 22:18:08 yeah.. Apr 25 22:18:32 and you might need hamlib too :) Apr 25 22:18:48 for radio stuff Apr 25 22:18:54 it's commented out in configure.ac now Apr 25 22:19:00 so I'll skip this for now Apr 25 22:19:06 it's for kristianpaul ;) Apr 25 22:21:54 who is kristianpaul is it reincarnation of Paul the octopus Apr 25 22:22:10 he is in nick list here :) Apr 25 22:22:33 one from jlime community :) Apr 25 22:25:06 oh.. goocanvas 2.0 is depends on gtk+3.0 Apr 25 22:25:29 well.. I'll port older version then Apr 25 22:30:04 03Michael 'Mickey' Lauer  07org.openembedded.dev * r1e2bf7175b 10openembedded.git/ (2 files in 2 dirs): cmtspeechtest: bump rev; also add to fso-autorev.inc Apr 25 22:39:34 wow.. Apr 25 22:39:46 goocanvas just built (tm) Apr 25 22:41:02 btw Apr 25 22:41:36 is there any scenario when some recipe A is depending on workdir of recipe B? Apr 25 22:42:07 I mean working directory exactly Apr 25 22:42:09 no Apr 25 22:42:12 that's not allowed Apr 25 22:42:36 so workdir may be cleaned right after build? Apr 25 22:42:56 which is why rm_work exists, and works fine, yes Apr 25 22:43:11 ah, yes.. Apr 25 22:43:31 rm_work tasks run at the end of each recipe, not the end of the build, so.. Apr 25 22:43:32 heh Apr 25 22:43:56 I'm just thinking about building on tmpfs.. Apr 25 22:44:09 but current scheduling doesn't allow that.. Apr 25 22:44:23 kergoth, just before it does the rm_work is when I'd like to gather stats. Apr 25 22:44:43 ka6sox-away: rm_work may be disabled Apr 25 22:44:51 rm_work does switch to the 'completion' scheduler, which runs a single recipe to completion before moving on Apr 25 22:44:52 but I got the point :) Apr 25 22:45:01 I'm nots ure how the scheduler could possibly work any better than that for workdir cleanup Apr 25 22:45:16 hm.. Apr 25 22:45:35 Jay7, when I/O time was more of an issue than space I just let it roll. Apr 25 22:45:35 I should re-measure disk space requirements then Apr 25 22:45:39 but now thats not an issue. Apr 25 22:46:27 may be good idea to introduce some fake task like 'end_of_recipe' Apr 25 22:47:02 Jay7, I would like to know the size of the workdir before its deleted with rm_work. Apr 25 22:47:03 end of recipe is exactly what do_build *is* Apr 25 22:47:04 then we may catch it and send stats about task Apr 25 22:47:06 and the rm_work task Apr 25 22:47:15 just do what rm_work does to replace the last task with your own Apr 25 22:47:28 see the bbclass Apr 25 22:47:28 hm.. Apr 25 22:47:35 alternatively, use an event handler Apr 25 22:47:36 interesting Apr 25 22:47:42 watch for the completion of hte ${BB_DEFAULT_TASK} Apr 25 22:48:01 kergoth: is poky's bb ready for this as well? Apr 25 22:48:13 what do you mean? Apr 25 22:48:14 to be oe-core-compatible :) Apr 25 22:48:16 this isn't new, or magic Apr 25 22:48:29 BB_DEFAULT_TASK has existed for 7 years or so now? Apr 25 22:48:45 I'm newbie in that area :) Apr 25 22:48:53 the completion scheduler has been around for years as well Apr 25 22:48:57 as has rm_work Apr 25 22:48:58 so.. Apr 25 22:50:37 ok then :) Apr 25 22:51:01 well... gpredict passes do_configure Apr 25 22:51:42 nice, built ok Apr 25 22:52:09 kristianpaul: gpredict was built ok :) Apr 25 22:53:24 khem, the build seems to be doing most of the I/O outside of the workdir. Apr 25 23:02:42 03Tom Rini  07master * rc1fdf6f456 10openembedded.git/recipes/bluez/ (6 files): Apr 25 23:02:42 bluez4: Remove old and unused versions. Apr 25 23:02:42 Signed-off-by: Tom Rini Apr 25 23:02:52 03Tom Rini  07master * r7eee068ed5 10openembedded.git/recipes/bluez/ (bluez4.inc bluez4_4.42.bb bluez4_4.84.bb bluez4_4.89.bb): Apr 25 23:02:52 bluez4: Move a patch from the .inc to the bb Apr 25 23:02:52 Now that we're down to just three versions, move sbc-thumb.patch into Apr 25 23:02:52 the recipes because it's not an always need thing, just needed in Apr 25 23:02:52 some versions. Apr 25 23:02:52 Signed-off-by: Tom Rini Apr 25 23:02:55 03Tom Rini  07master * rc83fb2e381 10openembedded.git/recipes/bluez/bluez4_4.91.bb: (log message trimmed) Apr 25 23:02:55 bluez4: Add 4.91 and make it default for Angstrom and SHR Apr 25 23:02:55 Compile tested on am37x-evm and angstrom-2010.x. SHR already Apr 25 23:02:55 uses this version in oe-core/meta-oe. Apr 25 23:02:55 Acked-by: Paul Menzel Apr 25 23:02:55 Acked-by: Koen Kooi Apr 25 23:02:56 Acked-by: Martin Jansa Apr 25 23:03:04 03Tom Rini  07master * reb6af335c8 10openembedded.git/recipes/bluez/ (bluez4_4.84.bb bluez4_4.89.bb): Apr 25 23:03:04 bluez4: Remove 4.8* now that 4.9x is default for Angstrom and SHR Apr 25 23:03:04 Signed-off-by: Tom Rini Apr 25 23:18:19 khem: ping Apr 25 23:19:01 03Andrea Adami  07org.openembedded.dev * r8d9d679902 10openembedded.git/ (4 files in 3 dirs): Apr 25 23:19:01 geodegx: move to nonworking, obsolete linux-geodegx 2.4 kernel Apr 25 23:19:01 * machines with 2.4 kernels only are not supported Apr 25 23:19:01 Signed-off-by: Andrea Adami Apr 25 23:19:01 03Andrea Adami  07org.openembedded.dev * r1710d7230e 10openembedded.git/ (4 files in 3 dirs): Apr 25 23:19:01 mnci: move to nonworking, obsolete mnci-ramses 2.4 kernel Apr 25 23:19:02 * machines with 2.4 kernels are not supported Apr 25 23:19:02 Signed-off-by: Andrea Adami Apr 25 23:19:03 03Andrea Adami  07org.openembedded.dev * r730c85ac8f 10openembedded.git/recipes/linux/ (8 files in 2 dirs): Apr 25 23:19:03 linux-colinux: move unsupported 2.4 kernel to /obsolete Apr 25 23:19:04 Signed-off-by: Andrea Adami Apr 25 23:19:05 03Andrea Adami  07org.openembedded.dev * rd532cfddbc 10openembedded.git/recipes/linux/ipod_2.4.24-ipod0.bb: Apr 25 23:19:05 ipod: move unsupported 2.4 kernel to /obsolete Apr 25 23:19:06 Signed-off-by: Andrea Adami Apr 25 23:20:27 well.. 4Gb of tmpfs is enough for 497 tasks (4 bb threads) Apr 25 23:21:03 and no rm_work was called in this time Apr 25 23:21:12 bb scheduler is completion Apr 25 23:28:23 03Andrea Adami  07org.openembedded.dev * r784a4184e0 10openembedded.git/recipes/linux/ (3 files in 2 dirs): Apr 25 23:28:23 handhelds: move 2.4 defconfigs to /obsolete Apr 25 23:28:23 Signed-off-by: Andrea Adami Apr 25 23:37:18 03Andrea Adami  07org.openembedded.dev * r79084da933 10openembedded.git/recipes/linux/ (linux-sun4cdm-2.4.26/defconfig linux-sun4cdm_2.4.26.bb): Apr 25 23:37:18 sun4cdm: move unsupported 2.4 kernel to /obsolete Apr 25 23:37:18 Signed-off-by: Andrea Adami Apr 25 23:40:41 ok, recipes/linux is 2.4 freee now Apr 25 23:41:15 no cruft hopefully Apr 25 23:45:17 well, one more Apr 25 23:45:19 03Andrea Adami  07org.openembedded.dev * r53dee8bf61 10openembedded.git/recipes/linux/linux/tosa/defconfig: Apr 25 23:45:19 linux_git: remove stale tosa 2.6.28 defconfig Apr 25 23:45:19 Signed-off-by: Andrea Adami Apr 25 23:50:43 03Andrea Adami  07org.openembedded.dev * r269d9f3e6d 10openembedded.git/recipes/linux/linux_git.bb: Apr 25 23:50:43 linux_git: fix SRCREV for 2.6.39-rc4 Apr 25 23:50:43 Signed-off-by: Andrea Adami Apr 26 01:01:57 pingswept: you still working on the nginx recipe? Apr 26 01:16:55 can anyone explain what ${PV} is? the bitbake website hosting the manual is not responding Apr 26 01:18:19 or point to me where i can find more information about these acronym variables Apr 26 01:22:01 ahh version preference Apr 26 01:22:41 see example 4.7 - http://webcache.googleusercontent.com/search?q=cache:AV_H6RJc8CUJ:bitbake.berlios.de/manual/ch04s02.html+site:bitbake.berlios.de/manual&cd=2&hl=en&ct=clnk&client=ubuntu&source=www.google.com Apr 26 01:29:41 yeah, it's one of the most important variables, along with PN (the recipe name), PR (revision) and PE (epoch) Apr 26 01:29:45 (and others, of course) Apr 26 01:31:30 thanks kergoth. im stumbling around. Apr 26 01:32:00 there's lots to stumble around, and trip over, sadly :) Apr 26 01:32:09 i know my face hurts Apr 26 01:32:36 and if i want to customise the pv of a recipe in my local.conf, do i do something like.. BLAH_PV = "1.0.0" ? Apr 26 01:32:37 I keep hitting walls...but hopefully the light @ the end of the tunnel isn't an oncoming train. Apr 26 01:33:11 JavaRocky: you can use the bitbake overrides mechanism Apr 26 01:33:17 heh in sydney, the train doesnt come Apr 26 01:33:31 overrides are essentially a way to structure the metadata in a layered way. from a technical perspective, it's just a way to conditionally define variables Apr 26 01:33:47 first, it's important to know that global configuration metadata flows into the recipes, they start from the config Apr 26 01:34:06 next, you can define any variable_ for any entry in the OVERRIDES variable (colon separated) Apr 26 01:34:20 e.g. FOO_arm will overwrite the regular FOO if 'arm' is in overrides Apr 26 01:34:23 (the case for arm targets) Apr 26 01:34:37 so it lets us start with the most general information, and let the more specific, accurate information override it Apr 26 01:34:43 global -> architecture -> machine Apr 26 01:35:06 this is just background. to answer your specific question, you can define FOO_pn- Apr 26 01:35:11 because, pn-${PN} is in overrides Apr 26 01:35:52 I think the mental view of layers is helpful to grasping how the bits fit together, because it's not only the overrides that act this way, we also source our configuration files in roughly the same order, least specific to most specific Apr 26 01:36:20 (with one exception. local should be the most specific, since the user should always be able to override, but we don't know the machine/distro until the user sets them, so we can't parse the machine and distro files until after it) Apr 26 01:36:33 kergoth: reading & re-reading, im still here Apr 26 01:37:16 * kergoth thinks we need more high level conceptual documentation, since people seem to have a lot of trouble seeing the big picture and how the various bits fit together Apr 26 01:37:41 maybe i should try writing something up. there are a lot of pieces though, i'm not sure how to structure the document :) Apr 26 01:38:31 kergoth: i did read about the overrides concept in the bitbake manual, however all these variables ${PV} ${PN} i havent seen doco for it as yet Apr 26 01:39:05 the bitbake manual was basically written and then not touched again Apr 26 01:39:20 also it was primarily *reference* documentation, much about 1) the file format, and 2) commandline options Apr 26 01:39:29 phil and i write most of it many years ago Apr 26 01:40:14 kergoth: i am happy to help. i develop s/w as my day job so diving into these sorts of docs is a daily thing. so you can think of me as fresh meat, ive only been using OE for 2 days now Apr 26 01:40:58 that's always helpful, it's hard to go back and see from a fresh perspective when you're neck deep in it Apr 26 01:41:04 actually, if i could contrib to docs, hit me up! Apr 26 01:41:09 that would be better yet Apr 26 01:41:13 we don't have many doc folk at all Apr 26 01:41:18 i know exactly what you mean Apr 26 01:41:21 in my experience open source projects emphasize code way too much Apr 26 01:41:27 not enough doc, test, CM, etc Apr 26 01:41:48 when you are in the zone, docs are the last thing. heh Apr 26 01:42:15 i gotta jet now, but let me know how i can start contributing! Apr 26 01:44:34 what?? Apr 26 01:44:52 Jay7: gpredict, nice !! Apr 26 01:45:05 kergoth: BBL! Apr 26 01:46:21 khem: /win 13 Apr 26 01:46:23 oops Apr 26 01:46:44 khem: better no comments on the octopus comment ;) Apr 26 01:57:32 Yooo Apr 26 01:59:46 Yay. **** ENDING LOGGING AT Tue Apr 26 02:59:57 2011