**** BEGIN LOGGING AT Fri Apr 23 02:59:56 2010 Apr 23 05:18:49 Hey guys Apr 23 05:19:26 Can anyone point me to info on how I should put together a recipe that overrides some etc files in the final image? Apr 23 05:19:36 */etc Apr 23 05:21:40 porcoesphino, its best that you identify the recipe that puts the file into /etc in the first place Apr 23 05:22:03 this may not necessarily be easy Apr 23 05:22:11 it's not that hard Apr 23 05:22:54 ok, well if you know the recipe then you can copy it to an overlay and make the modifications you need Apr 23 05:23:11 I'd just prefer to have all my changes to things as minor as config files in one place, rather then override an openembedded recipe and then worry about maintaining changes that are made to it.. Apr 23 05:23:44 i agree that would be nice. Let me know if you find a way. Apr 23 05:23:45 and how do I make sure that changes are merged or that there is some notification of a difference when something changes upstream? Apr 23 05:24:10 it hardly seems worth overriding a recipe to change /etc/inteface, amoung others.. Apr 23 05:24:53 if the recipe uses file:// for the file, which it usually does, you can use FILESPATHBASE Apr 23 05:25:00 FILESPATHBASE =. "${TOPDIR}/files:" Apr 23 05:25:08 mkdir files/base-files Apr 23 05:25:17 cp openembedded/recipes/base-files/base-files/somefile files/base-files/ Apr 23 05:25:23 *edit as appropriate* Apr 23 05:25:42 alternatively, you can use ROOTFS_POSTPROCESS_COMMAND to run arbitrary shell commands to modify the rootfs Apr 23 05:25:50 those are the main options if you want to avoid modifying the recipe, at this time Apr 23 05:26:00 (or the recipe's files inside teh repo) Apr 23 05:27:09 so this? Apr 23 05:27:11 ROOTFS_POSTPROCESS_COMMAND += "patch -p /etc/interface < mypatch" Apr 23 05:28:11 ${D}/etc/interface Apr 23 05:28:14 but otherwise yes Apr 23 05:28:24 ${D} is the path to the rootfs in that context Apr 23 05:28:30 iirc anyway. Apr 23 05:30:04 cool cool Apr 23 05:30:08 I'll see how that goes Apr 23 05:30:09 thanks Apr 23 05:33:20 wait, surely I wouldn't reference the patch like that. Should I be saving it in SRC_URI? Apr 23 05:56:49 * BitWraith recompiles the kernel again, with cpu frequencey scaling... just to push his luck a little further. X-D Apr 23 06:45:01 good morning Apr 23 07:14:51 re Apr 23 09:06:15 good morning Apr 23 11:40:59 good morning Apr 23 11:41:16 good morning Apr 23 11:58:35 hi recalcati Apr 23 14:15:46 morning Apr 23 14:17:58 hi kergoth Apr 23 14:20:10 hey woglinde Apr 23 14:20:13 how's it going? Apr 23 14:22:01 damn scratchbox Apr 23 14:22:29 ther perl-deps always put the sb-perl deps in Apr 23 14:22:54 now I hardcoded the diablo stuff in the control file Apr 23 14:56:46 woglinde: heh, fun fun Apr 23 15:43:52 [STABLE] Hej! I'm getting "opkg: Cannot create directory `/var/lib/opkg': Permission denied" during do_rootfs in STABLE but not in .dev. Apr 23 15:43:52 What's the commit in .dev which fixed it? Someone remember? Was it just upgrading to svn r527? Apr 23 15:44:34 + opkg-cl -f /home/oe/build/out.angstrom/staging/i686-linux/etc/opkg.conf -o /home/oe/build/out.angstrom/rootfs -t /home/oe/build/out.angstrom/rootfs-tmp -force-depends remove 'kernel-image-*' update-modules Apr 23 15:44:35 opkg: Cannot create directory `/var/lib/opkg': Permission denied Apr 23 15:44:35 No packages removed. Apr 23 15:44:35 It should work, all options provided :-( Apr 23 16:15:10 Ahh, it's r527! My first try didn't pick up the newer version from svn Apr 23 17:02:46 03Koen Kooi  07org.openembedded.dev * ra63e30ddd7 10openembedded.git/recipes/xfce-base/ (13 files): xfce-base apps: convert to new style staging Apr 23 17:02:48 03David-John Willis  07org.openembedded.dev * re3df76459f 10openembedded.git/ (2 files in 2 dirs): xfce4-panel: Add 4.6.3 and support new style checksums in xfce46.bbclass. Apr 23 17:02:48 03Koen Kooi  07org.openembedded.dev * r7b077b8f6c 10openembedded.git/recipes/gpicview/ (gpicview_0.1.5.bb gpicview_0.2.1.bb): gpicview: add 0.2.1 Apr 23 17:02:49 03Koen Kooi  07org.openembedded.dev * r653467ce15 10openembedded.git/recipes/xfce-base/ (6 files): xfce-base libs: convert to new style staging Apr 23 17:02:50 03Koen Kooi  07org.openembedded.dev * r5b7e93d0c9 10openembedded.git/recipes/xfce-base/ (2 files in 2 dirs): xfce4-settings: add 4.6.4 Apr 23 17:02:50 03Koen Kooi  07org.openembedded.dev * r925f2861c2 10openembedded.git/recipes/tasks/angstrom-task-lxde.bb: angstrom-task-lxde: add tasks to base an LXDE image on Apr 23 17:02:58 03Koen Kooi  07org.openembedded.dev * reae2a94ea7 10openembedded.git/recipes/pmount/pmount_0.9.19.bb: pmount: fix config.rpath problem Apr 23 17:09:08 so silent today :) Apr 23 17:13:45 03Klaus Kurzmann  07org.openembedded.dev * ra641e40f32 10openembedded.git/recipes/freesmartphone/frameworkd_git.bb: Apr 23 17:13:45 frameworkd_git.bb: bump rev to get message timestamp fix Apr 23 17:13:45 Signed-off-by: Klaus Kurzmann Apr 23 17:15:30 03Klaus Kurzmann  07org.openembedded.dev * rfa68e15214 10openembedded.git/recipes/shr/e-wm-config-illume2-shr_git.bb: Apr 23 17:15:30 e-wm-config-illume2-shr_git.bb: bump rev to get adjusted config for current illume2 Apr 23 17:15:30 Signed-off-by: Klaus Kurzmann Apr 23 17:51:14 anyone have any comments on http://github.com/kergoth/OpenEmbedded/commit/2c130ea5209eadd902a698db2ac4229186424cb1, good or bad? Apr 23 18:04:05 hmm, we should think about adding gettext bits to bitbake's UIs at some point Apr 23 18:20:20 kergoth: patch looks ok to me Apr 23 18:27:58 03Koen Kooi  07org.openembedded.dev * r6246a13e68 10openembedded.git/recipes/opencv/ (opencv-samples_svn.bb opencv_svn.bb): opencv + samples: update to 2.1.0 Apr 23 18:44:00 kergoth: Did you have a chance to look into the problem about fetching that we discussed yesterday Apr 23 18:44:20 kergoth: I am still seeing it after deleting the tmp folder Apr 23 18:45:22 didn't have much luck reproducing it, but I also didn't spend a great deal of time on it. I'll see if I can reproduce it again this afternoon Apr 23 18:48:25 kergoth: hmmm Apr 23 18:48:39 could be I missed a step Apr 23 18:48:45 kergoth: I am looking at base_do_fetch Apr 23 18:49:16 bb.fetch.go(localdata) is called before base_chk_file Apr 23 18:49:38 so it should have figured out that archive has to be downloaded in bb.fetch.go Apr 23 18:49:40 isnt it Apr 23 18:50:23 03Roman I Khimov  07org.openembedded.dev * r41477e0b1d 10openembedded.git/recipes/linux-ha/ (2 files in 2 dirs): Apr 23 18:50:23 cluster-glue: fix uclibc build Apr 23 18:50:23 Stack protection is not available. Apr 23 18:50:23 Signed-off-by: Roman I Khimov Apr 23 18:50:25 03Roman I Khimov  07org.openembedded.dev * r76e6c7c4ae 10openembedded.git/recipes/linux-ha/ (2 files in 2 dirs): Apr 23 18:50:26 cluster-resource-agents: disable doc build Apr 23 18:50:26 Unreliable (breaks from time to time) and VERY time consuming. Apr 23 18:50:26 Signed-off-by: Roman I Khimov Apr 23 18:50:28 03Roman I Khimov  07org.openembedded.dev * r4fc81a5cf4 10openembedded.git/recipes/corosync/ (corosync_1.2.0.bb files/corosync.conf): Apr 23 18:50:28 corosync: update to version 1.2.1 Apr 23 18:50:28 Signed-off-by: Roman I Khimov Apr 23 18:50:33 03Roman I Khimov  07org.openembedded.dev * r2aafa2db54 10openembedded.git/recipes/linux-ha/ (2 files in 2 dirs): Apr 23 18:50:33 cluster-resource-agents: fix uclibc build Apr 23 18:50:33 Software stack protection is not available. Apr 23 18:50:33 Signed-off-by: Roman I Khimov Apr 23 18:50:44 03Roman I Khimov  07org.openembedded.dev * r7b77d80145 10openembedded.git/recipes/linux-ha/cluster-resource-agents_1.0.2.bb: Apr 23 18:50:44 cluster-resource-agents: update to version 1.0.3 Apr 23 18:50:44 Signed-off-by: Roman I Khimov Apr 23 18:50:48 03Roman I Khimov  07org.openembedded.dev * rf6f1291f09 10openembedded.git/recipes/corosync/ (corosync_1.2.0.bb files/volatiles): Apr 23 18:50:48 corosync: fix volatiles Apr 23 18:50:48 populate-volatile.sh doesn't like blank lines in volatile files. Apr 23 18:50:48 Signed-off-by: Roman I Khimov Apr 23 18:50:48 03Roman I Khimov  07org.openembedded.dev * rf5e68df4f2 10openembedded.git/recipes/linux-ha/cluster-glue_1.0.3.bb: Apr 23 18:50:48 cluster-glue: update to version 1.0.5 Apr 23 18:50:49 Signed-off-by: Roman I Khimov Apr 23 18:50:50 03Roman I Khimov  07org.openembedded.dev * rda70b0c0d4 10openembedded.git/recipes/pacemaker/pacemaker_1.0.8.bb: Apr 23 18:50:50 pacemaker: fix uclibc build Apr 23 18:50:50 Software stack protection is not available. Apr 23 18:50:51 Signed-off-by: Roman I Khimov Apr 23 18:50:51 03Roman I Khimov  07org.openembedded.dev * rc36d284c99 10openembedded.git/recipes/pacemaker/ (files/volatiles pacemaker_1.0.8.bb): Apr 23 18:50:52 pacemaker: fix volatiles Apr 23 18:50:52 populate-volatile.sh doesn't like blank lines in volatile files. Apr 23 18:50:57 Signed-off-by: Roman I Khimov Apr 23 18:53:46 03Klaus Kurzmann  07org.openembedded.dev * r7db2b7b8cc 10openembedded.git/recipes/pyphonelog/pyphonelog_git.bb: Apr 23 18:53:46 pyphonelog_git.bb: bump rev to get fixes for changed opimd Apr 23 18:53:46 Signed-off-by: Klaus Kurzmann Apr 23 19:01:42 03Vladimir Sorokin  07org.openembedded.dev * r241a98e177 10openembedded.git/recipes/perl/ (perl_5.10.1.bb perl_5.8.8.bb): Apr 23 19:01:42 perl: undef 'futimes' for uclibc build Apr 23 19:01:42 Function 'futimes' doesn't exist in uClibc. Apr 23 19:01:42 Signed-off-by: Vladimir Sorokin Apr 23 19:01:42 Signed-off-by: Roman I Khimov Apr 23 19:01:52 03Roman I Khimov  07org.openembedded.dev * r5ef19cb729 10openembedded.git/recipes/proxy-libintl/proxy-libintl_20080418.bb: Apr 23 19:01:52 proxy-libintl: fix build for x86_64 Apr 23 19:01:53 Signed-off-by: Roman I Khimov Apr 23 19:04:13 kergoth: I think I know whats going on Apr 23 19:04:25 kergoth: I deleted the tar file but not the .md5 file Apr 23 19:04:32 ah! Apr 23 19:04:36 kergoth: and code in __init__.py Apr 23 19:04:46 checks if not m.forcefetch(u, ud, d) and os.path.exists(ud.md5): Apr 23 19:04:57 m.forcefetch always returns null Apr 23 19:05:00 or false Apr 23 19:05:29 kergoth: shouldnt it check for the real tar also ? Apr 23 19:05:38 along with existence of .md5 Apr 23 19:05:40 would think so, yeah Apr 23 19:05:47 the thing is Apr 23 19:05:52 the md5 is the indication of its completion Apr 23 19:06:02 otherwise the tarball could be interrupted/incomplete Apr 23 19:06:10 I see Apr 23 19:06:10 but it should handle that and re-fetch if it has to.. Apr 23 19:06:16 iirc, anyway Apr 23 19:07:30 so wud you say that adding an extra check for having both files in order to consider the download complete is a fair thing Apr 23 19:10:58 yeah, I'd think so Apr 23 19:11:17 exploding horribly when the tarball vanishes is clearly not ideal :) Apr 23 19:13:42 right Apr 23 19:14:04 kergoth: ud.localfile is pointing to gcc-4.5.0.tar.bz2 is that ok ? Apr 23 19:14:22 I thought it should be null if the tar file is not there isnt it ? Apr 23 19:15:04 or does it mean the file part of url only Apr 23 19:19:32 kergoth: http://khem.pastey.net/135615 Apr 23 19:19:39 kergoth: does this patch look ok to you ? Apr 23 19:22:37 khem: seems okay Apr 23 19:23:26 kergoth: ok I will post it to bitbake-dev then Apr 23 20:21:19 Greetings. Linux question if I may. I have a default oe-dev build for the at91sam9261ek board. I run my application and it takes ~5 seconds to load (which is fine). It is responsive when switching frames within the application (<1 sec). If I leave the application at the main window for a few hours and come back, and try to switch to a different frame it takes ~7 seconds to switch for the first switch. There is no laptop mode, or hibernation that I know about Apr 23 20:22:35 probably got swapped to disk and had to be read back into ram when you came back Apr 23 20:23:07 that would be my guess too. Apr 23 20:23:16 When I do "free" it says "Swap: 0 0 0" so I was assuming swap was disabled? Apr 23 20:24:23 But I could be mistaken, I'm not too familar with linux yet. Apr 23 20:34:51 kergoth: ping Apr 23 20:35:19 need your bb expertise :) Apr 23 20:37:36 denix: boo Apr 23 20:41:19 everyone wants a piece of kergoth these days... Apr 23 20:42:03 hi, I'm trying x11-sato-image in my beagleboard however it does n't load the matchbox desktop Apr 23 20:42:28 are there any pieces left? :) Apr 23 20:42:38 oh i guess denix0 can have some of the dregs Apr 23 20:42:41 denix0: whats up? Apr 23 20:42:53 dregs? Apr 23 20:43:05 data registers (in compiler) Apr 23 20:43:22 or it could be the stuff left in the bottom of the cup when the coffee is all gone... Apr 23 20:43:29 starting manually xserver i get gconf error "GConf Error: Failed to contact configuration server;" Apr 23 20:43:55 :) Apr 23 20:44:09 denix0: http://dictionary.reference.com/browse/dregs?r=75&src=ref&ch=dic Apr 23 20:44:45 gotcha :) Apr 23 20:46:04 mwester: lol Apr 23 20:46:20 kergoth: anyway, I'm trying to modify DEPENDS from amend.inc and having issues... if I use '=', I get Python recursion exception, and if I use ':=', I get BB circular deps :( Apr 23 20:46:25 any hints? Apr 23 20:46:42 what are you setting it to? Apr 23 20:48:51 kergoth: actually, I'm trying to filter some unneded deps out with oe_filter_out() Apr 23 20:49:08 DEPENDS := "${@oe_filter_out('mysql', bb.data.getVar('DEPENDS', d, 1), d)}" Apr 23 20:49:46 the problem is when the getVar happens Apr 23 20:49:56 by the time it runs, DEPENDS is the new one, not the old one Apr 23 20:50:01 that works with other vars, but not DEPENDS Apr 23 20:50:09 not likely. Apr 23 20:50:13 its just plain wrong. Apr 23 20:51:06 ORIG_DEPENDS := "${DEPENDS}" .. then getVar ORIG_DEPENDS, or do DEPENDS := "${@oe_filter_out('mysql', '${DEPENDS}', d)}" Apr 23 20:51:13 either of those should solve it for you Apr 23 20:51:17 can you save DEPENDS in a temp Apr 23 20:51:20 yeah Apr 23 20:52:26 ah, so ${var} resolution in python code still works? when does it get resolved? Apr 23 20:53:14 it does. if you're using :=, it gets expanded immediately, though the temp var is still the safest way to go Apr 23 20:53:20 * kergoth thinks we should really be avoiding := Apr 23 20:53:42 * kergoth also thinks there's no reason we shouldn't allow DEPENDS = "${DEPENDS} foo", it should be able to know that we mean "the previous version of DEPENDS" Apr 23 20:56:17 * mwester is somewhat skeptical that "_append" is a good idea, as well. Apr 23 20:57:33 But then I'm an old fogey who still prefers that a variable's name is its name, and doesn't imply operations... Apr 23 20:57:47 yeah, I'd like to see append/prepend die in favor of a smarter, more lazy +=/=+/.=/=. Apr 23 20:57:54 we also need a different character than _ for overrides Apr 23 20:58:30 ok, DEPENDS := "${@oe_filter_out('mysql', '${DEPENDS}', d)}" didn't work - still circular deps Apr 23 20:58:37 The underscore was an unfortunate choice, I think... Apr 23 20:58:54 denix0: hmm, okay. guess you're best off with the temporary var Apr 23 20:59:06 kergoth: trying it now... Apr 23 20:59:14 mwester: yeah, very, since its used elsewhere, *and* is valid in variable names. never know whats an override and what isn't Apr 23 21:00:03 I'm glad I'm not the only one who feels that way. Apr 23 21:00:13 mwester: I agree. Apr 23 21:00:24 kergoth: btw, it' Apr 23 21:00:27 oops Apr 23 21:01:23 it's not recursive exception, it's BB dependency loops - ERROR: Dependency loop #1 found.. Apr 23 21:02:46 mwester: there are lots of things like that i'd like to see fixed in a .bb2 format or something. the fact that 'inherit' is an immediate operation as well. some shit is more declarative, some is more imperative, it's confusing Apr 23 21:03:27 kergoth: having a set of keywords which are more like CPP tokens Apr 23 21:03:36 and second pass really parsing it would claridy Apr 23 21:03:40 clarify Apr 23 21:04:33 now that there's an AST internally in bitbake it makes it a bit easier Apr 23 21:04:47 making 'inherit' actually result in loading the class files -before- the recipe was a 4 line change Apr 23 21:04:59 CPP + a parser would be natural to more coders, I think. Apr 23 21:05:08 but maybe INHERIT += would be better in the recipe too Apr 23 21:05:15 less a "command", more a declaration Apr 23 21:05:17 * kergoth shrugs Apr 23 21:06:36 Perhaps both, but a consistent syntax that would make it clear which form is which (declarative vs imperative) would be nice. Upper/lower case might suffice. Apr 23 21:06:45 yeah, true Apr 23 21:06:58 I think it'd be nice to have 'overrides' / conditional *blocks*, not just an individual var, too Apr 23 21:07:07 +1 Apr 23 21:07:16 arm [ Apr 23 21:07:16 foo=bar Apr 23 21:07:16 bar=baz Apr 23 21:07:17 ] Apr 23 21:07:19 would be lovely Apr 23 21:07:49 or better yet, TARGET_ARCH == arm [] Apr 23 21:07:59 use a limited eval() Apr 23 21:08:21 A case structure would be ideal, I think. Apr 23 21:08:56 kergoth: still no go with the temp var - ORIG_DEPENDS := ${DEPENDS} DEPENDS = "${@oe_filter_out('mysql', bb.data.getVar('ORIG_DEPENDS', d, 1), d)}" Apr 23 21:09:20 denix0: well, that does make the change to depends, so dunno what to tell you. if there's a loop, check the value of DEPENDS with bitbake -e Apr 23 21:09:25 gives me this: ERROR: Unbuildable tasks were found. These are usually caused by circular dependencies and any circular dependency chains found will be printed below. Increase the debug level to see a list of unbuildable tasks. Identifying dependency loops (this may take a short while)... Apr 23 21:09:39 * mwester has vague recollections of something similar to what denix0 is struggling with, but cannot recall the solution exactly. Apr 23 21:10:38 kergoth: and I tell you, it works perfectly with another variable - I was changing config flags the same exact way and it worked before beautifully! :) Apr 23 21:10:47 I seem to recall that you have to alter the original DEPENDS assignment to the temp var, so that the recipe results in only a single assignment to DEPENDS. Apr 23 21:10:53 hi Apr 23 21:10:57 something special about DEPENDS Apr 23 21:10:58 it's important to remember that amend.inc's are parsed during the finalize(), but after anonymous python snippets that came before it, so certain things have already happened that normally haven't during normal evaluation Apr 23 21:11:06 becuase of how its implemented Apr 23 21:11:20 mwester: let me try that... Apr 23 21:11:21 i.e. overrides Apr 23 21:11:32 my intent is to give bitbake master direct support for this sort of thing Apr 23 21:11:40 BB_EXTRA_FILES += "/path/to/amend.inc" or something Apr 23 21:11:46 let it pull it in at the -right- place Apr 23 21:11:56 kergoth: I guess amend.inc is not meant to override DEPENDS and friends Apr 23 21:12:35 well, i know there are a few things that can't be easily done with it at all due to how its implemented, though i can't recall what all they are Apr 23 21:12:43 anonymous python snippets in an amend.inc arent' evaluated at all, for example Apr 23 21:12:56 I've seen some very, very ugly problems trying to use BBCLASSEXTEND in an amend.inc too Apr 23 21:13:04 due to when things are happening Apr 23 21:13:48 kergoth: may be summarizing the limitation and intent of its use would set the expectation right Apr 23 21:14:03 well, ideally you coudl do everything there you can do in the recipe Apr 23 21:14:10 the limitations are just a function of its implementation Apr 23 21:14:12 oh Apr 23 21:14:26 then denix's problem is a bug Apr 23 21:14:38 yes, but one that likely can't be fixed without modifying bitbake Apr 23 21:15:03 (i'm assuming his problem is within this class of limitations, i'm not sure of that, though) Apr 23 21:15:20 and bitbake -e indeed shows modified DEPENDS! Apr 23 21:16:32 I was trying to build images/helloworld, but I get "opkg_install_cmd: Cannot install package xxx". Whole things here: http://nopaste.info/ae53dbd180_nl.html. Maybe someone is able to help me!? Apr 23 21:17:19 denix0: weird. Apr 23 21:19:33 kergoth: there are 2 dependency loops identified - one with do_configure, do_qa_configure, do_compile, do_install, do_populate_staging and another with the same except for missing do_qa_configure... Apr 23 21:20:25 okay, screw it Apr 23 21:20:46 :) Apr 23 21:20:59 03Chris Larson  07org.openembedded.dev * r5816ec4860 10openembedded.git/classes/base.bbclass: Apr 23 21:20:59 base.bbclass: add /lib for each dir in BBPATH to the python search path Apr 23 21:20:59 Also makes the os, time, and bb package availability explicit. Apr 23 21:20:59 Signed-off-by: Chris Larson Apr 23 21:21:03 * kergoth tries to repro the loop Apr 23 21:21:06 03Chris Larson  07org.openembedded.dev * r5c8bfd6873 10openembedded.git/classes/base.bbclass: (log message trimmed) Apr 23 21:21:06 do_unpack: shift some responsibility around, clean things up Apr 23 21:21:06 - Move the unpacking message printing into do_unpack Apr 23 21:21:06 - Move the destination directory determination into do_unpack Apr 23 21:21:07 - Use subprocess's ability to pass in PATH and cwd rather than mangling the cmd Apr 23 21:21:07 - Use shutil.copy2/copytree for ordinary file "unpack" Apr 23 21:21:08 - Use the existing urldata from bb.fetch.init rather than re-decodeurl'ing the urls Apr 23 21:21:09 03Chris Larson  07org.openembedded.dev * r5c923fd35c 10openembedded.git/lib/oe/path.py: Apr 23 21:21:10 oe.path.relative: switch to a different appraoch Apr 23 21:21:10 Signed-off-by: Chris Larson Apr 23 21:21:10 03Chris Larson  07org.openembedded.dev * r38cf80fce6 10openembedded.git/lib/oe/path.py: Apr 23 21:21:11 oe.path.join: simplify a bit Apr 23 21:21:11 Signed-off-by: Chris Larson Apr 23 21:21:12 03Chris Larson  07org.openembedded.dev * r707bbb0667 10openembedded.git/lib/oe/path.py: Apr 23 21:21:12 oe.path.relative: leverage os.path.relpath if available Apr 23 21:21:13 Signed-off-by: Chris Larson Apr 23 21:21:30 Signed-off-by: Chris Larson Apr 23 21:24:35 mwester: i've been thinking about playing with setting the eval() context to the datastore, so you could "${@TARGET_ARCH + 'foo'}". make things slightly less ugly Apr 23 21:25:49 mwester: your suggestion also didn't work... Apr 23 21:26:22 * kergoth poners Apr 23 21:26:26 * kergoth ponders, too Apr 23 21:26:36 i.e. read into OLD_DEPS, modify into NEW_DEPS and set result into DEPENDS Apr 23 21:29:19 I guess it conflicts with __depends in amend.bbclass, but I'm not that familiar with bitbake's internals :) Apr 23 21:30:04 the_maddin: what happens if you retry Apr 23 21:30:08 different meaning of depends Apr 23 21:30:15 __depends is the list of *file* dependencies Apr 23 21:30:20 this includes that, this inherits otherthing Apr 23 21:30:39 its modified by amend.bbclass to ensure that creating an amend.inc that didn't exist previously will invalidate the cache and force a reparse of the recipe & amend.inc Apr 23 21:31:51 retry: http://nopaste.info/ecff4dfffb_nl.html Apr 23 21:32:34 logfile shows the same. Cannot install package helloworld... Apr 23 21:34:09 the_maddin: can you bitbake other recipes Apr 23 21:34:27 for example? Apr 23 21:35:04 khem: console-base-image fails exactly at the same point Apr 23 21:35:30 the_maddin: strange, is something wrong with permissions on your system Apr 23 21:36:45 git downloads and whole structure is owned by the same user, everything is in his ~ Apr 23 21:38:46 umask is 022 Apr 23 21:39:42 khem: aehm... files are rw-r--r--, folders rwxr-xr-x Apr 23 21:39:49 seems to be ok Apr 23 21:41:15 khem: yeah, umask is 022 Apr 23 21:41:50 what does the error cannot install package ... mean? can I enable more verbose output? Apr 23 21:45:57 the_maddin: no idea Apr 23 21:46:39 khem: ok, thanks. btw: distro is gentoo. Apr 23 21:47:13 may building in a clean debian/ubuntu in virtualbox help? Apr 23 23:20:34 * kergoth grumbles Apr 23 23:29:27 * Tartarus crosses fingers, sets exchange out of office stuff Apr 23 23:38:28 Tartarus: traveling? Apr 23 23:38:35 nope, just taking a day off Apr 23 23:38:40 kergoth: figured it out! Apr 23 23:38:47 nice, what was it? Apr 23 23:38:52 Tartarus: it's a weekend! :) Apr 23 23:39:01 heh Apr 23 23:39:16 shit, where did i put the shlex based recipe parser again? Apr 23 23:39:46 * kergoth thinks Apr 23 23:39:58 kergoth: there was a python ${@} code in DEPENDS based on PN. when changing DEPENDS in amend.inc PN gets resolved as "amend" resulting in a wrong DEPENDS Apr 23 23:40:18 haha, doh Apr 23 23:40:47 yeah Apr 23 23:41:16 any way to amend DEPENDS variable w/o evaluating the python snippet in it? Apr 23 23:42:22 adding to it is easy, but that's not what you're trying to do, you're trying to remove something from it, which can't be done without either := or an anonymous python snippet Apr 23 23:44:40 ok, either way it's kludgy... Apr 23 23:46:22 for now I'm just removing the fallout of early eval of DEPENDS, while I'm already need to remove something from it... Apr 23 23:56:18 when i needed to do that, i just copied & pasted the DEPENDS definition into the amend.inc. ugly, but works Apr 23 23:56:30 * kergoth is trying to figure out how to do a smart signature of a datastore Apr 23 23:56:36 really complecated :| Apr 24 00:35:19 03Khem Raj  07org.openembedded.dev * r61706420f3 10openembedded.git/recipes/uclibc/ (13 files in 6 dirs): Apr 24 00:35:19 uclibc/nptl: Remove nptl recipes. Apr 24 00:35:19 * uclibc nptl has been merged into master upstream. Apr 24 00:35:19 therefore remove the recipes for nptl branch. Apr 24 00:35:19 * Use nptl as default pthreads for uclibc git recipes. Apr 24 00:35:20 Signed-off-by: Khem Raj Apr 24 00:53:16 hmmmm.. should be able to use shlex to tokenize a shell function, with a basic parser on top of it we should be able to track shell function executions of one another **** ENDING LOGGING AT Sat Apr 24 02:59:56 2010