**** BEGIN LOGGING AT Thu Jan 06 02:59:58 2011 Jan 06 05:27:23 anyone around for bitbake questions Jan 06 05:27:24 ? Jan 06 07:36:55 gm Jan 06 09:19:17 03Koen Kooi  07org.openembedded.dev * r3f3a8b2c70 10openembedded.git/recipes/ti/ (5 files): Jan 06 09:19:17 c6accel/dmai/gst-ti: only fix up linkerscript when using gcc >4.3.x Jan 06 09:19:17 Signed-off-by: Koen Kooi Jan 06 10:43:24 03Koen Kooi  07org.openembedded.dev * r75689e221c 10openembedded.git/conf/machine/ac100.conf: Jan 06 10:43:25 ac100: fix XSERVER Jan 06 10:43:25 Signed-off-by: Koen Kooi Jan 06 10:43:37 03Koen Kooi  07org.openembedded.dev * r7a69ae0f18 10openembedded.git/recipes/opencv/ (opencv-samples_svn.bb opencv_svn.bb): Jan 06 10:43:37 opencv(-samples): rename from svn to 2.1 in preparation of 2.2 Jan 06 10:43:37 Signed-off-by: Koen Kooi Jan 06 11:05:18 did someone mention ldconfig size here a few days back? (and is there something to be done about it, mine is also about 500k) Jan 06 11:46:21 hi obi Jan 06 11:56:54 03Andrea Adami  07org.openembedded.dev * rd4196731eb 10openembedded.git/recipes/klibc/klibc.inc: Jan 06 11:56:54 klibc: add missing space to SRC_URI_append Jan 06 11:56:54 * flaw exposed by recent klcc-cross commit, adding another SRC_URI += Jan 06 11:56:54 * right fix seems to be in klibc.inc Jan 06 11:56:54 * bump PR Jan 06 11:56:54 Signed-off-by: Andrea Adami Jan 06 12:15:32 hi woglinde Jan 06 12:32:35 heh finally microsoft came clean about the their worst kept secret, windows on arm :-) Jan 06 12:34:38 cant wait to see the first version of ndisloader arm Jan 06 12:35:27 xora jupp Jan 06 12:35:35 1080p on tegra2 Jan 06 12:35:39 they showed Jan 06 12:39:26 dont think that will affect OE at all, but Canonical will have to fight hard for the shared market I think Jan 06 12:41:17 yes I see this scenario too Jan 06 12:43:26 bah, I really do want to skip 2011 now, 2012 looks interesting Jan 06 12:43:45 maybe even WoW on arm :-) Jan 06 12:48:33 03Andreas Oberritter  07org.openembedded.dev * r7d83a24f8f 10openembedded.git/recipes/qt4/ (qt4-embedded.inc qt4-x11-free.inc qt4.inc): (log message trimmed) Jan 06 12:48:33 qt4: fix incorrect pkg-config files Jan 06 12:48:33 * Previously, include statements were generated like this: Jan 06 12:48:33 -I.../mipsel-oe-linux/usr/include/qtopia/QtCore Jan 06 12:48:33 -I.../mipsel-oe-linux/usr/include/qtopia/QtCore/qt4-embedded/QtCoreE Jan 06 12:48:33 -I.../mipsel-oe-linux/usr/include/qtopia/QtCore/qtopia Jan 06 12:48:34 * The sed rules ignored QT_LIBINFIX. Jan 06 13:08:08 03Koen Kooi  07org.openembedded.dev * r7fcfb42608 10openembedded.git/recipes/linux/ (53 files in 3 dirs): Jan 06 13:08:08 linux-omap 2.6.37rc: add DVFS patches from Thara Jan 06 13:08:08 Signed-off-by: Koen Kooi Jan 06 13:08:50 obi ah python-wifi 0.5 Jan 06 13:08:59 have that in my homebrach too Jan 06 13:09:03 with some patches Jan 06 13:10:05 what kind of patches? Jan 06 13:12:58 internal one Jan 06 13:13:07 for frequencies Jan 06 13:13:27 btw. where do you need it? Jan 06 13:17:37 I have to go Jan 06 13:58:56 mickeyl, ping Jan 06 13:59:31 Hello Jan 06 13:59:46 How to resolve the error of "ccache" variable ? Jan 06 14:17:49 03Michael 'Mickey' Lauer  07org.openembedded.dev * r0ad67bca7b 10openembedded.git/recipes/ttf-fonts/otf-baskervald_1.0.1.17.bb: otf-baskervald: new font recipe Jan 06 14:18:01 03Michael 'Mickey' Lauer  07org.openembedded.dev * rd120a61ed6 10openembedded.git/recipes/ttf-fonts/otf-junction_02.bb: otf-junction: new font recipe Jan 06 14:18:03 03Michael 'Mickey' Lauer  07org.openembedded.dev * r16824de8a3 10openembedded.git/recipes/ttf-fonts/otf-gillius_1.0.0.8.bb: otf-gillius: new font recipe Jan 06 14:29:01 eFfeM_work: we'll have to wait for khem -- he seems to be the guy who knows about ldconfig. His comment was that it was unnecessary, but I think that needs qualification... I think there must be some compromise by leaving it out of an image. Jan 06 14:31:03 mwester: thanks, I just peeked into it, it is statically linked (not sure if it is needed if libc is at the std place), but then again I don't think it is needed at all, on my embedded system I have no ld.so.conf or ld.so.cache still it runs fine (and I do have shared libs) Jan 06 14:31:43 as I won't be installing libs afterwards (and even if I would, they would be in /usr/lib) guess it can go Jan 06 14:31:58 .me tries Jan 06 14:33:16 i'd suspect in most cases the rpaths are sufficient Jan 06 14:33:20 It has occurred to me that if it served some very valuable purpose on small and embedded machines, busybox would have such an applet. Jan 06 14:37:40 kergoth: guess o Jan 06 14:38:01 as you noticed.. ld.so.conf and ld.so.cache are not required in any system. It's a pretty common myth people thing they're needed.. Jan 06 14:38:10 mwester: there is indeed no applet, I just removed it and the system starts fine. 560k saved Jan 06 14:38:34 As long as the _soname_ is encoded properly in the libraries, and they exist in common library paths -- or rpath (or RPATH) is used.. things will work fine w/o a ld cache.. Jan 06 14:38:48 if you need to run ldconfig for the system to function, it usually indicates a problem in sonames Jan 06 14:38:59 fray: thanks for the explanation Jan 06 14:39:06 (ldconfig has a side effect of generating symlinks for any soname mismatches) Jan 06 14:39:09 eFfeM_work: does it have its own package? Jan 06 14:39:32 mwester: don't think so, let me verify Jan 06 14:41:19 mwester: seems not, it is part of glibc, but then again I am not at git head Jan 06 14:41:46 Ok, so you just rm'd it from the image. There's actually code that does similar stuff (rm's the kernel image) for SlugOS -- but it would have been nice to exclude it in a more graceful fashion. :) Jan 06 14:42:08 grep from head: Jan 06 14:42:08 glibc-package.inc:glibcfiles = "${sysconfdir} ${libc_baselibs} ${base_sbindir}/ldconfig \ Jan 06 14:43:48 mwester: I think we will, but as we have baselined our version of OE for this product. I guess for now it'll end up in a postprocess cmd in the image Jan 06 14:44:10 but I am all in favour of putting it in a separate package Jan 06 14:44:22 I'll see if I can cook up something tonight Jan 06 14:44:43 * mwester wonders if/how minimal and micro somehow exclude ldconfig as well... Jan 06 14:45:33 mwester: good remark, found this: Jan 06 14:45:34 # Don't install ldconfig and associated gubbins Jan 06 14:45:35 USE_LDCONFIG = "0" Jan 06 14:45:43 q:q Jan 06 14:45:44 :q Jan 06 14:45:56 500K of space is huge considering the roofs is 4.5MB total -- with that gone, i can re-instate the raid drivers. :) Jan 06 14:46:02 that is for micro.conf Jan 06 14:46:06 Cool! Jan 06 14:46:28 * mwester googles "gubbins" Jan 06 14:46:38 my rfs needs to stay within 8 M compressed so the actual gain is about 280 k (my ldconfig is 560k gzip halves it) Jan 06 14:51:25 mwester: I stand corrected. USE_LDCONFIG is from classes/package.bbclass and only means that you run ldconfig during the postinst if needed Jan 06 14:52:05 btw eglibc has Jan 06 14:52:06 ./recipes/eglibc/eglibc-package.inc:libcfiles = "${libc_baselibs} ${libexecdir}/* ${@base_conditional('USE_LDCONFIG', '1', '${base_sbindir}/ldconfig', '', d)}" Jan 06 14:52:10 So you need to set that appropriately in order to ensure that packages _do not_ attempt to run ldconfig upon install. Jan 06 14:52:24 guess we should add such a line to glibc too Jan 06 14:54:35 * mwester would prefer two variables -- one to suppress the call to ldconfig in postinst, and another to suppress ldconfig itself. Jan 06 14:54:55 mwester: I am peeking at ./classes/package.bbclass actually if the USE flag is set *and* it is determined that ldconfig is needed, only in that case a postinst is generated, and this one says Jan 06 14:54:56 if [ -e /etc/ld.so.conf ] ; then Jan 06 14:54:57 [ -x /sbin/ldconfig ] && /sbin/ldconfig Jan 06 14:55:08 so if you do not have /etc/ld.so.conf it is moot anyway Jan 06 14:55:13 heh Jan 06 14:55:26 which makes ldconfig completely useless on my images... Jan 06 14:55:49 * mwester goes to see if the older releases ever had /etc/ld.so.conf present. Jan 06 14:56:38 same here, actually tried an ldconfig -p before I nuked it and it told me: Jan 06 14:56:39 ldconfig: Can't open cache file /var/run/ld.so.cache Jan 06 14:56:45 so it was not even using /etc Jan 06 14:57:29 Heh. Indeed, it seems that ldconfig has been nothing but a waste of space on even the older images released. It's never actually executed! Jan 06 14:57:57 here neither Jan 06 14:58:20 03Andreas Oberritter  07org.openembedded.dev * r87b86f347e 10openembedded.git/recipes/python/python-wifi_0.5.0.bb: Jan 06 14:58:20 python-wifi: add recipe for version 0.5.0 Jan 06 14:58:20 * package previously unpackaged files into -doc and -examples Jan 06 14:58:20 Signed-off-by: Andreas Oberritter Jan 06 14:58:20 Acked-by: Martin Jansa Jan 06 14:58:20 Acked-by: Michael 'Mickey' Lauer Jan 06 14:58:21 Acked-by: Frans Meulenbroeks Jan 06 14:58:22 03Andreas Oberritter  07org.openembedded.dev * rba7b0b6533 10openembedded.git/recipes/python/python-wifi_0.3.1.bb: Jan 06 14:58:22 python-wifi: remove version 0.3.1 Jan 06 14:58:23 Signed-off-by: Andreas Oberritter Jan 06 14:58:23 Acked-by: Martin Jansa Jan 06 14:58:24 Acked-by: Michael 'Mickey' Lauer Jan 06 14:58:25 Acked-by: Frans Meulenbroeks Jan 06 15:02:07 I have a tmpdir with quite a lot of stuff in it -- can I change DISTRO and/or MACHINE without requiring a new tmpdir? (I'm suffering from lack of disk space whilst travelling this week) Jan 06 15:02:20 machine, yes, distro, no Jan 06 15:02:31 Thanks! Jan 06 15:02:35 np Jan 06 15:03:24 * kergoth thinks ldconfig should go in its own package the way ldd does Jan 06 15:03:27 machine definitely; as a general solution you could say TMPDIR = tmp_${DISTRO} or so to solve the disto diffs Jan 06 15:03:38 kergoth: fine with me Jan 06 15:03:44 * mwester wonders what will get deleted in order to make room to test khem's changes... and eyes some customer data. :P Jan 06 15:04:20 mwester: I think I once tried to build slugos for nslu2le and nslu2be in one tree and ended up with problems though, never really investigated it though Jan 06 15:05:00 eFfeM_work: I'll add that to my list of things to look into -- it should work. rwhitby took great care to isolate the two, so I probably broke it! Jan 06 15:05:43 * mwester curses IT departments that saved $20 by selecting a 320GB drive instead of a 500GB... Jan 06 15:06:02 BTW I agree ldconfig doesn't belong in the base libc package.. Jan 06 15:07:35 mwester: when we started the testing activity initially I did not clean between them, that is how I found out. I thought to be able to save some cycles, when I had trouble I just added an rm -rf of TMPDIR to my script. slugos build does not take that long anyway (and the testing build runs in the weekend when our autobuilder is mostly idling anyway) Jan 06 15:09:56 AFAIK the official nslu2.org autobuilders do not rm -rf tmpdir -- the bulk of the time is building -native packages that should be re-used. So when I bring those back into functioning service, I'll need to fix this anyway. Jan 06 15:09:57 yeah, ldconfig probably should be packaged on its own. fwiw, there is already a switch you can use to stop it going into libc6. Jan 06 15:10:21 so, if you don't want ldconfig at all (which is probably true of most folks, actually, though they might not realise it) then the existing recipe is probably adequate Jan 06 15:10:52 # Set this to zero if you don't want ldconfig in the output package Jan 06 15:10:52 USE_LDCONFIG ?= "1" Jan 06 15:10:55 ... is the thing in question Jan 06 15:11:27 mwester: I find the bulk of the native time is spent on building gcc which is different for le and be anyway Jan 06 15:12:49 pb__: this flag tells that postinst ldconfig is needed (or not if set to 0). eglibc also uses this flag to decide on whether to package ldconfig, glibc always packages Jan 06 15:12:52 I guess I should get more precise measurements. :) Jan 06 15:13:52 eFfeM_work: really? glibc certainly didn't use to do that. if it's started ignoring that variable then that would be a bug. Jan 06 15:13:53 mwester: actually this is only from observing how a build progresses, I think for my image half of the time is spent on building the three gcc stages and on glibc Jan 06 15:15:19 pb: http://www.pastie.org/1434123 Jan 06 15:15:25 this is git head from a week ago or so Jan 06 15:16:24 hi likewise Jan 06 15:16:40 gm :-) Jan 06 15:17:03 eFfeM_work: hm, that seems a bit silly. classes/glibc-package.bbclass used to check it. Jan 06 15:17:09 glibcfiles = "${libc_baselibs} ${libexecdir}/* ${@base_conditional('USE_LDCONFIG', '1', '${base_sbindir}/ldconfig', '', d)}" Jan 06 15:17:19 ... is the line that I have in my copy of that file. I guess some joker must have removed it. Jan 06 15:18:20 pb__: that file does not exist any more in git head Jan 06 15:18:21 http://cgit.openembedded.org/cgit.cgi/openembedded/tree/classes Jan 06 15:18:54 ah, I wonder where its contents have gone Jan 06 15:20:15 I'm not good enough with git to find out when it was deleted Jan 06 15:20:43 4e6b1bfe289ad346cbd84926a43ad0f93760930d Jan 06 15:20:53 libc-package.bbclass probably? Jan 06 15:20:53 ("git log -- glibc-package.bbclass") Jan 06 15:21:08 --follow if it goes far enough back Jan 06 15:21:27 yeah, that rev is the one to blame Jan 06 15:21:28 -glibcfiles = "${libc_baselibs} ${libexecdir}/* ${@base_conditional('USE_LDCONFIG', '1', '${base_sbindir}/ldconfig', '', d)}" Jan 06 15:21:29 - Jan 06 15:21:40 +glibcfiles = "${sysconfdir} ${libc_baselibs} ${base_sbindir}/ldconfig \ Jan 06 15:21:40 + ${libexecdir}/* ${datadir}/zoneinfo" Jan 06 15:21:51 sigh Jan 06 15:22:27 it pulls in zoneinfo too? Jan 06 15:22:32 apparently Jan 06 15:22:36 ick Jan 06 15:22:54 the checkin comment says: Jan 06 15:23:00 eglibc/glibc: Use/add cross-localedef to generate locales Jan 06 15:23:00 Jan 06 15:23:00 * Add recipe for cross-localedef-native Jan 06 15:23:00 * Consolidate eglibc/glibc to share common contructs in bbclass Jan 06 15:23:00 * Move common code in eglibc to eglibc.inc Jan 06 15:23:00 * By default use cross-locaedef to generate locales now instead of qemu Jan 06 15:23:16 4e6b1bfe289ad346cbd84926a43ad0f93760930d Jan 06 15:23:29 oops wanted to paste http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=4e6b1bfe289ad346cbd84926a43ad0f93760930d Jan 06 15:24:16 the eglibc package I work with moves the binaries out of the base package and into a seperate tools package.. might be something to consider Jan 06 15:24:18 khem: can you fix ^^^ Jan 06 15:24:24 Use of USE_LDCONFIG moved from class to here: recipes/eglibc/eglibc-package.inc Jan 06 15:24:40 likewise: yes but for glibc it is now there Jan 06 15:25:11 sorry, I joined a bit too late to follow the issue Jan 06 15:25:39 I saw zoneinfo in my local build yesterday when I tried to see where my large rootfs came from Jan 06 15:26:00 eFfeM_work: what is the issue at hand? No longer builds for non-ldconfig targets? Jan 06 15:26:21 likewise: np, just found out that our products had /sbin/ldconfig but that it was not used at all Jan 06 15:26:40 so about 280k deadweight in a compressed image Jan 06 15:27:10 ldconfig is used to prune the cache, right? Jan 06 15:27:28 we do not have cache and it is not needed Jan 06 15:27:37 agreed Jan 06 15:27:42 ldconfig isn't needed in a normal system.. Jan 06 15:27:54 eFfeM_work: good find, especially if you have a slow firmware update system ;-) Jan 06 15:28:31 likewise: here is the relevant part of todays log, including frays explanation Jan 06 15:28:32 http://www.pastie.org/1434163 Jan 06 15:28:44 likewise: slow update? we ? :-) Jan 06 15:29:01 i only use web based upgrade, but I know what you mean Jan 06 15:32:33 talking about "postinst", I would like to see postinst (on the target) completely removed, for creating small firmware on the host without write access on the target (read-only firmware) Jan 06 15:33:44 likewise I am not sure if that is *always* possible, but I feel at least some postinst stuff could be done also when generating the rfs Jan 06 15:33:59 but ofc that is sometimes a little bit trickier wrt paths etc Jan 06 15:35:32 yes, agreed, most of the postinsts can/should just go away Jan 06 15:36:12 $ grep -l postinst */* |wc Jan 06 15:36:12 347 347 9898 Jan 06 15:39:55 anyway, calling it a day, back later today from home Jan 06 15:41:19 cya eFfeM_work thanks for this Jan 06 15:41:37 pb__: at least our build system should not assume the target has write access to the RFS Jan 06 15:41:53 pb__: maybe should be configurable as well (yay, another switch) Jan 06 15:41:54 yeah, indeed Jan 06 15:42:33 well, the build system doesn't assume that in any fundamental way, but it's certainly true that some of the recipes and classes do. Jan 06 15:43:12 I do use OE successfully to build images for a read-only target but that is with a very restricted set of recipes. Jan 06 15:44:38 pb__: me too (and that is what I left to eFfeM_work, he now works at my former employee) Jan 06 15:45:02 pb__: the distro's pull in too much cruft anyway, to my dislike Jan 06 15:45:24 pb__: I'ld like to go back to the point where at least we _can_ _easily_ build minimal ro firmwares Jan 06 15:45:42 I don't think micro is too bad in that respect, or at least it wasn't last time I checked. Jan 06 15:46:46 but yes, it would be great if it was easier to build small read-only images without a pile of junk getting into them. Jan 06 15:52:20 03Chase Maupin  07org.openembedded.dev * r15870da5ef 10openembedded.git/ (4 files in 4 dirs): Jan 06 15:52:20 linux-omap-psp: add OCF support Jan 06 15:52:20 * Add a patch to put OCF support into the linux-omap-psp kernel Jan 06 15:52:20 * Add support to the defconfig file for am37x-evm for OCF Jan 06 15:52:20 * Bump the MACHINE_KERNEL_PR Jan 06 15:52:20 Signed-off-by: Chase Maupin Jan 06 15:52:20 Signed-off-by: Koen Kooi Jan 06 15:52:21 03Chase Maupin  07org.openembedded.dev * r4826988d6f 10openembedded.git/classes/kernel.bbclass: (log message trimmed) Jan 06 15:52:22 kernel class: stage ocf directory Jan 06 15:52:22 * Added the crypto/ocf directory to the list of directories Jan 06 15:52:23 to be staged when building the kernel (if it is present) Jan 06 15:52:23 * This is required when building OCF modules that allow access Jan 06 15:52:24 to crypto hardware outside of the kernel tree. Jan 06 15:52:24 Signed-off-by: Chase Maupin Jan 06 15:52:25 03Chase Maupin  07org.openembedded.dev * r77cc42b7c2 10openembedded.git/recipes/ti/ti-ocf-crypto-module_1.0.bb: Jan 06 15:52:25 ti-ocf-crypto-module: add crypto module for hw accel Jan 06 16:05:52 03Steffen Sledz  07org.openembedded.dev * r5c108828be 10openembedded.git/recipes/linux/linux-2.6.24/hipox/defconfig: Jan 06 16:05:52 linux-2.6.24: enable audio support/modules for hipox machine Jan 06 16:05:52 Signed-off-by: Steffen Sledz Jan 06 16:17:58 03Koen Kooi  07org.openembedded.dev * r0d63bd4a60 10openembedded.git/recipes/ti/ti-ocf-crypto-module_1.0.bb: Jan 06 16:17:58 ti-ocf-crypto-module: remove hardcodes from makefile Jan 06 16:17:58 Signed-off-by: Koen Kooi Jan 06 16:18:02 03Koen Kooi  07org.openembedded.dev * r8eb70199bb 10openembedded.git/recipes/ntp/ntp_4.2.4p7.bb: Jan 06 16:18:02 ntp: fix crontab location, run oe-stylize while I'm here Jan 06 16:18:02 Signed-off-by: Koen Kooi Jan 06 17:00:29 03Tom Rini  07master * rbe7fe31242 10openembedded.git/classes/cpan.bbclass: Jan 06 17:00:30 cpan.bbclass: Perform more mangling for perl path Jan 06 17:00:30 On hosts where we may already be past the #! limit this is required Jan 06 17:00:30 and this is safe on the target as well. Jan 06 17:00:30 Signed-off-by: Tom Rini Jan 06 17:07:50 gm Jan 06 17:52:05 03Koen Kooi  07org.openembedded.dev * r74acf4ce5f 10openembedded.git/recipes/ntp/ (3 files in 2 dirs): Jan 06 17:52:06 ntp: update to 4.2.6p3 Jan 06 17:52:06 Signed-off-by: Koen Kooi Jan 06 18:24:13 03Frans Meulenbroeks  07org.openembedded.dev * r147981c799 10openembedded.git/recipes/linux/ (4 files in 4 dirs): Jan 06 18:24:13 linux_2.6.37.bb: added DP=1 for sheevaplug, openrd-base and openrd-client Jan 06 18:24:13 note that this is not the preferred kernel recipe for those systems yet Jan 06 18:24:13 but this way people can give it a try Jan 06 18:24:13 Signed-off-by: Frans Meulenbroeks Jan 06 19:10:06 khem, have you seen this sort of thing? http://ettus-apps.sourcerepo.com/redmine/ettus/issues/265 Jan 06 19:10:12 I think it is cosmetic Jan 06 20:07:45 Crofton|work: its a warning right Jan 06 20:07:57 hiho khem Jan 06 20:11:13 Crofton|work: we can move them into gdb's datadir Jan 06 20:11:30 and then gdb will be able to find them and ldconfig wont see them Jan 06 20:11:34 hi woglinde Jan 06 20:11:41 ~seen mwester Jan 06 20:14:51 ok Jan 06 20:15:06 not critical, but we should take care of it Jan 06 20:15:17 when it is convenient for you Jan 06 20:15:35 yeah Jan 06 20:15:58 dman users :) Jan 06 20:17:30 mwester: are you ok with my changes to slugos ? Jan 06 20:17:37 !seen mwester Jan 06 20:17:38 khem, mwester is right here! Jan 06 20:27:28 khem, I've not been able to test them due to disk space problems -- but if you have tested and the rootfs works, go ahead and commit. Jan 06 20:33:13 * mwester ducks out for YASM (Yet Another, um, Stupendous Meeting) Jan 06 20:36:48 ah assembly of people == meeting....now it's clear Jan 06 21:23:10 calling it a day, cya Jan 06 21:27:19 03Tom Rini  07master * r36636be6ef 10openembedded.git/recipes/autoconf/autoconf.inc: Jan 06 21:27:19 autoconf: Mangle to use /usr/bin/env perl Jan 06 21:27:19 On the host side we can have the full path exceed the limit for the #! Jan 06 21:27:19 method. On the target side, we already had to play a game to ensure Jan 06 21:27:19 the path would be valid. So we change to using /usr/bin/env there Jan 06 21:27:19 as well, which is safe. Jan 06 21:27:19 Signed-off-by: Tom Rini Jan 06 22:24:54 03Richard Purdie  07master * rcf37932ad6 10bitbake.git/lib/bb/build.py: Jan 06 22:24:54 bitbake/build.py: Ditch the extract_stamp function as its just going to complicate improving the stamp handling functions Jan 06 22:24:54 (From Poky rev: 51a2e8712a9b94de53253e198c64c0fa428e4349) Jan 06 22:24:54 Signed-off-by: Richard Purdie Jan 06 22:24:56 03Richard Purdie  07master * rbee33a9623 10bitbake.git/lib/bb/build.py: Jan 06 22:24:56 bitbake build.py: Stamp handling improvements Jan 06 22:24:56 * Move stamp file deletion out of the internal stamp helper function Jan 06 22:24:56 * Add a new function to return the path to a stamp for a given task Jan 06 22:24:56 (From Poky rev: ec6e4793dc79c61b780b74571db4cd66c1c04251) Jan 06 22:24:56 Signed-off-by: Richard Purdie Jan 06 22:24:57 03Richard Purdie  07master * r94709a0904 10bitbake.git/lib/bb/build.py: Jan 06 22:25:52 bitbake build.py: Use localdata for stamp handling, not d Jan 06 22:25:52 (From Poky rev: 4a10a5b438f49577fb29f4e277e2300cdce10ab8) Jan 06 22:25:52 Signed-off-by: Richard Purdie Jan 06 22:27:16 Tartarus: ping Jan 06 22:27:24 pong Jan 06 22:27:31 hello there Jan 06 22:27:34 hey Jan 06 22:27:37 klcc patch? :) Jan 06 22:27:52 I'm pleased to see you're doing more perl-digging ;) Jan 06 22:27:58 heh Jan 06 22:28:05 Trying to turn the autobuilder we have to pass Jan 06 22:28:33 building in say /var/hudson/workspace/sb_openembedded_autobuilder/BBPKGS/x11-image/DISTRO/minimal/MACHINE/qemux86/label/sb-rhel5/ exposes problems most folks don't see :( Jan 06 22:29:06 back to klcc (not cross), this should reside on target, installed by klibc-dev (we'll rename it debian-way later on) Jan 06 22:29:24 now, it reads from sysroot Jan 06 22:29:47 my question is: at which point would you mangle it? Jan 06 22:30:14 Well, /usr/bin/env perl is always safe Jan 06 22:30:48 there are a few places where /usr/bin/env perl can fail.. but it's far less used then the long path failures we've all seen... Jan 06 22:30:48 there are 3 or 4 patches touching klcc but about prefix and install, not from perl pov Jan 06 22:30:59 I highly recomment _always_ using /usr/bin/env perl... Jan 06 22:32:45 Well, /usr/bin/env perl -w doesn't work, yeah Jan 06 22:32:47 but what's left? Jan 06 22:33:28 see, my /bin/klcc starts with "#!/oe/build/tmp/sysroots/i686-linux/usr/bin/perl" Jan 06 22:33:42 so the patch for klcc-cross could apply here, isn't? Jan 06 22:33:54 yeah, or what I just did to automake could also cover it Jan 06 22:34:43 (I'm not a perl expert... but) the one case I was thinking of is there is a way for perl to re-exec a perl script and it just gets reloaded.. we found an instance where /usr/bin/env perl confused this and it didn't work.. I'm not even sure thats a problem anymore Jan 06 22:34:44 but this is not a klcc cross, still I find "$CROSS = "arm\-oe\-linux\-gnueabi\-";" Jan 06 22:35:12 and "$CC = "\/oe\/build\/tmp\/sysroots\/i686\-linux\/usr\/armv5te\/bin\/arm\-oe\-linux\-gnueabi\-gcc"; Jan 06 22:35:12 " Jan 06 22:35:52 ant__, yeah, sounds like klcc needs some general mangling to fixup that kind of stuff, like we used to do on automake Jan 06 22:36:01 * ant__ should not have revived /bin/klcc...was unpackaged, in oblivion.... Jan 06 22:36:52 I can hardly imagine it but somebody could compile natively on target ... Jan 06 22:37:44 Yes, there are compile on target advocates, heh Jan 06 22:38:32 btw, see how debian splòit the bits http://packages.debian.org/search?searchon=sourcenames&keywords=klibc Jan 06 22:39:08 iirc khem suggested to adopt that naming Jan 06 22:57:54 master is giving me this: ERROR: Error parsing /home/leon/sandbox/navarra/screen/openembedded/openembedded.git/recipes/images/x11-pimlico-image.bb: Could not inherit file classes/rootfs_${IMAGE_PKGTYPE}.bbclass Jan 06 22:58:22 anyone else seeing this? Jan 06 22:59:24 ^^ false alarm ^^, distro not set correctly (I would assume this would go detected BTW). Jan 06 23:00:21 gn all Jan 06 23:50:52 03Khem Raj  07master * r3f405f3578 10openembedded.git/ (4 files in 3 dirs): Jan 06 23:50:52 distro/slugos: Upgrade to use gcc 4.5 and libtool 2.4 Jan 06 23:50:52 * Depend on sane-toolchain Jan 06 23:50:52 * Upgrade few old pinned versions Jan 06 23:50:52 * Inherit own-mirrors to define slug src mirror Jan 06 23:50:53 * nslu2-mirrors.bbclass is not needed anymore Jan 06 23:50:53 Signed-off-by: Khem Raj Jan 07 02:21:55 huh, new release of pychecker. nice. now there's pep8, pyflakes, pychecker, and pylint that work now **** ENDING LOGGING AT Fri Jan 07 02:59:58 2011