**** BEGIN LOGGING AT Tue Jun 01 02:59:57 2010 Jun 01 06:51:54 gm everyone Jun 01 06:52:42 Hi eFfeM_work Jun 01 06:56:10 moin Jun 01 06:56:58 khem: I've seen your gcc -> svn commit, should I expect more issues fixed than 0527 snapshot I used yesterday? Jun 01 06:57:14 JaMa: thats the new way Jun 01 06:57:27 there should not be any relevant fixes Jun 01 06:57:40 JaMa: btw I are u using udev 151 ? Jun 01 06:57:47 khem: yes Jun 01 06:57:57 khem: and it didn't oom me on spitz Jun 01 06:57:59 it seems there is a gcc/udev bug that kicks in on qemu Jun 01 06:58:11 what ram size does it have Jun 01 06:58:25 64M Jun 01 06:58:31 I can have 256M on qemuarm Jun 01 06:58:34 still ooms Jun 01 06:58:43 and I have another 128M swap Jun 01 06:58:57 still its a bug Jun 01 06:59:05 124 works ok Jun 01 06:59:30 I could boot all the way Jun 01 06:59:57 now I have a strange problem with libiconv Jun 01 07:00:13 if I try to build cofigure fails Jun 01 07:00:26 but when I run run.do Jun 01 07:00:37 separately all works Jun 01 07:01:23 I guess bitbake is messing some param Jun 01 07:01:34 I'm rebuilding minimal-image with this new gcc to check it again, but haven't seen libiconf issue yesterday Jun 01 07:01:42 you wont Jun 01 07:01:49 you only run into those with uclibc Jun 01 07:01:50 only issue was missing tclsh in tcl do_install Jun 01 07:01:59 ah.. ok I'm using eglibc Jun 01 07:02:01 and actually I have huge upgrades Jun 01 07:02:05 locally to gnu tools Jun 01 07:03:22 btw with both images on both armv[45]t* I have really lot of Alignment traps (lot more than with gcc-4.4.4) Jun 01 07:03:50 but cannot say how many of them are caused by just one binary - busybox Jun 01 07:03:56 yeah thats a eglibc/loader Jun 01 07:04:02 thing I guess Jun 01 07:04:16 ah I should have coreutils in that image.. so not busybox Jun 01 07:04:17 so you will get a plenty Jun 01 07:04:30 I will fix that one day Jun 01 07:04:55 so its because eglibc-2.11 -> 2.12 upgrade? Jun 01 07:05:23 no Jun 01 07:05:33 its one issue that keeps popping Jun 01 07:05:35 because I don't see it on "stable" image with 2.12 built with gcc-4.4.4 Jun 01 07:05:38 every now and then Jun 01 07:05:44 ok, thanks Jun 01 07:05:47 dynamic linker is tricky piece Jun 01 07:09:02 morning Jun 01 07:55:46 Hello all Jun 01 07:55:57 hi Jun 01 07:56:34 first time ever I join this channel ! Jun 01 07:56:51 intressting Jun 01 07:59:16 is oe members can help for creating the ARM-based toolchain ? Jun 01 07:59:58 bitbake gcc-cross Jun 01 08:00:08 means ? Jun 01 08:00:15 otherwise read the manual Jun 01 08:01:21 can you point the link ? Jun 01 08:02:01 see web url in topic Jun 01 08:02:05 ok Jun 01 08:25:35 hi xora Jun 01 08:25:47 good morning Jun 01 08:26:57 hi florian Jun 01 08:30:14 hey woglinde florian Jun 01 08:30:21 and all the other dudes/dudesses Jun 01 08:31:28 hey XorA Jun 01 08:31:38 * florian is shocked Jun 01 08:31:48 keep away from those live wires :-D Jun 01 08:32:00 XorA, Hope you don't think we all dress like this: http://en.wikipedia.org/wiki/Dude :-) Jun 01 08:32:01 No Linux driver for the WM8983?! Jun 01 08:32:15 kick broonie Jun 01 08:32:23 * XorA bets there is Jun 01 08:32:38 XorA: heh Jun 01 08:33:00 if it hasnt changed from when I worked there, then its whatever chip is register compatible :-D Jun 01 08:33:02 I guess I will have to wait a little bit before he gets up :-) Jun 01 08:33:10 s/WMXXXX/WM8983/s Jun 01 08:33:30 florian: he normally crawls into work 10-1030 UK time Jun 01 08:33:42 thats late Jun 01 08:33:58 Jay7: ping Jun 01 08:34:06 yeah might be possibble... looking at the boards here XXXX == 8976 :-) Jun 01 08:34:26 new boards Jun 01 08:34:29 hihi Jun 01 08:34:48 * XorA thinks 8976 is what used to be in OM devices before bluetooth Jun 01 08:37:24 woglinde: mostly old ones... but having in mind the same designers both for the reference board and a custom board and the same company trying to support linux on these (without much of success) it likely they won't have changed the codec for something that is much of driver development effort Jun 01 08:38:25 hm Jun 01 08:38:27 okay Jun 01 08:41:05 MWelchUK_work__: awesome :-D Jun 01 08:41:18 hi mickeyl Jun 01 08:41:36 morning Jun 01 08:41:52 hi jay7 Jun 01 08:42:00 hey Jay7 Jun 01 08:42:11 woglinde: :) Jun 01 08:42:14 Jay7: parcel is on my desk waiting for DHL to pickup Jun 01 08:42:23 XorA: I've seen email :) Jun 01 08:42:28 good! :) Jun 01 08:42:41 * XorA is in office today which is easier Jun 01 08:43:41 spitz? Jun 01 08:49:54 anyone an idea where I could find the src of gcc 4.1 as provided in the past by code sourcery, i got some patches for it, but the sourcery gcc 4.1 seems to be 4.1.2 with some patches and changes Jun 01 08:59:15 XorA: You have an office? Wow. Jun 01 08:59:43 broonie: you know calvatec has an office :-D Jun 01 09:04:48 ~curse nvidia for lack of monitor hotplug Jun 01 09:04:50 May the fleas of a thousand camels infest your most sensitive regions, nvidia for lack of monitor hotplug ! Jun 01 09:05:52 woglinde: collie for me and tosa for lumag :) Jun 01 09:18:14 jay7 ah cool Jun 01 09:18:16 hi djwillis Jun 01 09:36:00 hi woglinde Jun 01 09:46:45 broonie: Is there an existing driver I can use for the WM8983? Jun 01 10:03:11 florian: If it's not on the web site then no (it doesn't look like a variant of an existing deviec either). Jun 01 10:04:40 broonie: I'm shocked ;-) Jun 01 10:10:43 03Koen Kooi  07org.openembedded.dev * rc0a1ed9f34 10openembedded.git/recipes/gstreamer/gst-plugins.inc: gst-plugins: dependency tweaks, enable more plugins Jun 01 10:10:44 03Koen Kooi  07org.openembedded.dev * refae564108 10openembedded.git/recipes/gnome/cheese_2.30.1.bb: cheese: add 2.30.1 Jun 01 10:10:45 03Koen Kooi  07org.openembedded.dev * r6edd8c56a7 10openembedded.git/recipes/vlc/x264_r2245.bb: x264: update snapshot and convert to new style staging Jun 01 10:10:45 03Koen Kooi  07org.openembedded.dev * r8dff535648 10openembedded.git/recipes/gstreamer/gst-plugins-bad_0.10.19.bb: gst-plugins-bad: add 0.10.19 Jun 01 10:10:46 03Koen Kooi  07org.openembedded.dev * r198cbb495a 10openembedded.git/recipes/gstreamer/gst-plugins-base_0.10.29.bb: gst-plugins-base: add 0.10.29 Jun 01 10:10:47 03Koen Kooi  07org.openembedded.dev * r8ea382d68d 10openembedded.git/recipes/gstreamer/gst-plugins-good_0.10.23.bb: gst-plugins-good: add 0.10.23 Jun 01 10:10:48 03Koen Kooi  07org.openembedded.dev * ra2834c64fb 10openembedded.git/recipes/gstreamer/gst-plugins-ugly_0.10.15.bb: gst-plugins-ugly: add 0.10.15 Jun 01 10:10:54 03Koen Kooi  07org.openembedded.dev * r3080332680 10openembedded.git/recipes/webm/libvpx_0.9.0.bb: libvpx: stage headers in ${includedir}/vpx Jun 01 10:11:00 03Koen Kooi  07org.openembedded.dev * rf6c193d1c9 10openembedded.git/recipes/gstreamer/gstreamer_0.10.29.bb: gstreamer: add 0.10.29 Jun 01 10:21:50 hehe that last long Jun 01 10:21:55 webm support Jun 01 12:36:31 03Martin Jansa  07org.openembedded.dev * r3d875160a7 10openembedded.git/ (3 files in 2 dirs): (log message trimmed) Jun 01 12:36:31 xfce46: remove checksum name atribute from bbclass Jun 01 12:36:31 * ie xfce-base/libxfce4util_4.6.1.bb cannot find it's checksums anymore Jun 01 12:36:31 * commit e3df76459f7710af61d1b51066e0685c822dbe7b introduced Jun 01 12:36:31 name=archive in xfce46.bbclass without adding archive name to all Jun 01 12:43:25 a little question: whats the most "modern" distribution in oe ? shr ? Jun 01 13:20:01 Jay7: parcel has left the building Jun 01 14:34:19 03Eric Bénard  07org.openembedded.dev * r055be8f41f 10openembedded.git/recipes/live555/ (live555.inc live555_20100409.bb live555_20100529.bb): Jun 01 14:34:19 live555: add 20100529 and create include Jun 01 14:34:19 Signed-off-by: Eric Bénard Jun 01 16:19:05 morning all Jun 01 16:31:36 XorA|gone: ok :) Jun 01 16:33:15 gm Jun 01 16:54:25 hi pwgen Jun 01 17:05:05 hi Jun 01 17:06:14 I've set PREFERRED_VERSION_tslib = "git" in my machine.conf and it keeps building the 1.0. Is it because DEFAULT_PREFERENCE = "-1" in tslib_git.bb ? Jun 01 17:09:04 ok it seems that's the problem Jun 01 17:09:16 what is the policy for DEFAULT_PREFERENCE in the recipes ? Jun 01 17:13:38 ericben: DEFAULT_PREFERENCE can be set in recipes yes if you are confident that the git version is good then set DEFAULT_PREFERENCE = "" Jun 01 17:14:25 ericben: "git" wont be the PREFERRED_VERSION it will be the PR Jun 01 17:14:45 value which would be x.y.z+gitrxxxxxxxx something like that usually Jun 01 17:15:03 unless you set PR explicitly to something known Jun 01 17:15:13 anyone ever encountered something like this: "error while loading shared libraries: /usr/lib/libsqlite.so.0: unexpected reloc type 0x03" ? Jun 01 17:15:27 khem: ok I understand better ! in fact removing DEFAULT_PREFERENCE=-1 in tslib_git.bb made it compile as a default but that was not linked with DEFAULT_VERSION Jun 01 17:15:33 rhc: which architecture Jun 01 17:15:39 thanks a lot Jun 01 17:15:57 khem: arm, davinci 6446 Jun 01 17:16:13 ericben: ok DEFAULT_PREFERENCE=-1 is the least preference order Jun 01 17:16:18 running montavista pro 5.0 Jun 01 17:16:29 rhc: ask montavista :) I dont work for them anymore Jun 01 17:16:47 Didnt you buy technical support from them Jun 01 17:17:09 yeah, i think we have "two questions" or something, but this is linking a library emitted from OE Jun 01 17:17:12 rhc: what libc are you usin Jun 01 17:17:38 2.4 I think? Jun 01 17:17:53 let me check what the package stuff says Jun 01 17:18:08 2.5.90 Jun 01 17:18:11 5.0 should have 2.6 Jun 01 17:18:14 there you go Jun 01 17:18:48 at first I thought it was a toolchain issue, but my OE env is configured to pickup the MV toolchain Jun 01 17:19:17 brb Jun 01 17:19:34 khem: PREFERRED_VERSION_tslib = "1.0+gitr0+43abb4bcac12bed4dfa7" works fine, I got lost between PR & PV thanks again I can finalize my image :-) Jun 01 17:25:10 03Andrea Adami  07org.openembedded.dev * re020cd9b2c 10openembedded.git/recipes/klibc/ (7 files): Jun 01 17:25:10 klibc: fix packaging Jun 01 17:25:10 * insmod was removed after klibc_1.5 Jun 01 17:25:10 * dmesg, sync and ls have been added with 1.5.15 Jun 01 17:25:10 * wc and losetup are provided by OpenEmbedded patches Jun 01 17:36:07 khem: there is something else strange : in the .dot I don't find any reference to tslib 1.0 but it gets installed in the image together with the git version. Jun 01 17:38:02 in fact, tslib 1.0's ipk don't even exist anymore in deploy/glibc/ipk, can the 1.0 version I find in the image be left from previous builds ? Jun 01 17:43:49 khem: sorry I was mistaken by installed-packages.txt which list several versions but on ly one gets installed ... I need to sleep ;-) Jun 01 18:02:30 ericben|away: hmmm RDEPENDS may be Jun 01 18:02:40 ericben|away: ok good night Jun 01 18:03:03 rhc: so its MV toolchain issue mostly Jun 01 18:08:25 re Jun 01 18:41:00 hey kergoth Jun 01 18:41:08 hey Jun 01 19:12:58 hi Jun 01 19:13:41 03Koen Kooi  07org.openembedded.dev * rcebadbfc30 10openembedded.git/recipes/xbmc/wavpack_4.60.1.bb: wavpack: add 4.60.1 Jun 01 19:13:42 03Koen Kooi  07org.openembedded.dev * r65c22160ee 10openembedded.git/recipes/xbmc/libmicrohttpd_0.4.6.bb: libmicrohttpd: add 0.4.6 Jun 01 19:13:49 how do we customize pre and post iinstall scripts of ipk with OE? Jun 01 19:21:51 khem: how do we customize the post and pre install scripts with OE and bb? Jun 01 19:31:27 shazkhan: pkg_postinst/pkg_preinst Jun 01 19:32:01 you can override them pkg_preinst_${PN} e.g. will do it for that package Jun 01 19:32:56 khem: thanx ... me having a look at it ... Jun 01 20:14:11 khem: just built rootfs (now without -fno-ira-spa) and it seems to work great! Jun 01 20:14:20 khem: not single segfault yet Jun 01 20:16:06 JaMa: cool nice to know Jun 01 20:16:24 it works ok for me except udev Jun 01 20:16:30 which I have workaround Jun 01 20:16:42 right now I am trying uclibc based rfs Jun 01 20:18:02 khem: last image with -fno-ira-spa had few segfault and lots of Alignment traps, now both issues seems gone Jun 01 20:18:32 JaMa: even better Jun 01 20:19:17 I rechecked twice if gcc-4.5 was really used, but was :) Jun 01 20:19:45 heh Jun 01 20:20:26 hi khem JaMa Jun 01 20:24:29 khem: any idea about renaming/refactoring that -static recipes, e.g. klibc-utils-static? Jun 01 20:26:36 ant__: pb__ had a concern to rename although logically i think the -static should be renames to dev-static because thats really what those packages are Jun 01 20:27:59 well, some are for initramfs on target... Jun 01 20:28:05 kexec Jun 01 20:28:12 hi mwester did your compiler woes end :) Jun 01 20:28:22 Nope. Jun 01 20:28:29 khem: sorry I mixed the windows Jun 01 20:28:33 :( Haven't built the feeds in a month now. Jun 01 20:28:40 hmm Jun 01 20:28:51 mwester: where are you stuck Jun 01 20:29:04 gcc do_compile task fails Jun 01 20:29:12 4.2.4 ? Jun 01 20:29:13 I have no idea at all on how to troubleshoot. Jun 01 20:29:22 4.2.4 IIRC Jun 01 20:29:48 Do you have option to upgrade gcc ? Jun 01 20:30:35 It can't find libstc++.so at some point, but I lack any understanding of where it's supposed to be, why it's looking where it's looking, how that works with sysroot, etc etc. As for upgrading, 4.4.2 causes failures in boost, mad-wifi, and rtorrent. Jun 01 20:31:12 its libtool helping you Jun 01 20:31:15 so the .la files Jun 01 20:31:35 mwester: did you try 4.4.4 Jun 01 20:32:29 if you are willing to move forward I can look into build failures for the packages Jun 01 20:45:28 khem, I'm not sure that 4.4.x will generate code that actually works :( I know that apart from the failing packages I've listed, 4.4.2 compiles the code. Since I'm travelling across the wacko-high-security border between the US and Canada, I can't easily take my devices, jtag boards, etc. to do any testing. (I tried that once, and learned that it is impossible to underestimate the intelligence of airport security) :D Jun 01 20:47:54 So I'm willing to move forward, but I have my doubts that 4.4.4 will work in both BE or LE modes as well as 4.2.2 -- but if that's the only way forward, then I guess we will be driven to 4.4.x :( Jun 01 20:48:34 I am optimistic that LE and BE will work Jun 01 20:48:50 mwester: I can also fix 4.2.4 but you know its not that cool :) Jun 01 20:49:18 mwester: plus 4.4.4 is being used by few distros in OE so its pretty ok Jun 01 20:49:40 mwester: do you have logs for failed packages Jun 01 20:49:55 mwester: I can fix 4.4.4 for BE if something is needed there Jun 01 20:50:03 I'll sync up and try again, and perhaps see if 4.4.4 will build the base image. Not sure if I'll have time to actually flash the image when I'm home this coming weekend for a few hours, though. Jun 01 20:50:44 mwester: pretty good. keep me informed Jun 01 20:51:01 I dont have any ARM BE hardware here Jun 01 20:51:15 and I doubt if qemu can do armeb Jun 01 20:57:16 JaMa: do you do any benchmarking Jun 01 20:57:39 JaMa: I would like to measure perf characteristics of 4.4 and 4.5 Jun 01 21:02:05 no, at least not yet, but I would like to see diff on armv4t (freerunner), because we were using gcc-4.4.* there and koen said it's generating too slow code (agains gcc-4.3.* used in Angstrom - which I was also using on spitz) Jun 01 21:03:26 JaMa: ok just to start with you can try to run dhrustone Jun 01 21:03:30 dhrystone Jun 01 21:04:29 khem: oops, I forgot again to bump the patch for armv4/eabi Jun 01 21:05:52 ant__: for klibc ? Jun 01 21:06:51 no, generally Jun 01 21:07:00 mom Jun 01 21:07:39 oh you mean in supported arches in sane-toolchain ? Jun 01 21:08:02 arm_eabi_supporting_arches Jun 01 21:08:04 yep Jun 01 21:09:00 Do all distros which use sane=toolchain deploy a compiler that supports EABI on armv4 ? Jun 01 21:09:17 if thats the case then we can add armv4 to supported arch list Jun 01 21:09:20 there Jun 01 21:12:05 well, Angstrom did not use it, so I think yes...let grep a bit Jun 01 21:12:26 I think the rest is >4.4.2 Jun 01 21:12:30 ok Jun 01 21:12:33 and new binutils Jun 01 21:12:34 then we can add it Jun 01 21:12:42 I tested minimal fwiw Jun 01 21:12:49 will you propose the patch Jun 01 21:12:56 just apply Jun 01 21:13:05 it, I have no armv4 here :D Jun 01 21:13:25 thx in advance Jun 01 21:13:27 woglinde has Jun 01 21:14:15 I built for collie Jun 01 21:14:56 and for the armv5te family, but bot yet with 4.5 :D Jun 01 21:15:05 *not Jun 01 21:15:13 ok Jun 01 21:15:33 4.5 is still a lot of dust to settle in but would help to test it if you could try Jun 01 21:15:57 argh again no space left on device Jun 01 21:16:14 60G is nothing for OE Jun 01 21:16:14 I was thinking there is a benchmark in opie..but opie image is currently broken... Jun 01 21:16:31 broken in what way ? Jun 01 21:16:41 dont build or dont run Jun 01 21:16:45 some missing headers made a package fail Jun 01 21:16:51 oh Jun 01 21:16:52 opie-button-settings iirc Jun 01 21:17:13 well, I looked in tinderbox and in git: no changes in the recipes... Jun 01 21:17:22 must be some OE change... Jun 01 21:17:31 could be Jun 01 21:17:55 * khem still is not comfortable browsing tinderbox Jun 01 21:18:02 some staging mismatch iirc...rebuilding soon Jun 01 21:19:34 khem: I took lisppaste out of most channels during a spam attack by the same group of persistent people that have been attacking freenode for a while now. Jun 01 21:20:08 khem: I believe the problem's solved now, but I'd like to get a channel owner / operator's OK before I add it back. Jun 01 21:21:20 khem: do you have recipe for dhrystone (maybe http://linux.maruhn.com/sec/dhrystone.html)? cannot find it in OE (except one patch for sdcc). Jun 01 21:29:11 hi, may someone explain me the functionality of PACKAGES_DYNAMIC ? Jun 01 21:29:29 JaMa: no I dont have a recipe yet Jun 01 21:29:36 ericben: dont use it Jun 01 21:30:02 khem: is it deprecated ? Jun 01 21:30:50 ericben: it can cause problems Jun 01 21:31:41 so, question : if I split qt4-embedded-fonts in several packages and want to install all these packages in one image : is there a way to list all the fonts packages in one shot (without writing all their names in the image recipe) ? Jun 01 21:33:25 oh. .that's a nice new failure: Jun 01 21:33:35 libvpx-0.9.0-r1/temp/log.do_compile.1458 Jun 01 21:33:46 ... /bin/sh: ccache: command not found ... Jun 01 21:33:53 ericben: if you're using ipkg (and probably opkg), you can do that with greedy depends. alternatively, if this is just for image construction, you can probably arrange for oe to collect a list of all the packages in question and then pass them to do_rootfs. Jun 01 21:33:59 I don't have ccache here, neither on Gentoo Jun 01 21:34:04 ant__ : I had this one this afternoon, installing ccache on the host solved it Jun 01 21:34:13 I don't want it ;) Jun 01 21:34:15 ericben: DEPENDS will be checked against all PACKAGES_DYNAMIC patterns and the corresponding recipe be selected Jun 01 21:34:36 ant__: as I removed ccache last week, I though this was a consequence of this removal Jun 01 21:34:52 I see, no, I'd say not Jun 01 21:35:06 I removed long time ago Jun 01 21:35:14 ant__: OK so there is something in this recipe (or in the sources) Jun 01 21:35:22 it seems Jun 01 21:35:41 mizar ~ # cc Jun 01 21:35:41 cc ccmake Jun 01 21:35:41 pb__: ok thanks, I'll list the packages that will be simpler ;-) Jun 01 21:35:46 just to be sure ;) Jun 01 21:36:24 ensc: OK thanks Jun 01 21:36:28 ericben: heh. yeah, that's probably true. Jun 01 21:40:00 ericben: he he http://fr.pastebin.ca/1875646 Jun 01 21:41:08 all comes from config.mk: CONFIGURE_ARGS?=--target=armv5te-linux-gcc --enable-vp8 --enable-libs --enable-ccache Jun 01 21:41:34 a.k.a. bad configure :/ Jun 01 21:46:03 ericben: it's easy patch # ccache only really works on gcc toolchains Jun 01 21:46:03 enabled gcc || soft_disable ccache Jun 01 21:46:51 but without autoconfig you cannot test for ccache...how to properly do it? Jun 01 21:50:06 ant__: why not simplt disabling ccache there ? Jun 01 21:50:09 in the configure ? Jun 01 21:50:47 sure Jun 01 21:51:21 byou see, it would be a conditional depends of the recipe Jun 01 21:51:51 ccache-native ? Jun 01 21:51:53 ...but we have no FEATURES / USE flags Jun 01 21:52:11 I mean it seems a bad idea pulling ccache... Jun 01 21:52:21 unconditionally in a recipe Jun 01 21:54:09 ant__: in the recipe there is "--enable-ccache " Jun 01 21:55:23 I didn't see this I tought this was in the code (in fact I didn't know what is libvpx until now, that's the new toy of google ;) Jun 01 21:59:07 oh..it's even in the recipe :D Jun 01 21:59:15 I went straight to the sources.. Jun 01 21:59:31 sorry, I'm looking at the libopie headers right now.. Jun 01 22:00:43 ant__: OK I'm testing this Jun 01 22:01:16 thx, please push then, this configure option is insane Jun 01 22:02:36 it built here Jun 01 22:11:13 03Eric Bénard  07org.openembedded.dev * re249502750 10openembedded.git/recipes/webm/libvpx_0.9.0.bb: Jun 01 22:11:13 libvpx: remove --enable-ccache configure option Jun 01 22:11:13 libvpx-0.9.0-r1/temp/log.do_compile.1458 Jun 01 22:11:13 ... /bin/sh: ccache: command not found ... Jun 01 22:11:13 Signed-off-by: Eric Bénard Jun 01 22:19:46 khem: don't ask me why but this evening opie compiles just fine for armv5te :p Jun 01 22:24:37 khem: it was fun when opie benchs's were showing more speed from uclibc. But there must be a trick in the roundings, in the code... ;)) Jun 01 22:26:47 fwiw benchs here (and following post explains the diff) Jun 01 22:26:54 http://www.mail-archive.com/angstrom-distro-devel@linuxtogo.org/msg02703.html Jun 01 23:33:42 ant__: interesting. Numbers Jun 01 23:34:12 this was one year ago.... Jun 01 23:34:27 ant__: can you do the same experiment with gcc 4.5 Jun 01 23:34:45 I'll try to build tonight for armv5te Jun 01 23:35:36 cool Jun 01 23:35:48 eglibc or uclibc? Jun 01 23:35:55 eglibc Jun 01 23:36:00 ok Jun 01 23:36:17 uclibc wont build yet Jun 01 23:36:17 just bumping gcc pref? nothing for binutils? Jun 01 23:36:26 yes Jun 01 23:36:31 ok, great Jun 01 23:36:36 u can use eglibc 2.12 Jun 01 23:37:46 ant__: prolly eglibc-2.12 + binutils-2.20.1 + gcc-4.5 Jun 01 23:39:49 ok, thx Jun 01 23:43:23 zecke: hello wie gehts Jun 01 23:50:26 khem: hey, boot in progress... so tired. Jun 01 23:50:59 hmm cool. are in eastern part of world still Jun 01 23:54:43 i am back from some weeks in europe Jun 01 23:55:34 how are you? Jun 02 00:00:02 zecke: I am ok. hanging on Jun 02 00:08:45 khem: again about headers. We stage mtd headers in kernel.bbclass. These were once staged by mtd-utils itself. Jun 02 00:09:12 Now if you do bitbake -c clean mtd-utils you loose the mtd & ubi headers Jun 02 00:09:20 staged by the kernel Jun 02 00:15:59 ant__: hmm we need to remove them from being packaged thru mtd-utils Jun 02 00:16:18 ant__: but the problem is that it has to be done only for certain kernels Jun 02 00:17:28 it's not a common situation: I have a recipe which DEPENDS = "mtd-utils" and I'm switching to virtual/kernel dependency Jun 02 00:17:30 it wud be nice if bitbake could use some recipe version information in the bbfiles Jun 02 00:17:51 normally one would not remove the mtd-utils probably... Jun 02 00:17:55 animist: are you sure it does not need anything else from mtd-utils Jun 02 00:18:10 no, just the two headers iirc Jun 02 00:19:42 khem: eventually one would need mtd-utils at runtime, as RRECOMMENDS or such, but not in depends Jun 02 00:19:45 nandlogical.c:30:26: error: mtd/mtd-user.h: No such file or directory Jun 02 00:21:46 ant__: which package is that which needs these headers Jun 02 00:22:51 khem: zaurus-utils/nandlogical-klibc-static finds the headers...has -I${STAGING_INCDIR} Jun 02 00:23:32 hmm Jun 02 00:23:50 all this after -c clean mtd-utils Jun 02 00:24:04 and bitbake virtual/kernel (2.6.26) Jun 02 00:25:00 he.. /oe/build/tmp/sysroots/armv5te-angstrom-linux-gnueabi/usr/include/mtd ─ is empty... Jun 02 00:25:06 I see it in do_install ok Jun 02 00:32:35 guh Jun 02 00:32:35 hm.. khem..in the ..install.cmd I see cmd_/oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/linux-libc-headers-2.6.31-r4/image/usr/include/mtd/.install Jun 02 00:32:45 i dunno if doing this for the varref/signature code is going to be worth it Jun 02 00:33:30 i'm debating grabbing calls to functions from bb/oe python package imports, searching for the .py in sys.path, reading that in as a string, compiling it to an ast, and analyzing that to see if the api functions reference any variables Jun 02 00:33:36 but i think i may be insane Jun 02 00:34:22 the difficulty there is dealing with inheritance and shit Jun 02 00:34:27 among other issues Jun 02 00:37:13 khem: ok, now zZZ.. but before I've launched the eglibc-2.12+gcc-4.5 builds.. Jun 02 00:37:19 thx and good night Jun 02 00:38:00 gn Jun 02 00:38:33 * kergoth grumbles Jun 02 00:39:12 this is proving to be one gigantic pain in my ass Jun 02 00:39:20 though parts are fun Jun 02 00:40:31 * kergoth thinks Jun 02 00:41:26 I think I'd better do the auditing next Jun 02 01:04:39 kergoth: khem asked me to add the lisppaste bot back to #oe. Is this still something you're interested in having in the channel? Jun 02 01:07:47 not familiar with lisppaste Jun 02 01:25:45 chandler: yes I am. I know it from #webkit and like it. Jun 02 01:25:55 chandler: but it is sadly not much used here. Jun 02 01:30:16 zecke: There was a spam attack a little while ago (by the same people who have been attacking freenode off and on for a while) and I pulled it from most channels while I got it sorted out. Jun 02 01:30:41 zecke: In many cases it was the first paste to that channel in a year or more, so I'm trying to re-verify what channels actually still want the bot. Jun 02 01:30:52 chandler: IIRC oe was already there on lisppaste notification list earlier Jun 02 01:31:16 It was, but I found a number of channels that had changed hands and had no idea what the bot was or why it was there. Jun 02 01:31:42 Anyway, just wanted to make sure it was OK with the channel admins. Jun 02 01:31:43 chandler: ok Jun 02 01:35:36 kergoth: So, is it OK to add it back? Jun 02 02:03:33 hi,all, after I update my oe, it can't bitbake any .bb files, the error is "/usr/bin/autoreconf: unrecognized option `--exclude=autopoint'" could someone help me pls? tks Jun 02 02:13:35 chandler: fine with me, no opinion on it really Jun 02 02:14:04 OK, thanks **** ENDING LOGGING AT Wed Jun 02 02:59:57 2010