**** BEGIN LOGGING AT Tue Feb 23 02:59:58 2010 Feb 23 04:08:10 Where is the appropriate place to report bugs in one of the distros (specifically angstrom) I asked on the mailing list about a few things, and the lists seem dead) Feb 23 05:53:33 james_l: unfortunately, that is the state of things. There currently is no serious enduser-friendly distro out there. You could try #angstrom. This channel is not the right place, I'm afraid. Feb 23 05:58:24 Well, I filed a few bugs on bugs.openembedded.org, and they got marked NOTOURPROBLEM. So it's very frustrating. (Also, I appear to have found a problem with wireless encryption for potentially many cards for kernels above 2.6.29, due to ieee80211 beign renamed to lib80211) but don't know where I should report that. I tried posting it to angstrom-devel, and haven't heard back. (The fix I have breaks 2.6.28 and below, so I was hoping to a) find Feb 23 05:58:25 out the proper way to handle that, and b) get it patched into openembedded, as so far few devices seem to use 2.6.29 or above, even in -dev) Feb 23 06:15:04 james_l: The only thing I can offer is http://launchpad.net/sonkei, but that is a project that does not even exist yet Feb 23 06:15:34 I have moved some on-device bugs, there, because as you say there is no other place to collect and discuss them. Feb 23 06:15:50 The situation before wasn't any better, though Feb 23 06:16:13 we'd accept distro bugs in the bug tracker, but nobody ever bothered looking at them Feb 23 06:17:07 So how does one get patches submitted to OE? Feb 23 06:17:12 that's arguably worse than the current policy where you have nowhere to report such issues. But at least that correctly represents their chances of actually being addressed. Feb 23 06:18:33 prepare the patch, format it with git (forgot the exact command, I think it's git-format-patch HEAD^^^^ or something like that) and sent the patches to the ml Feb 23 06:18:46 but even there, we're lagging behind quite a bit Feb 23 06:18:55 * Laibsch doing some expectation management Feb 23 06:19:18 I'm off Feb 23 06:19:19 bye Feb 23 06:19:22 Thanks Feb 23 06:19:35 you're welcome Feb 23 06:19:42 sorry I don't have a more exciting answer Feb 23 07:36:38 good morning Feb 23 07:40:35 hi all Feb 23 09:13:42 morning Feb 23 09:13:57 morning Feb 23 09:13:57 goodmorning :) Feb 23 09:21:27 can someone tell me what "=+" in some recipe/classes means ? (couldn't find any definition) Feb 23 09:21:56 morning Feb 23 09:22:24 morning ;) Feb 23 09:22:25 mlip2: prepend Feb 23 09:22:32 += is append Feb 23 09:22:44 but this is not general python, is it ? Feb 23 09:23:04 I thoughgt it was, but I hate python with a passion so wouldnt really know Feb 23 09:23:37 couldn't find it in the docs and somehow it generated (also += is) problems ;) Feb 23 09:24:05 meh Feb 23 09:24:30 I was sent here by tsjsieb, after a question in #angstrom Feb 23 09:24:36 hello Feb 23 09:24:41 He believed it was more OE related Feb 23 09:24:48 go for it then Feb 23 09:25:02 while we are still in that early morning happy period :-D Feb 23 09:25:33 :D and you can always blame me, if they're starting to become angry :P Feb 23 09:25:51 I'm trying to build angstrom according to the manual on http://www.angstrom-distribution.org/building-angstrom Feb 23 09:25:53 * XorA terminates tsjsieb just in case Feb 23 09:26:18 On a Ubuntu 9.10 quadcore box, SMP enabled in the local.conf file Feb 23 09:26:23 but it fails Feb 23 09:27:02 With a note (I'm looking it up, I have only 100 tabs opened in Firefox) Feb 23 09:27:44 * XorA had forgotten to join #angstrom for last couple of weeks Feb 23 09:28:11 09:34:04 < Romke> NOTE: Couldn't find shared library provider for libgcc_s.so.1 Feb 23 09:28:12 09:34:32 < Romke> But when I look for that library it is installed Feb 23 09:28:12 09:35:28 < Romke> Now, am I just doing something very wrong? Feb 23 09:28:26 It is installed on my system Feb 23 09:28:40 that is normal to see once I beleive Feb 23 09:28:48 09:53:42 < Romke> I've found this page, which exactly states my error message, but I don't quite understand which library it means Feb 23 09:28:52 09:54:45 < Romke> The cross compiled version? Or the version on my system? Feb 23 09:29:14 its only a NOTE so not an error unless it breaks something Feb 23 09:29:17 http://docs.openembedded.org/usermanual/html/commonuse_prebuilt_toolchain.html << this is the page which states that message exactly Feb 23 09:29:36 It's one of the last messages Feb 23 09:30:17 good morning Feb 23 09:30:37 While googling I use some other parts of another message (dunno if it was a note or error) which said things about: LEAD_SONAME 'libc.so' Feb 23 09:31:10 Romke, Can you past the complete log (last few 100 lines or something) (in pastebin). Feb 23 09:31:24 erhm.. Feb 23 09:32:00 I would have to restart the build Feb 23 09:32:24 Because I started looking for those files, browsed around in the dirs Feb 23 09:32:46 And screen doesn't have a large enough scrollback buffer for that :+ Feb 23 09:33:17 also, my build box has been rebooted Feb 23 09:36:43 well, the NOTE your provided doesn't necessarily indicate an error. Unless you know which package failed (then you could look up the log file), I think you have to rebuild and paste the lines. (tsjsieb,XorA correct me if I'm wrong) Feb 23 09:39:31 I agree, and (if I'm not wrong) it doesn't have to cost very much time, if you didn't remove your build/tmp directory Feb 23 09:40:28 I did.... Feb 23 09:40:39 But I have a quadcore :P Feb 23 09:41:08 At least I did not remove my downloads dir Feb 23 09:41:44 mlip2: your right, I said pretty much the same Feb 23 09:41:45 which saves download time and angry coworkers :P Feb 23 09:42:12 I always read that as cow-orkers for some reason Feb 23 09:42:53 hehe Feb 23 09:43:35 Time to check if the coffee machine is still working :) Feb 23 09:48:22 Yes, it does :) Feb 23 09:55:12 It's already working on task 1329 Feb 23 09:55:20 It fails around task 1427 Feb 23 09:55:44 which is after generating 23094829034829035720934809238490234 locales Feb 23 09:56:15 Which is something I would like to alter when I better understand OE and angstrom Feb 23 09:58:56 Is there a way to get a new password from the oe bugtracker? i forgot mine... Feb 23 09:59:11 oh, nvm Feb 23 10:04:50 Romke: past the error, by a pastebin. Feb 23 10:06:01 yeah, I'm still waiting :P Feb 23 10:06:10 It's still doing things Feb 23 10:06:17 hasn't failed yet.. Feb 23 10:06:34 It hasn't started generating locales yet Feb 23 10:07:47 * Romke whips the dell quadcore box, 'faster!' Feb 23 10:08:00 we must get the cross localgen working :-) Feb 23 10:08:01 http://marcin.juszkiewicz.com.pl/2010/02/23/gource-openembedded/ Feb 23 10:08:33 cross localgen? Feb 23 10:09:16 Romke: you might want to add this to your local.conf GLIBC_GENERATE_LOCALES = "en_GB.UTF-8 nl_NL.UTF-8" As the locales might be not very importand for you.. (please correct me if I'm wrong) Feb 23 10:09:42 It would have been nice if I coudl tell OE that I'm mainly interested in en_US and maybe en_GB and nl_NL Feb 23 10:09:45 oh Feb 23 10:09:48 well Feb 23 10:09:51 thanks for the tip :) Feb 23 10:11:14 Is it safe to interrupt bitbake using ^C ? Feb 23 10:11:32 yes Feb 23 10:11:36 ok Feb 23 10:13:37 Romke: you can Feb 23 10:13:48 Romke: tell it to limit locale gneration Feb 23 10:17:58 http://pastebin.org/95791 Feb 23 10:19:05 oh, wait Feb 23 10:19:09 that's not everything Feb 23 10:19:09 nothing wrong so far Feb 23 10:19:18 Romke: interesting looks like is blew up evaluating ${PYTHON_DIR} Feb 23 10:21:51 http://pastebin.org/95796 Feb 23 10:22:15 now let's check the log :+ Feb 23 10:25:19 yeah, off course... Feb 23 10:25:25 it failed because of 'interrupt' Feb 23 10:25:43 qemu: uncaught target signal 2 (Interrupt) - exiting Feb 23 10:25:45 which makes sense, since I did interrupt it... Feb 23 10:26:49 Then the question is, what is wrong with that... Feb 23 10:27:36 qemu 0.11 and host kernel 2.6.33-rc? Feb 23 10:28:31 uname -a: Linux quadbuild-ubuntu 2.6.31-19-generic-pae #56-Ubuntu SMP Thu Jan 28 02:29:51 UTC 2010 i686 GNU/Linux Feb 23 10:28:54 And I didn't install qemu manually Feb 23 10:30:29 cdromke@quadbuild-ubuntu:~/OE/angstrom-dev$ staging/i686-linux/usr/bin/qemu-arm Feb 23 10:30:29 qemu-arm version 0.10.3, Copyright (c) 2003-2008 Fabrice Bellard Feb 23 10:31:30 Hm, the logfile gives me this: http://pastebin.org/95800 Feb 23 10:31:42 it's a grep for 'error' Feb 23 10:32:42 odd Feb 23 10:35:22 hi cbrake Feb 23 10:38:26 hi , can java app run on directfb ? swing and awt ? Feb 23 10:41:06 03Graeme Gregory  07org.openembedded.dev * rf13a3b6d01 10openembedded.git/recipes/networkmanager/networkmanager_0.8.bb: networkmanager_0.8.bb : add new version Feb 23 10:41:07 03Graeme Gregory  07org.openembedded.dev * r930370692f 10openembedded.git/conf/checksums.ini: checksums.ini : new NM components checksums Feb 23 10:41:08 03Graeme Gregory  07org.openembedded.dev * r29ba1723bc 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom-2008-preferred-versions.inc : network manager updates Feb 23 10:41:09 03Graeme Gregory  07org.openembedded.dev * r9915d82f6f 10openembedded.git/recipes/networkmanager/network-manager-applet_0.8.bb: network-manager-applet_0.8.bb : add new version Feb 23 10:41:17 hmm.. So, is it a bug or is it just me being stupid? Feb 23 10:41:20 03Graeme Gregory  07org.openembedded.dev * r141074d901 10openembedded.git/recipes/networkmanager/networkmanager-openvpn_0.8.bb: networkmanager-openvpn_0.8.bb : add new version Feb 23 10:41:45 Romke: can you confirm gcc failed to compile? Feb 23 10:41:54 http://pastebin.org/95810 Feb 23 10:42:21 I'll try to put the full log somewhere Feb 23 10:42:29 thats a bug Feb 23 10:42:43 fsck knows what caused that though Feb 23 10:43:25 http://cdromke.users.cdromke.nl/log.do_compile.22504 Feb 23 10:43:31 here it is Feb 23 10:48:50 mkay Feb 23 10:48:59 let's tell the boss the erm... good news then Feb 23 10:49:09 Romke: this stable or .dev? Feb 23 10:49:27 dunno tbh Feb 23 10:49:37 anybody knows who brought the NGW100 at Fosdem? Feb 23 10:50:01 mckoan|training: nope, needing expertise or lost one? Feb 23 10:50:01 http://www.angstrom-distribution.org/building-angstrom << the manual speaks about 'stable' Feb 23 10:50:16 so i guess it's stable Feb 23 10:51:08 unfortuneately I dont have time to support stable :-( Feb 23 10:52:36 mckoan|training: Ulf iirc Feb 23 10:53:20 Romke: wait a minute, I think this is the gcc 4.4.X that is in ubuntu cant compile 4.2.X which OE uses Feb 23 10:53:34 Romke: I remeber someone else having this issue, you need to install older gcc in ubuntu Feb 23 10:53:36 hrw: thx Feb 23 10:53:40 Romke: which should be packages Feb 23 10:56:06 hi all , how can i do static compile on my projects that they wont require glibc on my OS ? Feb 23 10:57:46 hi all , how can i do static compile with bitbake , on my projects that they wont require glibc on my OS ? Feb 23 10:57:52 glibc doesn't really support being linked statically. Feb 23 10:58:14 I guess use uclibc then Feb 23 10:58:34 (upstream that is, rather than just in OE.) Feb 23 10:58:53 * XorA thinks its unlikely that question was the right question though Feb 23 10:59:07 i got problem when i compile codes , glibc version is not compatible with my bitbake do ? Feb 23 10:59:17 my OS using old glibc Feb 23 10:59:34 and my project i do compile uses more future glibc Feb 23 10:59:39 how can i fix this problem Feb 23 10:59:42 someone have idea? Feb 23 11:00:42 pull source of your old glibc into OE Feb 23 11:01:29 XorA , work directory ? Feb 23 11:01:58 JDuke128: make a recipe for it, use it Feb 23 11:02:11 JDuke128: chinook-compat distro does that for old maemo stuff Feb 23 11:05:26 XorA , btw java swing apps works on directfb ? Feb 23 11:05:45 JDuke128: do I look like a java weirdo :-D Feb 23 11:06:01 JDuke128: I have no idea sorry! Feb 23 11:06:33 just asked i thought its embedded question Feb 23 11:07:04 JDuke128: check out the jalimo project Feb 23 11:07:23 XorA: tried newer transmission? Feb 23 11:07:34 JDuke128: we rather use Java on top of X11 Feb 23 11:07:39 hrw: not yet, bug its changelog doesnt list any interesting fixes Feb 23 11:07:52 hrw: Im on 1.83 which works well Feb 23 11:08:00 XorA: I am planning to abandon angstrom on my sheeva Feb 23 11:08:03 mkay, I'll try XorA Feb 23 11:08:22 hrw: moving to fedora? Feb 23 11:08:28 XorA: debian Feb 23 11:08:32 sorry for the late response, I was preparing food Feb 23 11:08:37 XorA: why would have to go for fedora? Feb 23 11:08:40 Romke: where mine then :-D Feb 23 11:09:01 hrw: I guessed you meant debian, was just having a laugh Feb 23 11:10:17 XorA: I have encoding problems with samba shares, not working nfs server and few other strange things which I do not have a time to check Feb 23 11:11:15 hrw: odd, samba working well for me, but then I do speak to it in its native language :-) Feb 23 11:12:13 I use lot of !ascii characters Feb 23 11:17:27 debian installer on sheevaplug over serial is nice Feb 23 11:18:42 :) Feb 23 11:19:13 03Graeme Gregory  07org.openembedded.dev * r2f0e768e7e 10openembedded.git/recipes/transmission/transmission_1.91.bb: transmission_1.91.bb : add new version, few minor bug fixes Feb 23 11:19:35 too bad that 2.6.30.2 does not initialized usb ;( Feb 23 11:19:41 need to find SD card for rootfs Feb 23 11:20:36 * XorA uses ubifs root with USB data Feb 23 11:22:04 * Romke wants a dual-quadcore Feb 23 11:22:21 go faster, dammit Feb 23 11:22:30 * XorA has two dual quads here Feb 23 11:22:39 but this is a legitimate excuse for slacking off :) Feb 23 11:23:21 anyone in for a swordfight? (http://xkcd.com/303/) Feb 23 11:24:02 * XorA looks for his Lochaber axe Feb 23 11:25:02 * Romke would like this stuff to work that well that it could happen during the night Feb 23 11:35:03 XorA: I use such combo too, but Debian do not have ubifs root option Feb 23 11:35:10 bah Feb 23 11:35:50 ~hail sim.one for having 4GB sd Feb 23 11:35:51 * ibot bows down to sim.one for having 4GB sd and chants, "I'M NOT WORTHY!!" Feb 23 11:36:15 * XorA has some 16G SD for omap3 rootfs Feb 23 11:37:03 XorA: how much used? Feb 23 11:37:11 hrw: about 350M Feb 23 11:37:15 250M Feb 23 11:37:20 ;D Feb 23 11:43:08 heh Feb 23 11:46:27 hrw: Do we have support for the sim one in OE already? Feb 23 11:47:43 yes Feb 23 11:47:46 MACHINE=simone Feb 23 11:47:52 very nice! Feb 23 11:48:01 and thx for remind - I have kernel to check Feb 23 11:48:17 hrw: yw :) Feb 23 11:48:32 as ts72xx guys ported some of ep93xx patches to .32 Feb 23 11:48:46 florian: you got #00024 board? Feb 23 11:49:01 I should give it a try but currently time is very limited since Embedded World is coming! Feb 23 11:49:15 hrw: Nope I have #31 I think. Feb 23 11:49:55 goddammit Feb 23 11:49:56 I think I'll make this the platform for playing with really small images. Feb 23 11:50:01 now the kernel fails to compile :/ Feb 23 11:50:33 florian: move out of mmc as it is very slow (~250KB/s) due to being mmc-spi Feb 23 11:50:40 florian: usb also has some problems Feb 23 11:52:41 hrw: That's the idea... we need to add some more useful images to OE anway. And if I didn't miss anything we lack really small but functional ones. Feb 23 11:53:04 florian: just-busybox-image for example Feb 23 11:53:43 serial+console access only, nothing except busybox Feb 23 11:53:53 I thought that was micro-image :-) Feb 23 11:54:01 hrw: usb is a problem on many platforms. Look at he beagleboard... or currently I work with a TX51 - guess what fails ;) Feb 23 11:54:31 tx51? Feb 23 11:54:34 imx51? Feb 23 11:54:47 hrw: yes Feb 23 11:55:16 interesting Feb 23 11:55:32 fast cpu, 512MB ram etc Feb 23 11:56:01 ah. tx51 has only 128MB Feb 23 11:57:03 Yes that's a quite interesting piece of hardware. But RAM is a little bit limited... Feb 23 11:57:44 Its really fast but compared to an OMAP3 the i.MX is quite large and gets hot. Feb 23 11:58:34 There ia some Qseven board with i.MX51 and more RAM... Feb 23 11:58:52 03Koen Kooi  07org.openembedded.dev * r86975b82c6 10openembedded.git/recipes/gnome/gnome-system-tools_2.28.2.bb: gnome-system-tools: add 2.28.2 Feb 23 11:58:58 03Koen Kooi  07org.openembedded.dev * rb0c8e5a6f9 10openembedded.git/recipes/cpufreqd/ (cpufrequtils/fix-proc-stat-reading.patch cpufrequtils_006.bb): Feb 23 11:58:58 cpufrequtils 006: add patch from upstream to fix /proc/stat reading Feb 23 11:58:58 * This brings 006 up to 007 functionally, without all the buildsystem changes in between Feb 23 12:08:32 ~hail non-mainline devices... Feb 23 12:08:33 * ibot bows down to non-mainline devices... and chants, "I'M NOT WORTHY!!" Feb 23 12:08:38 ep93xx has 2 SPI drivers Feb 23 12:08:46 * hrw -> food Feb 23 12:39:18 IT COMPILES! Feb 23 12:39:22 SHIP IT! Feb 23 12:39:22 :D Feb 23 12:40:20 ~hail #oe people Feb 23 12:40:21 * ibot bows down to #oe people and chants, "I'M NOT WORTHY!!" Feb 23 12:42:16 * Romke kicks tsjsieb's ass Feb 23 13:08:31 git.openembedded.net needs more bandwidth :/ Feb 23 13:09:55 broonie: donate some? :D Feb 23 13:10:59 Oh, perhaps it's I/O constrained not bandwidth - bounced up now. Feb 23 13:11:32 We were dontaing a MTN server previously, but obviously no longer. Feb 23 13:11:42 git is running off the shared machine with slugos isnt it? was a monster machine from memory Feb 23 13:12:04 Wolfson mtn mirror saved the project a few times Feb 23 13:12:17 I'm getting download speeds bouncing between 15-100k . Feb 23 13:12:26 I used wolfson's one mostly Feb 23 13:12:46 although I deny anything to do with it Feb 23 13:18:07 Meh, yeah. This looks like I/O speed - fetching into an existing clone is *much* faster. Feb 23 13:18:25 I wonder if the repo needs repacking? Feb 23 13:21:00 I thought zecke had added a cron to do that, best ask him Feb 23 13:24:42 * broonie has been stuck on 19% clone for ~15 mins now :( Feb 23 13:25:22 sounds like monotone speeds :-) Feb 23 13:33:15 * XorA really wishes gnome guys would stop building python into everything Feb 23 13:34:25 there are a couple git-pack-object processes on the server using close to 200MB RES ... Feb 23 13:34:50 cbrake: But how much memory does it have? Feb 23 13:35:19 broonie: 1.5GB Feb 23 13:35:49 so it should be able to tolerate the repacks (though I/O might be an issue) Feb 23 13:37:30 If I remember correctly we mostly had I/O performace issues with the ols server at LinuxToGo. Feb 23 13:37:35 shit. I need to upgrade sheeva uboot Feb 23 13:37:53 hrw: last time I did that it was real easy :-) Feb 23 13:38:37 hrw: the alpha-6 sheeva installer works pretty well for me Feb 23 13:39:41 I already did that in past so know how to do it ;D Feb 23 13:39:47 just do not like uboot updates Feb 23 13:40:05 could be worse, could be gta01 one :-) Feb 23 13:40:15 yep Feb 23 13:40:37 hrw: what distro are you running on your plug? Feb 23 13:40:47 cbrake: moving to Debian Feb 23 13:40:50 from angstrom Feb 23 13:41:16 hrw: I've been using Ubuntu 9.04. Tried debian yesterday, but had all kinds of trouble with the network interface going up and down. Feb 23 13:41:33 ubuntu is dead end for sheeva Feb 23 13:41:41 9.04 was last release for armv5te Feb 23 13:41:43 hrw: yes, agreed Feb 23 13:42:13 hrw: you can shorten that to just ubuntu is dead end :-) Feb 23 13:42:17 9.10 was armv6, 10.04 will be armv7a Feb 23 13:43:39 heh... very clever Feb 23 13:43:49 * florian hugs OE Feb 23 13:50:04 cbrake: you use mainline uboot or marvell uboot? Feb 23 13:50:41 hrw: likely marvell -- whatever comes in their sheevainstaller Feb 23 13:50:44 can someone point me to the procedure of sending patches for bbclass files to the mailing list? git format-patch and then append it to the email or just c&p the patch into the email? Feb 23 13:51:33 mr_nice: use git send-email Feb 23 13:52:20 broonie: ok, thx Feb 23 13:52:32 cbrake: thats marvell one. start take ages Feb 23 13:55:05 anyway it has kind of working SD Feb 23 14:01:01 Debian GNU/Linux squeeze/sid sheevaplug ttyS0 Feb 23 14:04:28 hrw: you using the debian installer, or the deboostrap, or ...? Feb 23 14:07:01 cbrake: debian installer Feb 23 14:07:08 cbrake: worked nice Feb 23 14:08:20 removing not needed stuff now Feb 23 14:09:06 aptitude purge libc6 :-) Feb 23 14:11:50 * cbrake sets up an ext4 build partition ... Feb 23 14:12:08 XorA: ;D Feb 23 14:12:34 cbrake: hmm.. I should finally convert my /home/ to ext4 too Feb 23 14:13:36 brave hearts :) Feb 23 14:14:09 ynezz: hrw used to run reiser, we know he is brave :-D Feb 23 14:14:16 :D Feb 23 14:14:48 hrw: my home and project partitions are still ext3/sw raid-1, but I'm trying to make my scratch builds a little more efficient Feb 23 14:15:00 I will say one thing, reiser was really really fast, its surprising what a good optimisation not writing data to disk was Feb 23 14:15:18 heh Feb 23 14:16:29 XorA: I also used xfs Feb 23 14:16:38 and still think that reiserfs was better Feb 23 14:16:46 thats just scary Feb 23 14:18:22 * mwester has wondered about upgrading his RAID-5 to ext4 Feb 23 14:18:39 03Martin Jansa  07org.openembedded.dev * r206c704fa0 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Feb 23 14:18:39 EFL: bump SRCREV a bit, latest illume2 fixes and big python-efl cleanup Feb 23 14:18:39 Signed-off-by: Martin Jansa Feb 23 14:18:40 03Martin Jansa  07org.openembedded.dev * rcd2745db6c 10openembedded.git/conf/distro/include/preferred-shr-versions.inc: Feb 23 14:18:40 preferred-shr-versions: use mesa-dri from git Feb 23 14:18:40 Signed-off-by: Martin Jansa Feb 23 14:18:41 03Martin Jansa  07org.openembedded.dev * r9471f8941b 10openembedded.git/recipes/xorg-xserver/ (8 files in 2 dirs): Feb 23 14:18:42 xserver-xorg_git: add more pending patches for latest dri2 changes Feb 23 14:18:42 Signed-off-by: Martin Jansa Feb 23 14:18:43 03Martin Jansa  07org.openembedded.dev * r5c3846f9bb 10openembedded.git/recipes/xserver-kdrive-common/ (2 files in 2 dirs): Feb 23 14:18:43 xserver-kdrive-common: SHR - don't force user to calibrate his screen after first boot, but restore calibration if available. Feb 23 14:18:44 Signed-off-by: Martin Jansa Feb 23 14:19:42 I'm running into some serious system slowdown when building on ext3/sw-raid-1 partition (on a core-i7), so hoping a non-raid ext4 will be better Feb 23 14:20:59 raid1, raid5... Feb 23 14:21:10 how many hdd you guys have in your machines? Feb 23 14:21:33 I have just desktop for all things and 320+500GB inside Feb 23 14:22:43 hehe, my macbook has only 1 disk :+ Feb 23 14:22:58 which can hold all my crap for now :P Feb 23 14:23:11 Romke: my laptop has 80GB hdd Feb 23 14:24:50 cbrake: that's strange, I've never noticed any performance problem with ext3 on raid1. Feb 23 14:25:14 hrw, "blocks" (the autobuilder) was just tossed together out of spare parts -- I use RAID-5 because I need disk space and had lots of old disk drives lying about (and I was too cheap to buy a large new one!) :-D Feb 23 14:25:16 oh, 320GB here Feb 23 14:25:17 :+ Feb 23 14:25:18 one of our older arrays here is ext3 on raid5 and that one is a bit sluggish but I think it's more down to the slow disks than anything software related. Feb 23 14:25:27 and I don't own a desktop anymore Feb 23 14:26:11 I'd be interested to hear if switching to ext4 does make a difference for you though. If it does then I will have to try it for myself :-) Feb 23 14:27:03 mwester: I build second desktop from spare parts - now my youngest brother uses it at home instead of his athlonxp machine ;D Feb 23 14:27:05 I ended up putting the downloads directory on the RAID, moving a whole lot of other seldom-accessed data there as well, so that my TMPDIR can fit on a small non-RAID disk. Much faster build times that way, because RAID-5 is just slow writing to disk. Feb 23 14:27:46 mwester: I need to finish sheeva fixes - probably will move DL_DIR there for nfs mounting Feb 23 14:27:58 my DL_DIR is ~30GB Feb 23 14:29:33 pb_: yeah, it seems the gui gets sluggish at times, which is very annoying where there is still CPU bandwidth left. Perhaps I should just blame Ubuntu :-) Feb 23 14:31:17 hrw: I have 2, 1TB disks for a raid-1 setup, then several 250GB scratch/working drives -- everything in one system (workstation, file server, etc) Feb 23 14:56:36 anyone here got some info, why udev (151-r1.5) wouldn't execute a script in a rule a boottime, but successfully executes it when hotplugging the device ? Feb 23 14:59:14 actually I think I did something wrong, but i thought that just putting it in /etc/udev/rules.d/ with a number in front of it would the rules to be fired at boottime too Feb 23 15:01:00 mlip: check that udev trigger/sync stuff is still called on boot Feb 23 15:01:18 mlip: it may be for speed we stopped doing that Feb 23 15:01:54 XorA: how? haven't done too much with udev (despite rules/hotplugging) Feb 23 15:02:05 does it generate log entries ? Feb 23 15:02:47 probably not, udev starts before / is writeable Feb 23 15:06:24 libunique doesn't build because of -Werror Feb 23 15:16:46 ^can be solved by --enable-maintainer-flags=no Feb 23 15:21:05 XorA: udev trigger/settle are only called on first boot; otherwise something like trigger --subsystem-nomatch=tty .... Feb 23 15:21:08 is this the problem ? Feb 23 15:25:19 mlip: yes because it means scripts arent run Feb 23 15:28:01 XorA: any preferred procedure to get the scripts executed? or do I have to add an init scripts that runs udevadm with specific subsystem and attr-match ? Feb 23 15:28:34 mlip: bleugh, its all a balance between speed and usefulness, udev is really really slow at first boot Feb 23 15:29:00 mlip: Im not qualified to guess the correct method to fix Feb 23 15:29:24 well, I am updating gpsd and trying to integrate it into udevd (koen suggested running gpsd using a socket) Feb 23 15:29:47 i have successfully updated the daemon and all libs; it is also working when hottplugging it Feb 23 15:30:06 the only thing left is the "right" startup Feb 23 15:30:33 I think on my boards I just turned off the udev caching anyway Feb 23 15:30:39 hey Feb 23 15:30:48 hey kergoth Feb 23 15:30:58 hej Feb 23 15:31:30 ROOTFS_POSTPROCESS_COMMAND += "rm -f ${IMAGE_ROOTFS}/etc/default/udev;" :-D Feb 23 15:31:41 morning kergoth Feb 23 15:31:42 :) Feb 23 15:31:56 i dont think you would like that in the tree ;) Feb 23 15:32:12 * XorA never much cared for the caching Feb 23 15:32:34 devtmpfs + backgrounded udev could be cool Feb 23 15:33:18 or finding out why it takes udev 10s to scan /sys but mdev only 200ms Feb 23 15:33:36 haha Feb 23 15:33:40 wow, mdev is that fast? Feb 23 15:33:42 nice Feb 23 15:33:58 mdev is very fast Feb 23 15:33:59 kergoth: it was once upon a time, no guarantees it still is Feb 23 15:34:04 heh Feb 23 15:34:23 meh, i hate fixing the history when git-svn wasn't smart enough to pick up the branch points Feb 23 15:59:14 time to go home Feb 23 16:15:22 yeah, mdev is way faster. 200ms still sucks though, it would be nice to get rid of that dead time. Feb 23 16:15:38 udev is just so awful in terms of boot time that it doesn't bear thinking about. Feb 23 16:15:41 use devtmpfs? Feb 23 16:16:01 yeah, that is a possibility Feb 23 16:16:44 hrw: that doesnt get rules run though Feb 23 16:16:56 finding why udev sucks time so badly would be good Feb 23 16:17:14 XorA: speed or rules Feb 23 16:17:49 devtmpfs lacks rules but is in kernel. mdev has very simple rules but is fast, udev had complex rules but thats only plus Feb 23 16:18:13 hrw: udev is now integrated with *kit and dbus and ...... Feb 23 16:18:31 XorA: udev or libudev? Feb 23 16:18:35 soon you will need the whole of gnome to boot :-) Feb 23 16:18:53 hrw: isnt libudev a no-op without udev? Feb 23 16:20:00 did not checked Feb 23 16:20:14 but I know that HAL does not required udev - just udevinfo Feb 23 16:20:26 hal is gone though Feb 23 16:20:47 libudev can be simple way to get all infos which hal gave == udevinfo in library form Feb 23 16:21:03 me would like to have whole gnome to boot if it's still boot in < 2 mins :) Feb 23 16:21:33 JaMa: specially on headless ;D Feb 23 16:21:51 yes, there it can be called hidden bonus :) Feb 23 16:22:37 no, seriously, I'm looking for usable virtual keyboard working with touchscreen throught gpm.. Feb 23 16:22:45 * XorA resists the giving head jokes Feb 23 16:58:20 03Martin Jansa  07org.openembedded.dev * r00c31b00f8 10openembedded.git/ (3 files in 3 dirs): Feb 23 16:58:20 ffalarms: moved from svn to git, update PV, remove applied patch Feb 23 16:58:20 Signed-off-by: Martin Jansa Feb 23 16:58:30 03Martin Jansa  07org.openembedded.dev * ref928c4436 10openembedded.git/recipes/ffalarms/ffalarms_git.bb: Feb 23 16:58:30 ffalarms: oe-stylize Feb 23 16:58:30 Signed-off-by: Martin Jansa Feb 23 17:01:16 03Martin Jansa  07org.openembedded.dev * r6eb74e2f10 10openembedded.git/recipes/ffalarms/ffalarms_git.bb: ffalarms: kill SRCPV Feb 23 17:03:21 03Cliff Brake  07org.openembedded.dev * r86d55b2f5a 10openembedded.git/: Merge branch 'org.openembedded.dev' of git://git.openembedded.net/openembedded into org.openembedded.dev Feb 23 17:03:32 03Cliff Brake  07org.openembedded.dev * r2d8dc03f6f 10openembedded.git/contrib/weekly-changelog-report.py: weekly-changelog-report.py: add staging-branch Feb 23 17:05:56 pb_: thoughts on using gcc's --with-sysroot=? Feb 23 17:12:12 kergoth: what about it? Feb 23 17:12:24 kergoth: Effecitively we already do... Feb 23 17:13:54 erm, miswrote, i meant the --sysroot= compiler argument as an alternative to the -isystem, etc, not the configure argument :) Feb 23 17:14:19 kergoth: We actually got rid of that entirely in poky Feb 23 17:14:45 * kergoth really needs to pay closer attention to poky development, apparently :) Feb 23 17:14:49 kergoth: Just never dared in OE as I did have to cleanup some packages in poky to make them work properly Feb 23 17:15:04 kergoth: It was ages and ages ago Feb 23 17:15:09 ah Feb 23 17:15:58 so just needs some tlc / massaging Feb 23 17:17:46 kergoth: http://git.pokylinux.org/cgit.cgi/poky/commit/?id=ba2e1f4d933c37b372d6749d64614f2510ee9d7b Feb 23 17:17:56 thanks, appreciate it Feb 23 17:18:11 kergoth: Two years+ ago :) Feb 23 17:18:17 heh :) Feb 23 17:18:23 now that's a backlog :) Feb 23 17:18:40 kergoth: The thing is I have a definitive test suite for Poky Feb 23 17:18:46 I can't say the same for OE Feb 23 17:18:57 Therefore I never risk pushing changes like this :( Feb 23 17:19:18 ah :( Feb 23 17:19:47 well, it's not .stable, we can't afford to let it stagnate just to avoid all risk.. need to find a balance Feb 23 17:20:00 have a nice rest of day Feb 23 17:20:15 I know but I just can't afford the time it takes to push a change like that, then catch all the fallout :( Feb 23 17:20:18 * kergoth nods Feb 23 17:20:19 How do I tell OE to add the '-g' option to gcc, in order to get some debugging info? Feb 23 17:20:25 i don't mind taking the heat :) Feb 23 17:20:44 kergoth: There are a few nice changes in Poky it would be good to push... Feb 23 17:21:27 feel free to pass em my way, or add em to the google wave i'm playing with Feb 23 17:21:39 kergoth: Like the top part of http://git.pokylinux.org/cgit.cgi/poky/commit/?id=7fc5d6eeea6c633682f4cc4759f9e9e99cb2c309 Feb 23 17:22:08 The mangling is still in OE, I just removed it from Poky Feb 23 17:22:35 wjward: if all you want is debug symbols and the like, but want to retain the optimizations, you can just add -g to CFLAGS. If you want to avoid the optimizations as well to simplify the debugging, you can set DEBUG_BUILD = "1". see line 501 in openembedded/conf/bitbake.conf Feb 23 17:23:09 kergoth: great! thanks very much. Feb 23 17:23:47 wjward: note that debug symbols are always split out of the main binaries into separate packages. if you want to debug, say, nano, make sure you install nano-dbg as well as nano into your filesystem Feb 23 17:23:58 right Feb 23 17:52:10 just for the record, I was told by someone that Angstrom's boot time was great :) Feb 23 17:52:24 of course, the reference was a desktop linux box Feb 23 18:03:15 heh Feb 23 18:44:31 forgive my ignorance but should the SA-1110 "for an ipaq h3800" be linux-uclibc or linux as the target os can't really find any thing rock solid on google to make a choice Feb 23 18:45:03 either would be fine Feb 23 18:45:18 ok thank you Feb 23 18:52:11 anyone tried building for at91sam9261ek recently? Feb 23 18:52:17 pb___: thoughts on something like http://kergoth.pastey.net/133359 to go along with the toolchain desuck, perhaps? Feb 23 19:12:37 hi eFfeM Feb 23 19:19:33 kergoth: yeah, using --sysroot seems like a sensible idea. I am not all that enamoured of ${TOOLCHAIN_OPTIONS} as a variable, but I think the principle of the patch is a good one. Feb 23 19:24:19 pb___: thanks :) Feb 23 19:24:28 And yeah, I didn't like the var name either, but it was there Feb 23 19:24:51 yeah, fair enough Feb 23 19:25:03 * Tartarus puts this laptop back in the playpen (where his son can't get at it) and goes back to the work laptop Feb 23 19:25:35 I wonder whether it would be reasonable to have native.bbclass add a --sysroot=${STAGING_DIR_NATIVE} as well Feb 23 19:25:57 rather than just zapping it to nothing Feb 23 19:27:11 yeah i was wondering about that too Feb 23 19:28:53 pb___: if you add --sysroot then linker wont look beyond it Feb 23 19:29:20 I think we use some libraries from build too Feb 23 19:29:23 Tartarus: did the stagefile bits hit the ML for review and I didn't see it? kernel.bbclass isn't a recipe, it is subject to the usual two-ack process Feb 23 19:31:48 and some native toolchains (e.g. ubuntu) dont enable their linkers to support sysroot Feb 23 19:36:09 03Tom Rini  07org.openembedded.dev * r8fe1beae6d 10openembedded.git/recipes/python/python-pysqlite2-native_2.4.0.bb: python-pysqlite2-native: New recipe Feb 23 19:36:12 03Tom Rini  07org.openembedded.dev * r072cad0100 10openembedded.git/recipes/initscripts/ (files/umountfs initscripts_1.0.bb): initscripts: Make umountfs a bit more robust, bump PR Feb 23 19:42:35 pb___: No, it would not Feb 23 19:42:53 pb___: Which libc do you want to link against? Feb 23 19:52:30 re Feb 23 19:52:38 hi florian , could I ask an embarrassing question(for me): I don't remember the name of the person sitting on the chair that is on the right,behind his laptop in this picture: http://picasaweb.google.com/fransmeulenbroeks/Fosdem2010?authkey=Gv1sRgCPfSh7qcjv2siQE#5435938084619204290 Feb 23 19:53:31 GNUtoo: gremlin :) Feb 23 19:53:35 ok thanks a lot Feb 23 20:04:55 lovely, the cmake recipe isn't building :) Feb 23 20:11:48 GNUtoo: isn't that Gremlin? Feb 23 20:11:59 Oh, I see it's already been answered Feb 23 20:12:07 yes btw thanks a lot Feb 23 20:28:25 is there any way to prevent RRECOMMENDS from getting into the image? Feb 23 20:34:06 re Feb 23 20:38:17 denix: BAD_RRECOMMENDS Feb 23 20:38:31 denix: but use with care ;) Feb 23 20:40:17 RP: it doesn't seem completely unreasonable to either symlink the host libs that we know we want into the sysroot, or build a glibc-native. Feb 23 20:40:47 symlinking sounds nice, it'd be good to do that for the binaries too Feb 23 20:40:53 RP: that way, you would avoid the problem of host libraries leaking unexpectedly into the build in place of accidentally-omited dependencies on -native packages. Feb 23 20:41:35 DJWillis: BAD_RECOMMENDATIONS, thanks for the pointer! :) Feb 23 20:42:05 pb___: Its a nice idea although I suspect it could be a pain in practise... Feb 23 21:07:06 RP: any news about BBLAYERS ? Feb 23 21:09:03 mckoan: being discussed on the mailing list Feb 23 21:09:10 but I need something in poky soon :/ Feb 23 21:09:54 oneshel: any problem with at91sam9261ek? Feb 23 21:12:02 how many double precition floatingpoint operations u think an arm9 can stand at 433 mhz? what u guess? Feb 23 21:12:37 3? Feb 23 21:12:50 mckoan: I'm just picking it back up, just curious if anyone had been using it Feb 23 21:15:53 oneshel: I think it's ok, but actually I'm using at91sam9263ek and at91sam9g20ek without problems (until today) Feb 23 21:16:47 hmm, can't seem to get BAD_RECOMMENDATIONS to work. any ideas? Feb 23 21:17:07 oneshel: I'm testing Angstrom with a build from scratch because I'm facing to a problem with dbus-glib-1 after last patches Feb 23 21:20:48 mckoan: that's good to hear, I have a build running right now using the 2009 stable branch Feb 23 21:22:23 oneshel: I didn't use stable for ages Feb 23 21:22:59 mckoan: yeah? I figure if this build works I'll go with something more recent Feb 23 21:23:08 mckoan: that's more good news though :) Feb 23 21:25:32 * mckoan confirm last dbus patches broke Angstrom Feb 23 21:28:09 eggdbus_0.6.bb do_configure failed http://pastebin.com/NEuN3JCE Feb 23 21:33:40 enough for today Feb 23 21:33:46 'nite Feb 23 21:36:30 03Denys Dmytriyenko  07org.openembedded.dev * r250e384fec 10openembedded.git/recipes/eggdbus/eggdbus_0.6.bb: eggdbus: add dependency on dbus-glib Feb 23 21:36:48 mckoan|away: ^^^ Feb 23 21:54:37 pb___: Any chance you own one of those distros that has FEATURE_SUID=y ? Feb 23 21:57:12 Tartarus: I guess I sort of own micro. Feb 23 21:58:02 I think we might want to turn FEATURE_SUID off in the longer term, there, but I am happy to leave it on for the time being. Feb 23 22:02:56 ok Feb 23 22:03:15 Like I said in the start of the email, I know folks have or had strong opinions about it Feb 23 22:03:27 And since this would make it actually active... Feb 23 22:10:08 * RP__1 is getting as bad as pb___ :) Feb 23 22:27:34 RP__________1: heh Feb 23 22:28:08 Tartarus: yah, it is a bit weird that it hasn't actually been active until now. I wonder how angstrom has actually been working without that. Feb 23 22:28:42 for micro, it isn't much of an issue because we don't (in a standard install) actually use busybox for anything much that requires setuid. Feb 23 22:30:14 hail zecke Feb 23 22:33:54 jo Feb 23 22:37:20 hi woglinde Feb 23 22:39:08 pb___: For angstrom, I bet the answer is similar in that they don't really use that parts that need suid too Feb 23 22:39:25 That said, it's nice to have when it works :) Feb 23 22:42:03 * Crofton guesses not many people actually use the SUID stuff :) Feb 23 22:42:22 * Crofton notes he is logged in as root anyway Feb 23 22:42:36 hello folks Feb 23 22:46:23 Tartarus: yeah, I guess. I am a bit surprised that angstrom isn't using the setuid bits. Feb 23 22:47:12 but, as you say, it would be nice to have them working. I think you can probably go ahead and check in your patch. Feb 23 22:47:40 Tartarus, when did you send that email to the list? Feb 23 22:47:55 few hours ago Feb 23 22:48:18 you should wait a day, so that people in different TZ's have a chance to see it Feb 23 22:48:32 irc is not authoritative :) Feb 23 22:48:42 and, has anyone built cmake lately :) Feb 23 22:48:46 ? Feb 23 22:48:56 crofton nope Feb 23 22:49:14 yeah, it is failing Feb 23 22:49:20 I fear I will need to look at it Feb 23 22:50:17 woglinde: I'm ending up thinking that we need two recipes of klibc: one for 2.6.2x pre-headers_install and one for 2.6.3x which I'll try to fix... Feb 23 22:50:36 yep, it is on my list of things to worry about Feb 23 22:51:00 So, oh yes Feb 23 22:51:05 I'm not trying to get it in right now Feb 23 22:51:14 I just wanna make sure at least certain eyes do see it Feb 23 22:51:46 Since it's making a semi controversial feature actually work, I want folks to know and opt out if they don't really want it on Feb 23 22:52:03 Tartarus, thanks Feb 23 22:52:08 ant yo I saw it on the ml Feb 23 22:52:33 ah, I missed it then Feb 23 22:53:48 fwiw, minimal will embrace the busybox SUID change Feb 23 22:55:03 mickey|zzZZzz: very good Feb 23 22:56:05 i've been complaining about tinylogin since ages Feb 23 22:56:27 anyone: am having difficulties with bitbaked images and images from online builder. when i untar them, i am getting broken symlinks. invoke with 'target_dir$ tar xvf ball.tar'... ??? Feb 23 23:04:18 03Chris Larson  07org.openembedded.dev * r54373795ff 10openembedded.git/classes/insane.bbclass: Feb 23 23:04:18 insane: check for .debug as a path component, not part of the path string Feb 23 23:04:18 Signed-off-by: Chris Larson Feb 23 23:08:39 Hmm, how can I specify an include early on in a conf file which I want to include last to make weak assignments if they've not been made? :/ Feb 23 23:09:20 I was talking to tom about weak assignments the other day, was thinking it could be nice to make ?= lazier. make the last ?= be the one thats used, rather than the first. thoughts? Feb 23 23:09:33 that way the most specific default would be used Feb 23 23:10:25 woglinde: opinion about the ln -sf ${STAGING_KERNEL_DIR} linux in klibc do_configure? Feb 23 23:12:00 kergoth: I'm not sure that changing the semantics of ?= is a good thing, seems like too much risk of introducing subtle failures. Adding a new "??=" operator that does what you suggest seems like a fine idea though. Feb 23 23:12:45 then we just need a "?!?!?!!=" operator and we'll be all set. Feb 23 23:12:47 Yeah, that's a good point, that sort of change is likely a bad idea, particularly lacking any sort of versioning of the file format or something Feb 23 23:12:49 rofl Feb 23 23:12:49 hehe Feb 23 23:12:58 * XorA thinks pb___ has been on the perl today Feb 23 23:13:01 Well, we could find out Feb 23 23:13:10 Make bitbake print something when it does hit it Feb 23 23:13:18 Then go try a bunch of combos Feb 23 23:13:29 Check and see if there's really any ?= that get goofed Feb 23 23:13:40 the problem is, we can't check external repositories directly Feb 23 23:13:41 (Or conversley if we're fixing something) Feb 23 23:14:03 it is confusing being 3 hours left of my normal TZ Feb 23 23:14:07 kergoth: right, but we can also loudly mention the change Feb 23 23:14:09 kergoth: ?= is useful, I just want a way to say "only use this if nothing else wants to set it" Feb 23 23:14:20 And if people still get bit, well, that does happen with out of mainline stuff Feb 23 23:14:32 Or am I just too much of a kernel guy at heart still? :) Feb 23 23:14:51 Tartarus: heh, I think you are. don't worry, it'll pass. Feb 23 23:15:02 ha Feb 23 23:15:26 but, more seriously, I think the existing ?= semantics are sensible and probably worth preserving in some form. Feb 23 23:15:33 Tartarus: I think the behaviour of ?= is very relied upon. It would screw poky up to change it for certain Feb 23 23:15:35 I maintain some of these external ones, so it's not like I don't have some skin in the game, so to speak :) Feb 23 23:15:49 if you want to add a new operator with your new semantics, that'd be fine and I think you should absolutely do it. Feb 23 23:15:58 RP: ?= *is* useful, but i suspect the semantics when there are multiple ?='s around the config files could confuse. though, i suspect the majority will just go away when more default values move into the classes Feb 23 23:16:09 pb__, RP, really? OK, seriously, lemme hear it. I'm sure there's stuff I don't think about :) Feb 23 23:16:48 I just don't get ?= gets the first one is all Feb 23 23:17:10 (So today we're doing ?= then = then = later still) Feb 23 23:17:21 Which works, but is also funny looking Feb 23 23:17:36 (and depends a little on us knowing some of the order) Feb 23 23:17:43 ?:=? Feb 23 23:18:02 Tartarus: classes doing ?= and code using the classes that need something specific do = Feb 23 23:18:11 Tartarus: well, that's just the way that bitbake assignments traditionally work. they're always immediate at the point of parsing, at least as far as the lvalue is concerned. Feb 23 23:18:12 nobody need care about the order Feb 23 23:18:18 i don't think there's any getting away from knowing the order, it's just too important in too many ways. the layering via config file load order and overrides is critical to how we do things Feb 23 23:18:36 and people do use ?= as part of that Feb 23 23:18:39 which is fine Feb 23 23:18:49 I'm tempted by a ??= though Feb 23 23:19:40 Tartarus: if you want to add a new one which defers the assignment until later, that is definitely a different set of semantics. I think it is a useful set of semantics, but it is not how bitbake has traditionally behaved. Feb 23 23:19:47 on a related note, i think not enough people know they can override the distro/machine using an override, rather than hacking the distro/machine .conf to use ?= Feb 23 23:19:52 _local is our friend Feb 23 23:19:58 also, fwiw, you can emulate ??= today using an anonymous function, whereas you couldn't emulate the existing ?= that way. Feb 23 23:20:40 pb___: I did wonder about an anonymous function but for the number of variables I want to set, it would be ugly as sin Feb 23 23:20:57 I guess there is a slightly murky issue around what happens if you have two ??= assignments to the same var and which one wins. Feb 23 23:21:14 pb___: Question is first or last? Feb 23 23:21:54 RP: yeah, I didn't mean the anonymous function as an answer to your issue specifically. that was more in response to tartarus asking why the existing ?= semantics were valuable and should be kept. Feb 23 23:21:57 So, maybe we aren't being clear about the use for ??= Feb 23 23:22:11 In say pstaging stuff Feb 23 23:22:19 pb___: I'd say last wins as otherwise its impossible to change a value Feb 23 23:22:20 We default to some parts of the path with ?= Feb 23 23:22:47 In our distro, we want to change those default, but still have the notion of "these are still just suggestions" Feb 23 23:23:01 Whereas when I see a '=' I think, ok, that needs to be set to that, for good reason Feb 23 23:23:24 So what we can do is just comment and = our new suggestion Feb 23 23:24:10 Which is what got me wondering why ?= is a first found not a last found Feb 23 23:24:35 Since today you don't get what I expected as the behavior (2nd ?= is ignored) Feb 23 23:25:32 Tartarus: well, yeah, but I think that is just because your expectations are slightly misplaced. the semantics of ?= are well documented: it means to set the variable if it wasn't previously set. Feb 23 23:25:53 now, that may well not be what you wanted in this particular instance, but it doesn't mean that those semantics are broken. Feb 23 23:27:03 Yeah Feb 23 23:27:10 the 2nd ?= isn't exactly ignored, it's just that it (correctly) determines that the variable is already set and hence doesn't re-set it. Feb 23 23:27:12 I'm willing to accept that Feb 23 23:27:23 I'm just asking if really, that's a good way to go Feb 23 23:27:56 ie today if there's any 2+ ?='s being found, is that a hidden bug or an intentional feature Feb 23 23:28:08 Tartarus: I suspect we're thinking about the same thing Feb 23 23:28:34 So a ??= in bitbake 1.10.0 it is? Feb 23 23:28:34 Specifically I'd like my layer.conf file to specify an include file where a load of default values can be obtained from Feb 23 23:28:46 Tartarus: Tempting :) Feb 23 23:29:07 Seriously, if you guys are sure it's a good behavior choice, that's fine. I'm just asking :) Feb 23 23:29:21 What I hate about file format changes is getting nice errors. add that, then try to parse it with an old bitbake and it explodes and dies unpleasantly :\ Feb 23 23:29:21 And speaking of 1.10.0, got a swag at when it'll be out? :) Feb 23 23:30:07 kergoth: hence the minimum version requirement checks Feb 23 23:30:15 kergoth: heh. that has more to do with bitbake parser suckage than any fundamental problem with file format changes. Feb 23 23:30:30 RP: won't it explode before it even gets to run the version requirement check? Feb 23 23:30:33 RP: and that'll help with a *parse error* how exactly? it'll freak out before it ever gets to the finalise() Feb 23 23:32:17 03Martin Jansa  07org.openembedded.dev * r2022a92200 10openembedded.git/recipes/tasks/task-shr-feed.bb: Feb 23 23:32:17 task-shr-feed: add mcnavi Feb 23 23:32:17 Signed-off-by: Martin Jansa Feb 23 23:32:19 03Thomas Zimmermann  07org.openembedded.dev * r31dd962614 10openembedded.git/recipes/openmoko-3rdparty/mcnavi_0.2.5.bb: Feb 23 23:32:19 mcnavi: new recipe Feb 23 23:32:19 Signed-off-by: Martin Jansa Feb 23 23:32:21 kergoth: now, if only we could rename bitbake to, say, oemake and put it inside the oe tree... :-} Feb 23 23:32:39 heh :) Feb 23 23:33:11 then each distro could customize bitbake Feb 23 23:33:15 i was thinking about playing with a zc.buildout. clone oe, run the buildout script, it grabs the bitbake you need Feb 23 23:38:57 RP: ? :) Feb 23 23:48:32 Any ML admins around? Feb 23 23:52:55 depends what you need. I can deal with generic email issues on ltg for you, but for list-specific mailman config tweaks you probably need überhacker mickey|zzZZzz Feb 23 23:54:34 Ah Feb 23 23:54:39 Well, I keep getting bounce disabled Feb 23 23:54:50 And I'm gonna assume when I file an IT ticket, they'll need an example or two Feb 23 23:55:02 So not you I assume :) Feb 23 23:55:24 let me have a quick look. it's possible that mailman keeps its bounces in some easy to find place. Feb 23 23:56:11 oh yeah, you are quite bouncy. Feb 23 23:56:13 Feb 24 00:11:02 2010 (14959) openembedded-devel: tom_rini@mentor.com current bounce score: 5.0 Feb 23 23:56:13 Feb 24 00:11:02 2010 (14959) openembedded-devel: tom_rini@mentor.com disabling due to bounce score 5.0 >= 5.0 Feb 23 23:56:30 Yeah, got that email at least, and re-enabled Feb 23 23:56:38 It's like the 3rd or 4th time :( Feb 23 23:58:14 sadly no, it doesn't seem that mailman keeps the actual bounces anywhere that I can conveniently get at them. Feb 23 23:58:38 if you send a note to mickey and florian then I expect one of them can sort you out. Feb 23 23:58:56 k, thanks Feb 23 23:58:57 failing that, if you poke me again tomorrow then no doubt I can crack the mailman system given enough time :-} Feb 23 23:59:26 * pb___ bedtime now Feb 23 23:59:27 night all Feb 23 23:59:48 nite pb Feb 24 00:00:33 Sorry, got distracted Feb 24 00:00:49 Yes, of course it would explode, I'm not thinking straight Feb 24 00:01:10 I also don't know how to solve that problem although putting the version check in the layer.conf file could work... Feb 24 00:01:19 assuming that doesn't use the new syntac Feb 24 00:01:22 syntax Feb 24 00:03:59 woglinde: argh..I discovered the second getline is the one on my buildhost :/ Feb 24 00:04:08 something obviously wrong... Feb 24 00:04:53 scripts/unifdef.c:209: error: conflicting types for 'getline' Feb 24 00:04:53 | /usr/include/stdio.h:651: error: previous declaration of 'getline' was here Feb 24 00:05:06 Ah glibc 2.10 Feb 24 00:05:12 ^^this is x86 stdio.h Feb 24 00:06:06 ant__: I put like a million patches for that in OE :-) Feb 24 00:06:13 ant__: its real simple Feb 24 00:06:38 XorA: the bug must be in the single line of the headers_install patch I added Feb 24 00:07:01 klibc tries to install headers using kernel script Feb 24 00:07:35 ant__: recipes/linux-libc-headers/files/unifdef.patch Feb 24 00:08:04 +<----->$(Q)make -C $(KLIBCKERNELSRC) ARCH=$(KLIBCARCH) INSTALL_HDR_PATH=$(INSTALLROOT)$(INSTALLDIR)/$(KCROSS) headers_install Feb 24 00:08:05 recipes/uclibc/files/unifdef-fix.patch Feb 24 00:08:25 do you see anything obviously wrong? Feb 24 00:08:33 ant__: you are not looking Feb 24 00:08:36 yes Feb 24 00:08:43 the issue was not present Feb 24 00:08:45 ant__: look at those patches, they show exactly what is wrong Feb 24 00:09:10 oki Feb 24 00:09:32 I had fun finding all those fixes when fc10 became fc11 Feb 24 00:10:44 I see, is the renaming like woglinde suggested Feb 24 00:11:15 ant__: just apply that patch to your source :-) Feb 24 00:11:25 ant__: or wherever your unifdef is coming from Feb 24 00:12:02 he..I'm using a symlink to the staged linux Feb 24 00:12:19 ln -sf ${STAGING_KERNEL_DIR} linux in klibc do_configure Feb 24 00:12:32 so the script comes from there Feb 24 00:13:05 ant__: hmmm, would be better to make kernel.bbclass replace the internal unifdef with the one from unifdef-native Feb 24 00:13:11 ant__: then you only need to fix once Feb 24 00:13:36 anyway more vampire diaries for me Feb 24 00:13:44 he Feb 24 00:18:01 good night Feb 24 01:21:36 So, hmm Feb 24 01:21:49 Why does sdk.bbclass say LDFLAGS=$BUILD_LDFLAGS and not $SDK_LDFLAGS ? Feb 24 01:24:30 ah nm :( Feb 24 01:31:13 OK, this looks workable, just need for my coworker to finish unbitrotting the mingw stuff Feb 24 01:31:27 And then I think we can use SDK_LDFLAGS in *-cross-sdk Feb 24 01:31:35 And still be correct in *-canadian-sdk Feb 24 01:31:51 And then get back to talking about what is or isn't insane for making sdks with OE Feb 24 01:32:02 * Tartarus looks for RP Feb 24 01:32:45 what's the correct way to call python function from shell function? Feb 24 01:35:31 Tartarus, +1 on figuring out a better name for a package that installs a working toolchain on the device :) Feb 24 01:36:17 * Crofton suspects RP may be in bed Feb 24 01:36:28 but my tz skillz are poor atm Feb 24 01:37:58 03Tom Rini  07org.openembedded.dev * r82dc1d8cf6 10openembedded.git/recipes/gdb/ (3 files): Feb 24 01:37:58 gdb-cross-sdk: Add zlib-sdk to DEPENDS. Feb 24 01:37:58 No PR bump since we don't actually link vs the -sdk libs. Feb 24 01:41:13 he probably is, heh Feb 24 01:41:38 has anyone worked with the DSP on the omap processor? Feb 24 01:41:46 But, this will be an improvement over what we don't do now, see that last commit message Feb 24 01:41:51 gdb-cross-sdk links w/ BUILD_LDFLAGS Feb 24 01:42:11 So having zlib-sdk in depends (and ncurses-sdk too) is nice, but if zlib-native wasn't built, it won't find -lz anyhow Feb 24 02:20:54 denix: there's no way to do that. add a new task :) Feb 24 02:21:16 kergoth: that's what I'm trying to do... :) Feb 24 02:23:11 kergoth: since there are 2 shell functions calling one another, I can't addtask a python func in between? should I instead generate shell commands from that python func and ${@func(d)}, is it correct? Feb 24 02:23:35 could do that. ugly, but might be the only way to get what you want Feb 24 02:26:44 thanks **** ENDING LOGGING AT Wed Feb 24 03:00:02 2010