**** BEGIN LOGGING AT Mon Aug 31 02:59:57 2009 Aug 31 03:21:22 hi Aug 31 03:21:29 there are sombody? Aug 31 03:25:00 somewhere there is always somebody Aug 31 06:41:07 good morning all ! Aug 31 07:12:48 indeed. good morning. Aug 31 07:24:49 good morning Aug 31 08:10:32 good morning Aug 31 08:15:35 hi pH5 Aug 31 08:16:04 hej florian Aug 31 08:37:19 hi Aug 31 08:41:30 hi darxkies Aug 31 09:20:58 Hello, how can i build or where can i get package libgcc-dev on angstrom console-image kernel 2.6.29 ? Aug 31 09:24:03 it should be built as part of gcc. Aug 31 09:24:54 faad2 sources are in .zip archive, so I get 'sh: unzip: command not found' when unpacking it. DEPENDS = "unzip-native" doesn't work though Aug 31 09:25:37 morning Aug 31 09:26:08 mckoan: hi, got my mails? Aug 31 09:28:16 pb_: i found only this: http://www.pastebin.org/13610 Aug 31 09:32:43 is there a tips&tricks section about how to make the rootfs readonly? right now i turned the udev/caching off but dropbear fails to start due to /var/lib not being writable Aug 31 09:36:24 pb_: the libgcc-dev should be created after 'bitbake libgcc' ? Aug 31 09:36:26 no suggestions for unzip missing problem? Aug 31 09:40:18 is there someone who use qt-embedded on console image with -qws and his touchscreen works? Aug 31 09:44:04 darxkies, usually /var/lib would be in tmpfs or some such if / is readonly. Aug 31 09:48:03 i did append in fstab ro to defaults for / Aug 31 09:48:22 but /var/lib is not of type tmpfs Aug 31 09:48:31 btw i am using angstrom Aug 31 09:49:18 you might want to try #angstrom, I think that's where the angstrom h4x0rs hang out. Aug 31 09:50:17 or maybe post on their mailing list. I thought angstrom did use tmpfs for /var but I don't really know much about that distro myself. Aug 31 09:50:48 no Aug 31 09:51:04 ok, thanke you :) Aug 31 09:51:08 angstrom uses tmpfs for /var/volatile/ and symlink some of /var/ dirs there Aug 31 09:51:10 there is a /var/volatile Aug 31 09:51:12 yes Aug 31 09:51:24 though not /var/lib Aug 31 09:53:01 hrw: hi, received, I'm building Aug 31 09:53:35 mckoan: video drivers also need update - check my local.conf on quad Aug 31 09:53:50 hrw: ok Aug 31 10:02:02 mckoan: you also need to update gtk+/pango/glib-2.0 versions Aug 31 10:03:47 mckoan: reload my local.conf Aug 31 10:18:51 Where bitbake looking for includes (for example ) when compiling my own code? Aug 31 10:24:16 in staging/.../include Aug 31 10:26:07 good morning everyone Aug 31 10:26:56 morning Aug 31 10:31:33 pb_: thanks Aug 31 10:37:49 pb_: have a moment? Aug 31 10:37:54 sure Aug 31 10:38:06 pb_: http://marcin.juszkiewicz.com.pl/download/diffs/xserver-common/ contains my patches for xserver-common 1.3x Aug 31 10:38:20 pb_: the reason for 0024 is simple: make xserver-nodm-init works with xserver-common 1.3x Aug 31 10:38:30 pb_: now my at91sam9263ek boots to x11 session directly (with XKdrive) Aug 31 10:39:17 i wonder if we have reaced a state where we can get rid of xserver-kdrive-common Aug 31 10:39:18 ok, looks reasonable to me Aug 31 10:39:44 I'm not familiar with xserver-nodm-init but patch 0024 looks fine anyway Aug 31 10:39:46 florian: Xsession/Xinit scripts would be nice to compare Aug 31 10:40:40 pb_: my version of xserver-nodm-init calls ". /etc/X11/xserver-common;xinit /etc/X11/Xsession -- `which $XSERVER` $ARGS &" Aug 31 10:45:11 thx Aug 31 10:45:47 hrw, have you seen the new sam9 g45? Aug 31 10:47:17 czr: no Aug 31 10:47:39 hrw, k, nevermind then :-) Aug 31 10:48:09 czr: mckoan has used it iirc Aug 31 10:48:36 mckoan, did you have any problems with it? Aug 31 10:48:49 I'm hoping to get an ekes soon. Aug 31 10:50:16 czr: he will be online in ~1.5h Aug 31 10:50:28 * czr nods at hrw Aug 31 10:59:24 hi mickeyl Aug 31 10:59:28 morning hrw Aug 31 11:02:29 hi woglinde Aug 31 11:02:42 hi hrw Aug 31 11:03:08 bin/run_calibrate.sh is funny Aug 31 11:03:34 if there is xtscal, run it. else if there is ts_calibrate then run it. after all anyway run xtscal Aug 31 11:04:32 hrw: maybe the user hurried and copied it in the proper way... Aug 31 11:04:44 hi mickeyl Aug 31 11:05:05 morning pb_, I didn't forget the agenda bits. will send you something later today Aug 31 11:05:29 mickeyl: okay, awesome. Aug 31 11:06:05 Does the .bb recipe for compiling .cpp must be different than that which is used to compile .c file? Aug 31 11:07:22 03Marcin Juszkiewicz  07org.openembedded.dev * rb1aad586d8 10openembedded.git/recipes/xorg-driver/xf86-video-vesa_2.2.0.bb: xf86-video-vesa: fixed RDEPENDS Aug 31 11:07:33 mickeyl: put it in a wiki? Aug 31 11:07:33 03Marcin Juszkiewicz  07org.openembedded.dev * raa19cf6294 10openembedded.git/recipes/xserver-common/ (4 files in 2 dirs): (log message trimmed) Aug 31 11:07:33 xserver-common 1.30: another set of updates Aug 31 11:07:33 0024 moves functionality of /etc/X11/Xserver to separate file (except Aug 31 11:07:33 running X11 server) - this allows to use xserver-common package in Aug 31 11:07:35 systems where xinit is used to run X11 session as root (look at Aug 31 11:07:37 xserver-nodm-init package) Aug 31 11:07:39 0025 moves xmodmap files to /etc/X11/xmodmap/ and installs Aug 31 11:31:35 ho all again ... Aug 31 11:31:38 ping mickeyl Aug 31 11:33:53 khem, pb: I sent proposal for SPE/EABI handling Aug 31 11:37:22 hi gremlin[it] Aug 31 11:37:29 florian, hi ! Aug 31 11:37:43 florian, have you info abou the pxy document dor the OE GA ? Aug 31 11:37:55 florian, proxy document Aug 31 11:38:38 gremlin[it]: the instructions should be in the pdf attachment Aug 31 11:40:07 florian, never seen, i'll look better on backup email ... Aug 31 12:02:48 czr: sam9 g45 is in design phase, I haven't any board yet Aug 31 12:03:03 gremlin[it]: hi, how are you? :-) Aug 31 12:03:38 hi mckoan, fine thanks ! Aug 31 12:04:57 mckoan, how are u ? Aug 31 12:05:59 gremlin[it]: I'm lazily trying to restart working Aug 31 12:06:34 mckoan: then read my query d: Aug 31 12:06:39 mckoan, about me too ... i rub bitbake yestarday after some weeks ;) Aug 31 12:08:04 rub -> run Aug 31 12:22:49 hrw: thanks, I saw your mail. I still feel though that the abi selection should primarily be a question for the DISTRO, not for the MACHINE. Aug 31 12:23:13 pb_: sure, but how distro know which devices are capable of spe or eabi? Aug 31 12:23:34 pb_: what if distro will be used with external target device config? Aug 31 12:23:58 hrw: I don't quite understand what you mean by "external target device config"? Aug 31 12:24:36 pb_: if I will try to build Angstrom for my umbaumba strongarm based device I will have to change all angstrom configs to add umbaumba as "arm not capable of doing EABI" - thats today Aug 31 12:25:09 pb_: if eabi (or spe) is just 'feature' then it is done automatically and I do not have to change any configs Aug 31 12:26:55 what do you think? Aug 31 12:27:14 yeah, I guess that boils down to how you define a distro. angstrom is sort of a funny case because it has these various unbound parameters floating around. Aug 31 12:27:36 in that sense I kind of consider angstrom itself to just be a "meta-distro" rather than an actual concrete embodiment of a distro. Aug 31 12:27:39 pb_: it was just example Aug 31 12:28:05 hrw: sure, but my point is that most other distros want to pick a single abi and then stick with it. Aug 31 12:28:13 ok Aug 31 12:28:17 I see point Aug 31 12:28:31 because, in some sense, the definition of a "distro" is a group of packages that are interoperable with each other. Aug 31 12:29:05 pb_: but your way makes me need to build micro-spe, micro-armeabi, micro-armoabi, micro(normal ppc) etc - few distros just because my devices are different? Aug 31 12:29:15 if you have multiple different abis floating around then I don't really consider that to be a coherent single distro anymore, it's rather a group of distinct distros. Aug 31 12:29:48 hrw: if your devices are incompatible, yes. obviously nothing forces you to build the eabi version, you could run oabi on all arm devices. Aug 31 12:30:00 likewise, nothing actually forces you to build spe, you could run regular powerpc on all ppc devices. Aug 31 12:30:08 sure Aug 31 12:30:58 but yes, if you want to build binaries with all those different abis then I would consider those to all be different distros since you can't take a package from one and install it on another. Aug 31 12:31:24 unless it is noarch one Aug 31 12:31:30 indeed Aug 31 12:34:14 so, perhaps this would be a good item for the oedem agenda. Aug 31 12:34:27 good idea Aug 31 12:34:40 btw - offtopic question Aug 31 12:35:02 what is a proper name for kitchen sync tap? Aug 31 12:35:18 just 'kitchen sync tap' or something other? Aug 31 12:35:33 "kitchen sink tap", I guess Aug 31 12:35:45 thx Aug 31 12:35:48 I meant sink ;D Aug 31 12:35:56 heh Aug 31 12:36:58 there are more specific terms for different types of tap that you might find on a sink, but I don't think there's any particular word for a generic kind of tap. Aug 31 12:38:40 my pl<>en dictionary do not cover that anyway. time to find something better Aug 31 12:39:20 03Rui Miguel Silva Seabra  07shr/import * r4e6b5c46d4 10openembedded.git/recipes/curl/curl.inc: curl: enable SSL for curl. Blessed by dos1. Aug 31 12:51:22 03Sebastian Krzyszkowiak  07shr/import * rf64c8b9ed3 10openembedded.git/recipes/packagekit/ (2 files in 2 dirs): packagekit: fix opkg backend to install multiple packages Aug 31 13:05:23 hrw: kitchen faucet, in north america at least :) Aug 31 13:05:55 msmith_: faucet is US and I prefer UK version of English Aug 31 13:06:14 anywya I found on google Aug 31 13:06:16 i'm canadian so.. stuck in the middle Aug 31 13:06:53 the centre, i should say ;) Aug 31 14:25:06 morning Aug 31 14:25:53 hi kergoth booxter Aug 31 14:26:54 hi hrw Aug 31 14:27:49 hi kergoth Aug 31 14:30:46 03Koen Kooi  07org.openembedded.dev * rd1d7c677ae 10openembedded.git/ (conf/checksums.ini recipes/evince/evince_2.26.2.bb): evince: add 2.26.2 Aug 31 14:30:47 03Koen Kooi  07org.openembedded.dev * r2698abdc5e 10openembedded.git/ (3 files in 3 dirs): epdfview: update to 0.1.7 Aug 31 14:30:48 03Koen Kooi  07org.openembedded.dev * r2d5f5e12bb 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 31 14:30:48 03Koen Kooi  07org.openembedded.dev * rcac4ab3d9c 10openembedded.git/ (conf/checksums.ini recipes/poppler/poppler_0.11.3.bb): poppler: add 0.11.3 (0.12rc) Aug 31 14:31:33 hi kergoth Aug 31 14:31:36 hi booxter Aug 31 14:35:27 * kergoth ponders Aug 31 14:57:58 the checksum on asterisk changed, wtf Aug 31 15:08:33 people that change tarballs on servers should be shot Aug 31 15:09:02 you are too human Aug 31 15:09:12 :) Aug 31 15:10:02 that date changes Aug 31 15:10:10 and the sha1sum on the server is ok Aug 31 15:10:15 they frakking changed it Aug 31 15:10:24 no note in the dl dir Aug 31 15:10:26 or did they.. Aug 31 15:10:40 yeah Aug 31 15:10:42 maybe it's the NSA proxy that is broken. Aug 31 15:10:45 this used to work Aug 31 15:10:47 ROFL Aug 31 15:11:42 hm, maybe man in the middle?? :) Aug 31 15:11:51 or UFO guests? Aug 31 15:11:56 well, the checksum on the server is OK Aug 31 15:12:06 bu the man in the middle could be altering it Aug 31 15:12:25 Crofton, I had a broken proxy once fussing up and keeping an broken version of a tarball for ever Aug 31 15:12:40 (the RAM was full of unwanted entropy on that box) Aug 31 15:13:17 Crofton|work: maybe they repacked tarballs with new gzip parameters.. Aug 31 15:13:27 not all of them Aug 31 15:13:29 only some Aug 31 15:14:00 maybe they found a security hole and assumed it's better to fix it silently? ;)) Aug 31 15:14:39 or maybe someone found a security hole and fixed it for them? Aug 31 15:14:49 could be even both. Aug 31 15:14:51 hehe Aug 31 15:15:37 Crofton|work: maybe you should ask upstream for explanations. Maybe they really have security problems with their servers... Aug 31 15:15:55 in #asterisk .... Aug 31 15:16:00 hah. going through tickets for last couple of months for our own project. my boss managed to make one.. "(XXXd dies mysteriously before lunch damn it) created by foo" Aug 31 15:16:11 if not, you are free to punish them in any way you like most Aug 31 15:16:11 they replaced all the MoH with a different provider. Something about lawyers having a pissing contest or something like that. Aug 31 15:19:26 http://blogs.digium.com/2009/08/18/asterisk-music-on-hold-changes/ Aug 31 15:22:33 morning Aug 31 15:42:52 Crofton|work: do you know if the old wiki contents from the original design discussions about OE are still around in any form? Aug 31 15:47:13 kergoth: Laibsch can know more Aug 31 15:47:20 k Aug 31 15:47:42 i know we used to still have them somewhere. Aug 31 15:47:43 hmm Aug 31 15:53:51 bye Aug 31 15:56:30 * kergoth adds mvl6 to successstories Aug 31 15:57:06 yeah, that would probably qualify :-) Aug 31 15:58:20 03Evan Doiron  07org.openembedded.dev * r91c6116d7f 10openembedded.git/ (5 files in 3 dirs): Aug 31 15:58:20 add sscep: v0.0.2008.1211 Aug 31 15:58:20 http://patchwork.openembedded.org/patch/998/ Aug 31 15:58:20 Signed-off-by: Evan Doiron Aug 31 15:58:20 Signed-off-by: Michael Smith Aug 31 15:58:22 03Evan Doiron  07org.openembedded.dev * r199c29075e 10openembedded.git/ (3 files in 3 dirs): Aug 31 15:58:27 net-snmp: add v5.4.2.1 Aug 31 15:58:29 Signed-off-by: Evan Doiron Aug 31 15:58:31 Signed-off-by: Michael Smith Aug 31 15:58:35 03Evan Doiron  07org.openembedded.dev * r72834ed878 10openembedded.git/ (2 files in 2 dirs): Aug 31 15:58:38 usb-modeswitch: add v1.0.2 Aug 31 15:58:40 Signed-off-by: Evan Doiron Aug 31 15:58:42 Signed-off-by: Michael Smith Aug 31 15:58:46 Acked-by: Khem Raj Aug 31 16:05:47 I'm thinking that MV's usage is quite different from most. Since they actually ship bitbake/oe and are subject to their customers, who actually use it, MV will probably do more driving of improvements to the tools/infrastructure than most, since it's more than just themselves who would benefit from it. Aug 31 16:39:59 03Koen Kooi  07org.openembedded.dev * r0ef4a105d0 10openembedded.git/ (4 files in 2 dirs): python: fix -netserver and -devel packaging, convert 2.6.1 to inc_pr Aug 31 16:40:00 03Koen Kooi  07org.openembedded.dev * ref79d5bc59 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 31 16:48:49 hi all Aug 31 16:49:16 i'm looking on the mailing list about info about OE GA ... specially about the proxy document ... but i'm unable to find ;( Aug 31 17:08:07 kergoth, I do not know Aug 31 17:13:45 any opinion on adding oe_popen/oe_system wrappers around subprocess to base.bbclass, which pass along an environment that includes all exported vars from the datastore? Aug 31 17:13:55 then we could ditch the explicit PATH= everywhere Aug 31 17:15:39 hmm, what's the general consensus among python devs about the use of setuptools entry_points as a plugin / extensibility points mechanism? Aug 31 18:05:20 * kergoth ponders Aug 31 18:25:54 re Aug 31 18:26:13 wb Aug 31 18:59:11 kergoth: I support the wrappers for popen and prolly replace system calls Aug 31 18:59:17 with popen Aug 31 19:08:49 pb__: should I expect to build console-image with micro or is it too much ? I can certainly make it do that with some more stuff in distro conf Aug 31 19:26:59 khem: yes, it ought to work. Aug 31 19:27:47 console-image does pull in a pile of useless-seeming stuff, and we can probably do better than that for micro users, but I don't think there's any reason it shouldn't be buildable. Aug 31 19:28:22 mickeyl: thanks for your email, that's exactly what I needed. Aug 31 19:29:58 ok, cool. perhaps we'll find some more along those lines Aug 31 19:31:15 * pb_ stabs debian Aug 31 19:31:40 stupid x seems to stop working almost every time I upgrade Aug 31 19:31:44 pb_: ok then I guess micro will need some love Aug 31 19:32:02 khem: what goes wrong for you? I built console-image for qemuarm a couple of days ago and I don't remember it being very traumatic. Aug 31 19:32:22 pb_: I wonder what does _ signify in your nick :) sometimes there are more than one Aug 31 19:32:34 well, it was fairly traumatic watching the list of things that got compiled, but I don't remember needing to take much action. Aug 31 19:32:43 pb_: I have uclibc Aug 31 19:32:48 khem: it's my tail: sometimes I wind it out, and sometimes I reel it in. Aug 31 19:32:52 pb_: so its micro-uclibc Aug 31 19:32:58 pb_: lol Aug 31 19:33:12 ah, right, I was using glibc Aug 31 19:33:18 my bitchX completion of nicks show the tail nicely Aug 31 19:33:34 I don't think I've ever tried a micro-uclibc build of console-image. Aug 31 19:33:51 * kergoth wonders how best to integrate configobj (or similar) with optparse Aug 31 19:34:54 mickeyl: Can you ack my patch for minimal distro posted here http://patchwork.openembedded.org/patch/1010/ Aug 31 19:36:52 khem: sent Aug 31 19:37:10 mickeyl: danke Aug 31 19:37:15 :) Aug 31 19:37:16 pb_: http://pastey.net/124403 Aug 31 19:37:28 re Aug 31 19:37:32 this is the cause of my error down the line Aug 31 19:39:34 what's the actual error? Aug 31 19:39:57 none of those (except libx11-trim, which is death) look like they ought to cause an obvious build failure Aug 31 19:41:24 obviously if oe ends up picking avahi-python when you wanted avahi, you might end up with an inadvertent dependency on python that you would have preferred to avoid, but it shouldn't get much worse than that. Aug 31 19:42:53 khem: http://kergoth.pastey.net/124404 - how's something like that look? Aug 31 19:42:56 if there are packages masquerading as providers of a virtual that they don't actually implement correctly, I would prefer to take them out and burn them (or fix them, if appropriate) individually rather than working around it with distro config options. Aug 31 19:45:27 kergoth: looks good, though I am a bit nervous about the "only set preexec_fn if it wasn't already" semantics. Aug 31 19:45:37 true Aug 31 19:46:05 seems like that is an open manhole waiting for someone to step in it. Aug 31 19:46:06 03Philip Balister  07org.openembedded.dev * rb0e97c1ea8 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 31 19:46:17 03Philip Balister  07org.openembedded.dev * r122b402e9a 10openembedded.git/ (conf/checksums.ini recipes/asterisk/asterisk_1.4.23.1.bb): (log message trimmed) Aug 31 19:46:17 asterisk-1.4.23.1 : Checksum changed on source package. Aug 31 19:46:17 * Updated checksum to new values. Aug 31 19:46:17 * Dropped sounds.xml.patch so recipe builds. Aug 31 19:46:19 The Asterisk people changed the source tarballs for legal reasons. This Aug 31 19:46:21 blog entry explains the music-on-hold changes: Aug 31 19:46:23 http://blogs.digium.com/2009/08/18/asterisk-music-on-hold-changes/ Aug 31 19:46:41 I think you should either always set preexec_fn (and, effectively, disallow the user to pass it in) or implement a stacking system so that you can bolt the SIGPIPE thing onto whatever other function the user wants to get called. Aug 31 19:47:10 yeah, that would be better, indeed Aug 31 19:47:12 if the user really, truly wants to avoid the sigpipe uncatching then they can always call subprocess.Popen() directly. Aug 31 19:47:41 really, i think this should go into bitbake, there are parts of the bb package that do the same forking of external processes without the preexec fix Aug 31 19:47:56 but then you have to deal with bitbake version dependency and all, so this will do.. Aug 31 19:48:10 stupid lawyers Aug 31 19:48:22 yeah, quite. can always merge it into bitbake later, and then drop the local copy from oe at some future date when a new bitbake gets released. Aug 31 19:48:31 yeah Aug 31 19:50:18 I'm curious why asterisk guys didn't release new revision of their software for music-on-hold changes :-\ Aug 31 19:51:05 I think they had to remove stuff from old releases Aug 31 19:51:12 but yeah Aug 31 19:51:24 they could delete old files from their servers and put updated Aug 31 19:51:26 pb_: http://pastey.net/124406 is actual error it could be something else too as I see Aug 31 19:51:31 they could have removed the old tarball, and bumped the minor number Aug 31 19:52:02 exactly, that's what a good upstream should do Aug 31 19:57:03 kergoth: looks good to me preexec function could be overridden and set here as pb_ suggested Aug 31 20:22:27 ~curse imagemagick dumb 'developers' Aug 31 20:22:28 May the fleas of a thousand camels infest your most sensitive regions, imagemagick dumb 'developers' ! Aug 31 20:29:13 khem: ah, that's the old problem with bitbake not understanding versioned depends. Aug 31 20:30:05 which package had that error? Aug 31 20:39:45 pb_: hi Aug 31 20:40:07 hi ant__ Aug 31 20:40:18 pb_: about atomizing mtd-utils, you mean a -base package andan -extra ? Aug 31 20:40:41 no, I was thinking more of splitting into functional subsets. Aug 31 20:40:49 jffs2-utils, ubifs-utils, ... Aug 31 20:40:54 I see Aug 31 20:41:14 I use only flash_eraseall and nandwrite Aug 31 20:41:49 or better, I'd like to ship just these two binaries Aug 31 20:42:17 yeah, those should probably be packaged individually. Aug 31 20:43:00 pb_: are you skilled with X? Aug 31 20:43:21 I still don't understand why none of the X images work for qemu Aug 31 20:43:53 Laibsch: I wouldn't describe myself as an X-Man, though I do know something about it. Aug 31 20:44:01 what's the particular failure that you're seeing? Aug 31 20:44:10 and remind me, is this qemuarm or qemux86? Aug 31 20:44:23 both Aug 31 20:44:34 http://oss.leggewie.org/oe/X_in_qemu/ Aug 31 20:44:43 Those are the images I compiled a while ago Aug 31 20:44:50 particular failure? Aug 31 20:44:53 I couldn't say Aug 31 20:44:58 X just doesn't start Aug 31 20:45:10 And I don't see a way to understand why Aug 31 20:46:28 can you get access to a shell prompt? Aug 31 20:46:33 no Aug 31 20:46:42 I can boot the shell console image Aug 31 20:46:42 without xorg.log Aug 31 20:46:44 no chance Aug 31 20:47:07 I'm not very skilled with qemu, either Aug 31 20:47:11 how about if you boot with 'init=/bin/sh' or some such and then adjust the initscripts to stop it trying to start x? Aug 31 20:47:50 I wrote down a few of the things I learned by poking around at http://wiki.openembedded.net/index.php/Testing_with_QEMU Aug 31 20:48:10 pb_: how do I make that adjustment? Aug 31 20:48:14 I'm downloading the arm images now. I'll have a look and see if I can figure out what's up. Aug 31 20:48:21 that would be really nice Aug 31 20:48:25 Thank you Aug 31 20:48:49 Laibsch: probably just remove one of the symlinks in /etc/rc2.d/ Aug 31 20:49:07 or whatever your default runlevel is, might be 5. Aug 31 20:51:12 looks like the download will take a while, my connection to that site seems to be slow. Aug 31 20:52:06 yes, my upstream is very limited Aug 31 20:52:11 but stable Aug 31 20:54:03 righto, no problem Aug 31 20:55:09 brb Aug 31 21:01:02 You let irc be optional ? that's the problem :) Aug 31 21:01:04 ga Aug 31 21:04:51 heh, never any need for fortune(6) when Tartarus is around. Aug 31 21:05:21 heh Aug 31 21:06:57 glad to be of some help still :) Aug 31 21:07:16 Tartarus: how is the big merger going? Aug 31 21:08:01 still going :) Aug 31 21:09:20 Tartarus: good deal :-) I should look Matt P. up again one of these days. We live in the same town (relatively) Aug 31 21:10:04 heh. merger has him quite busy :) Aug 31 21:13:30 03Khem Raj  07org.openembedded.dev * r6cf6b423d9 10openembedded.git/conf/distro/ (micro-uclibc.conf minimal-uclibc.conf): (log message trimmed) Aug 31 21:13:30 micro-uclibc, minimal-uclibc: Reorder including the base distro config. Aug 31 21:13:30 A lot of stuff in these conf's are already set and Aug 31 21:13:30 is duplicated. This is removed. The base conf file Aug 31 21:13:32 should be included at the end after assigning LIBC Aug 31 21:13:34 so that correct toolchain configuration is picked. Aug 31 21:13:36 Signed-off-by: Khem Raj Aug 31 21:15:14 * kergoth mulls over possibilities for changing how configuration metadata is handled Aug 31 21:16:33 eeks Aug 31 21:16:53 bitbake gtk+ pulls in some iphone sdk here Aug 31 21:18:30 DEPENDS gone wild Aug 31 21:25:15 florian: yah, that is annoying. it happened to me yesterday. Aug 31 21:25:23 I think iphone-sdk should probably be default_preference -1 Aug 31 21:25:37 pb_: any idea what pulls this in? Aug 31 21:25:48 or, if I had my way, -99999999999999999, but I would compromise on -1 Aug 31 21:26:01 heh Aug 31 21:26:37 florian: I don't remember exactly. It PROVIDES some core package and gets selected by mistake under some (obscure) set of circumstances. Aug 31 21:27:18 hmm... bitbake doesn't have some kind of "--whatpullsin" switch? ;) Aug 31 21:27:39 florian -DDD Aug 31 21:27:51 woglinde: -DDD is... eh... messy Aug 31 21:27:52 florian: with debugging enabled it should say "selecting crazy-iphone-sdk to satisfy some-virtual" Aug 31 21:28:19 some combination of -D and grep usually yields the answer. or -g probably works too. Aug 31 21:29:04 ok, so i should take a look Aug 31 21:29:28 florian: you can bump debugging for specific messaging domains Aug 31 21:29:51 -l Depends -l Depends -l Depends, perhaps Aug 31 21:30:00 see lib/bb/msg.py's domain enum Aug 31 21:30:19 perhaps -l TaskData to see why it's selecting providers Aug 31 21:30:25 each -l bumps the level by one for that domain Aug 31 21:30:30 -DDD bumps it by 3 for all Aug 31 21:30:40 cute :) Aug 31 21:30:44 let's try this Aug 31 21:30:50 aye, it works well for trimming down the output Aug 31 21:31:23 of course, a large portion of the messages emitted go to the default domain, and there's no -l to show the default domain specifically, so it's either all or nothing for those :) Aug 31 21:31:45 but for your case, i'm sure -l will help Aug 31 21:32:36 * kergoth never used it, until he was making his mvl6 troubleshooting presentation Aug 31 21:33:15 this reminds me... the sdks don't work properly for eglibc Aug 31 21:33:43 any eglibc guru here? :) Aug 31 21:34:00 florian khem is the guru Aug 31 21:34:14 ah right Aug 31 21:34:38 pb_: this is xorg-lib/libxext_1.0.5.bb Aug 31 21:34:50 he khem *g* Aug 31 21:35:03 ...and Tartarus the toolchain guru :) Aug 31 21:36:28 khem, Tartarus: glibc-thread-db is guilty - it pulls the glibc dependency into task-sdk-bare Aug 31 21:36:32 florian: what was specific problem with sdk Aug 31 21:37:19 and glibc-dev into task-sdk-base, but this is trivial to fix Aug 31 21:40:19 pb_: virtual/libconv Aug 31 21:41:41 hm right khem removed the setting Aug 31 21:41:49 for minimal Aug 31 21:41:54 or micro Aug 31 21:41:59 both Aug 31 21:42:30 libiconv gets chosen based on your libc now Aug 31 21:43:02 khem: hmm... when? Aug 31 21:43:41 could some ack this patch -> http://patchwork.openembedded.org/patch/1007/? Aug 31 21:43:44 woglinde: can you look into the uclibc patch I posted http://patchwork.openembedded.org/patch/1012/ Aug 31 21:44:02 florian: recently may be on saturday Aug 31 21:44:14 khem hm yes saw this patch Aug 31 21:44:38 khem: strange... eglibc is in the list of possible providers. Aug 31 21:44:38 khem didnt I ack this? Aug 31 21:44:58 not yet :) bless it Aug 31 21:45:09 khem is an ack here enough? Aug 31 21:45:27 florian hm Aug 31 21:45:36 for me yes but may be for records patchwork ack is better Aug 31 21:45:48 florian I only had strange thinks with RDEPENDS an virtuals Aug 31 21:46:00 florian is it bases on RDEPENDS? Aug 31 21:46:50 florian: yes eglibc provided all packages that glibc provides Aug 31 21:47:06 so if there is contention it will raise its hand too Aug 31 21:47:12 woglinde: it looks like this patch puts the x264-presets into the main package if x264 is enabled? Aug 31 21:47:20 is there any situation where somebody would want x264 in ffmpeg, but not the presets? Aug 31 21:47:52 how can I avoid these merge messages getting push into logs upstream Aug 31 21:48:15 khem: instead of git pull, do git fetch ; git rebase origin/org.openembedded.dev Aug 31 21:48:30 that will pop your changes to the top of the stack instead of doing a real merge Aug 31 21:48:57 msmith_??? nope the patch pulls the presets only when you compile against libx264 Aug 31 21:49:06 which we dont do in oe Aug 31 21:49:15 root Aug 31 21:49:15 msmith_: ah yes Aug 31 21:49:17 oops Aug 31 21:49:19 woglinde: not sure... i din't yet know where it starts... but i think its a normal dependency Aug 31 21:49:36 koen invented this package because he was not fine about my solution deleting this files Aug 31 21:49:39 * khem was waiting for pb_ typing in passwd next Aug 31 21:49:51 woglinde: so the recipe as it was before the patch, doesn't build? Or just always puts the presets in the ffmpeg-x264-presets package? what's wrong with leaving them in that package? Aug 31 21:50:19 i once lost a root password to irc.. then while changing it, the mouse rolled back into the irc window and i lost the replacement too :P Aug 31 21:50:27 heh Aug 31 21:50:30 khem: do we have a virtual provider to get rid of glibc-thread-db in the sdk? Aug 31 21:50:57 msmith_ nope the intesions was to fix the QA issue because of not shipping the presets in any package Aug 31 21:51:32 git branch -a shows remotes/origin/org.openembedded.dev and remotes/origin/origin/org.openembedded.dev Aug 31 21:51:39 whats the difference Aug 31 21:51:56 florian: I have to look Aug 31 21:52:04 msmith_ my first solution was to delete the files Aug 31 21:52:31 but as I said koen made a useless package to fix this QA-warning Aug 31 21:52:47 Laibsch: it looks like the x startup scripts in your x11-image are expecting kdrive, but you have xorg. Aug 31 21:52:59 woglinde: yeah, seems either way works... i'm just thinking, let's say someday I want to build x264 support into ffmpeg, but i don't need these presets... then with koen's solution i still don't need them, they're in package i don't install. Aug 31 21:53:16 khem: you can do git show remotes/origin/org.openembedded.dev ; git show remotes/origin/origin/org.openembedded.dev and see if they're both being kept up to date Aug 31 21:53:24 if not you can delete the old one Aug 31 21:53:41 Laibsch: so, /etc/X11/Xserver tries to do 'exec Xorg -br -pn -rgba vrgb', and xorg chokes on -rgba. Aug 31 21:53:55 I suspect "vrgb" is wrong for qemu anyway, but that's kind of beside the point. Aug 31 21:54:55 hrm... xorg seems to be a step backwards on quite some platforms Aug 31 21:55:13 git show remotes/origin/origin/org.openembedded.dev stopped updating after Jun 12 Aug 31 21:55:36 yeah, I'm not quite sure what the perceived advantage is of using xorg if all you have is a flat framebuffer. Aug 31 21:55:41 remotes/origin/org.openembedded.dev is current Aug 31 21:56:24 msmith_: Did you try compiling xserver-xorg? I reported compile problems a few days ago. Aug 31 21:56:49 pb_: thank you for the information Aug 31 21:56:52 Laibsch: turns out that the easiest way to gain a shell prompt is "sudo mount x11-image-qemuarm.ext2 /mnt/image -o loop", then go to /mnt/image/etc and rm rcS.d/S01psplash rc5.d/S99gpe-dm Aug 31 21:56:57 I wonder what is the best way to deal with that Aug 31 21:57:07 pb_: maybe because people need something to fill ram and flash and do not like to have screen rotation ;-) Aug 31 21:57:14 also need to vi passwd and remove the * line from root in order to be able to log in Aug 31 21:57:15 Laibsch: yes, i have compiled xserver-xorg Aug 31 21:57:42 Laibsch: the easiest thing is probably just to switch to kdrive for those images. I don't think you would lose anything by doing that. Aug 31 21:57:58 alternatively, I think hrw|gone has been working on teaching the scripts to play better with xorg. Aug 31 21:58:05 florian might know more about the status of that. Aug 31 21:58:07 where is that set? Aug 31 21:58:19 something preferred_provider? Aug 31 21:58:26 yeah, virtual/xserver probably Aug 31 21:59:03 and/or possibly ${XSERVER} depending on how the images are set up. Aug 31 22:00:15 ok Aug 31 22:00:29 there is still some awkward settings about that Aug 31 22:01:02 I started working on it (I think that should be a machine specific setting and not so much an image setting like it currently often is) Aug 31 22:01:40 I wasn't able to finish it and dropped the ball because a "friendly" person interfered. Aug 31 22:01:58 Apparently nobody else picked up the work to straighten this out Aug 31 22:03:53 BTW, I think there is good reason to switch to xorg: kdrive is dead upstream Aug 31 22:04:56 hum well... if it works and does its job? Aug 31 22:05:18 does anyone knows how to just change the image when linux is starting... Aug 31 22:05:56 i mean when the OS is starting Aug 31 22:06:28 msmith_: It still fails for me: http://tinderbox.openembedded.net/builds/27271/ Aug 31 22:07:36 hmm. i built for x86 Aug 31 22:07:56 and v1.6.1 Aug 31 22:08:05 * * OE Bug 5279 has been RESOLVED (WONTFIX) by pascal_kesseli(AT)hotmail.com Aug 31 22:08:07 * * glibc 2.2.5 : "can't resolve '_GLBOAL_OFFSET_TABLE_' Aug 31 22:08:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5279 Aug 31 22:08:34 ssgo: you mean the tiny logo or the boot splash screen Aug 31 22:08:36 ? Aug 31 22:08:42 Laibsch: is there any way to see the configure log on this tinderbox thingy? Aug 31 22:08:52 i mean the boot splash screen Aug 31 22:08:54 i want to check if it passed --enable-glx-tls or not Aug 31 22:10:20 msmith_: Click on "failed" and then it's right there Aug 31 22:10:27 The do_compile log is there, yes Aug 31 22:10:29 but not the do_configure Aug 31 22:10:41 OK Aug 31 22:10:45 wait a minute Aug 31 22:11:13 ssgo: that's compiled into the psplash application. the recipes/psplash dir has some examples how to customize it and the necessary tool to generate the header file is mentioned in one of the bb files Aug 31 22:11:40 Laibsch: Oh also which libc are you using Aug 31 22:11:59 if it's not uclibc, then it definitely passed --enable-glx-tls.. is that ok for arm? Aug 31 22:13:43 msmith_: http://paste.debian.net/45422/ Aug 31 22:14:05 * * OE Bug 5315 has been created by pascal_kesseli(AT)hotmail.com Aug 31 22:14:07 * * helloworld-image fails on 'opkg_unarchive.patch' Aug 31 22:14:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5315 Aug 31 22:14:11 msmith_: I believe minimal is using eglibc now after khem changed that Aug 31 22:14:45 ok.. So is there any reason why eglibc on arm wouldn't have pthread_once()? Aug 31 22:16:27 no idea Aug 31 22:17:54 * kergoth tries to go through his OpenEmbedded task list to figure out what he should work on next on his own time Aug 31 22:19:11 Laibsch: can you paste up the do_configure log for your eglibc? to check if it has tls support Aug 31 22:20:56 http://paste.debian.net/45424/ Aug 31 22:21:11 checking for ARM TLS support... yes Aug 31 22:21:17 ok.. so that's fine Aug 31 22:21:36 msmith_ only when you have the updated tls.m4 macro Aug 31 22:21:38 msmith_: there are two patches I gave links to Laibsch other day which are needed Aug 31 22:21:47 otherwise you can get false results Aug 31 22:22:11 * khem searches the links Aug 31 22:23:24 khem: remember, those patches did not apply cleanly Aug 31 22:24:31 http://cgit.freedesktop.org/xorg/xserver/commit/?id=ae38151ddda9984effca5bb7c582540061201dce Aug 31 22:24:49 http://cgit.freedesktop.org/xorg/xserver/commit/?id=efb723e166e5fa89e90c7b400fb4c7979b1f50c0 Aug 31 22:24:55 yes they might need porting Aug 31 22:25:01 or some other patches on top Aug 31 22:25:05 or below Aug 31 22:25:22 yeah it was on 08/27 logs tell me Aug 31 22:25:37 it seems v1.5.3 isn't properly linking against -lpthread when --enable-glx-tls Aug 31 22:26:02 but 1.6.1 is Aug 31 22:26:08 trivial patches Aug 31 22:26:14 presumably.. because these patches were applied Aug 31 22:26:47 what we could do is move the "inherit glx-use-tls" from xorg-xserver-common to xserver-xorg 1.6.1 Aug 31 22:27:19 anything that gets Laibsch build back to normal Aug 31 22:27:41 xserver should be using tls.m4 too Aug 31 22:28:02 so we do not hassle about it with python foo Aug 31 22:28:23 woglinde: i'm not familiar with tls.m4 but yeah, xserver and mesa-dri just have to agree on tls or no Aug 31 22:29:41 let me see if I can provide something Aug 31 22:39:37 pb_: conf/distro/minimal.conf:XSERVER ?= "xserver-xorg xf86-input-evdev xf86-input-keyboard xf86-input-mouse xf86-video-fbdev" Aug 31 22:39:46 that is probably what needs to be changed Aug 31 22:40:07 but I'm still not sure if moving away from xorg back to kdrive is the best option Aug 31 22:49:13 03Khem Raj  07org.openembedded.dev * r3c1aac7f71 10openembedded.git/recipes/uclibc/ (8 files): (log message trimmed) Aug 31 22:49:13 uclibc: Pass ARCH=" on make commandline. Aug 31 22:49:13 Like linux kernel Git version of uclibc mandates that ARCH is passed Aug 31 22:49:13 down from make instead of being computed magically. This will be Aug 31 22:49:13 there in next release 0.9.31 of uclibc. Defining it on make Aug 31 22:49:15 invocation for older releases is no harm. Aug 31 22:49:19 Also move uclibc to use INC_PR Aug 31 22:52:34 03Graham Gower  07org.openembedded.dev * r5e51870a3e 10openembedded.git/recipes/netsurf/ (5 files in 2 dirs): Aug 31 22:52:34 netsurf: fixes for GCC-4.4, from netsurf svn. Aug 31 22:52:34 Netsurf appears to be statically linked with these libraries, so I Aug 31 22:52:34 bumped its PR too. Aug 31 22:52:34 Signed-off-by: Graham Gower Aug 31 22:52:36 Signed-off-by: Khem Raj Aug 31 22:53:16 all: need help with omap gfx .bb (libgles-omap3_xxx.bb). anyone available? Aug 31 22:53:33 jeaneus hm whats your porblem? Aug 31 22:54:57 I do not want to make GFX. I have a different GFX that I want to use that I am planning to place in the file system later. I have been trying to remove this or turn this off, but I don;t know where to do this. Aug 31 22:55:47 jeaneus: preferred provider for qt4x11 Aug 31 22:56:02 Laibsch nope Aug 31 22:56:16 you can have gles without qt Aug 31 22:56:38 FYI: I am somewhat new to oe and bitbake so I am learning my way around.... Aug 31 22:56:51 jeaneus what you exactly building? Aug 31 22:57:04 woglinde: my understanding is that gles is unwanted here Aug 31 22:57:12 laibsch right Aug 31 22:57:27 but we have to find out by which stuff it is draged in Aug 31 22:58:47 03Ming Chow 周明  07org.openembedded.dev * rccd18852f3 10openembedded.git/ (7 files in 3 dirs): (log message trimmed) Aug 31 22:58:47 compcache: Add version 0.5+0.6pre3 and version 0.6 Aug 31 22:58:47 * compcache creates RAM based block device (named ramzswap) which acts as swap disk. Pages swapped to this disk are compressed and stored in memory itself. Aug 31 22:58:49 * Compressing pages and keeping them in RAM virtually increases its capacity. This allows more applications to fit in given amount of memory. Aug 31 22:58:52 * added modules_install target in top Makefile to fit the automatically modules build process Aug 31 22:58:54 * added KERNELDIR variable to make 2.6 style KBUILD system work Aug 31 22:58:56 * modified the Makefile for rzscontrol tool to use cross toolchain instead of local gcc Aug 31 22:59:20 woglinde: I have a gpe .bb that build X11 then build some other apps for the target. I do not want gles to be included in the build also I can care less for the kernel. I can load a newer version with a different kernel later on. Aug 31 23:00:14 jeaneus you build your own image.bb? Aug 31 23:01:04 woglinde: I would like just to build oe using angstrom for a zoom2 (unstable branch) with no GFX support, since GFX seems to based on beagle (for now).... Aug 31 23:01:51 jeaneus again please show your bitbake line Aug 31 23:02:50 03Florian Boor  07org.openembedded.dev * ra8b95ab31b 10openembedded.git/recipes/xserver-common/xserver-common_1.31.bb: xserver-common: add 1.31 Aug 31 23:02:51 03Florian Boor  07org.openembedded.dev * rbc3852e7f0 10openembedded.git/recipes/tasks/task-sdk-base.bb: task-sdk-base: Use virtual libc development package to use the selected libc instead of pulling in glibc. Aug 31 23:02:52 03Florian Boor  07org.openembedded.dev * r2695f790ec 10openembedded.git/recipes/gtkhtml/gtkhtml-3.8_3.18.0.bb: gtkhtml-3.8: Fix staging of 3.18.0 Aug 31 23:05:17 is it a custom: bitbake gpe-zoom2.bb that Aug 31 23:06:42 hmm...a zoom2 for gpe development - cool :) Aug 31 23:06:45 ah then show us the content Aug 31 23:08:26 good night Aug 31 23:08:34 nite florian Aug 31 23:08:36 give me min.... Aug 31 23:08:48 03Ming Chow 周明  07org.openembedded.dev * r40dd422071 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Aug 31 23:08:48 linux-nokia800-2.6.21-osso71.bb: Fixed undefined reference to __aeabi_uldivmod Aug 31 23:08:48 * This was a problem when using gcc-4.3 compile linux-nokia800-2.6.21-osso71 Aug 31 23:08:48 Signed-off-by: Ming Zhou Aug 31 23:08:50 Signed-off-by: Khem Raj Aug 31 23:10:22 khem do you update patchwork too? Aug 31 23:11:41 yes Aug 31 23:13:06 XSERVER ?= "xserver-xorg \ Aug 31 23:13:06 xf86-input-evdev \ Aug 31 23:13:06 xf86-input-tslib \ Aug 31 23:13:06 xf86-input-mouse \ Aug 31 23:13:06 xf86-video-fbdev \ Aug 31 23:13:08 xf86-input-keyboard" Aug 31 23:13:12 ECONFIG ?= "places e-wm-config-angstrom e-wm-config-default" Aug 31 23:13:14 ECONFIG ?= "places e-wm-config-angstrom e-wm-config-default" Aug 31 23:13:18 EXTRA_ZOOM2_STUFF += "\ Aug 31 23:13:20 EXTRA_ZOOM2_STUFF += "\ Aug 31 23:13:22 dbus \ Aug 31 23:13:24 dbus-glib \ Aug 31 23:13:26 dbus-x11 \ Aug 31 23:13:28 dbus-lib \ Aug 31 23:13:29 args Aug 31 23:13:30 gst-meta-base \ Aug 31 23:13:32 gst-meta-audio \ Aug 31 23:13:34 gst-meta-video \ Aug 31 23:13:34 ~pastebin Aug 31 23:13:35 [~pastebin] A "pastebin" is a web-based service where you can paste anything over 3 lines without flooding the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://www.rafb.net/paste Aug 31 23:13:36 xf86-input-tslib \ Aug 31 23:13:38 mime-support e-wm ${ECONFIG} exhibit \ Aug 31 23:13:40 tslib-conf tslib-tests tslib-calibrate \ Aug 31 23:13:42 xmodmap \ Aug 31 23:13:44 xev \ Aug 31 23:13:46 pointercal \ Aug 31 23:13:48 ogre-egl \ Aug 31 23:13:50 ogre-egl-samples" Aug 31 23:13:52 Aug 31 23:13:56 DEPENDS = "${MACHINE_TASK_PROVIDER} \ Aug 31 23:13:58 DEPENDS = "${MACHINE_TASK_PROVIDER} \ Aug 31 23:14:00 virtual/xserver" Aug 31 23:14:06 IMAGE_INSTALL = "\ Aug 31 23:14:08 IMAGE_INSTALL = "\ Aug 31 23:14:10 task-proper-tools \ Aug 31 23:14:12 task-base-extended \ Aug 31 23:14:14 angstrom-zeroconf-audio \ Aug 31 23:14:16 angstrom-led-config \ Aug 31 23:14:18 ${MACHINE_TASK_PROVIDER} \ Aug 31 23:14:20 openssh-scp openssh-ssh \ Aug 31 23:14:22 ${XSERVER} \ Aug 31 23:14:24 ${EXTRA_ZOOM2_STUFF}" Aug 31 23:14:28 inherit image Aug 31 23:14:30 inherit image Aug 31 23:14:32 Sorry about that.... Aug 31 23:14:37 ~pastebin Aug 31 23:14:38 [~pastebin] A "pastebin" is a web-based service where you can paste anything over 3 lines without flooding the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://www.rafb.net/paste Aug 31 23:16:31 jeaneus do you need the ogre stuff? Aug 31 23:16:36 http://pastebin.com/mc8f97f1 Aug 31 23:16:42 thats where gles comes ine Aug 31 23:16:54 yes... Aug 31 23:17:14 how should this work? Aug 31 23:17:24 you cannt compile the stuff with gl support Aug 31 23:19:36 okay then the only option is to include the GFX from a local dir and add it prior to build ogre? Aug 31 23:20:50 hm Aug 31 23:21:00 yes Aug 31 23:21:17 but I dont know if ASSUME_PROVIDED works on virtual package Aug 31 23:23:24 I was able to use this an build a beagle env with ogre on angstrom so I assumed it would work with zoom2 Aug 31 23:24:35 I can adjust libgles-omap3xxx.bb to point to a local GFX package then. Aug 31 23:25:07 nope Aug 31 23:25:10 the way is Aug 31 23:25:15 provide somewhere Aug 31 23:25:35 and then try the ASSUME_PROVIDED line in your local.conf Aug 31 23:26:39 ASSUME_PROVIDED += "virtual/egl" Aug 31 23:29:32 okay... I'll have another go at it.... Aug 31 23:29:46 woglinde: Thanks for the input... Sep 01 00:15:04 my ide slc dom arrived today Sep 01 00:15:07 quite small Sep 01 00:15:29 innodisk edc 8000 2gb Sep 01 02:44:47 is there a way i can populate /dev with udev/mdev after initial boot? Sep 01 02:49:59 m4t, as opposed to? Sep 01 02:50:41 you could create a static /dev and run udev/mdev at any point after boot. Sep 01 02:50:51 erm, okay well all i can fit into this initramfs is micro-image Sep 01 02:51:05 you want a static /dev Sep 01 02:51:06 i am limited to transferring files via netcat currently Sep 01 02:51:12 :) Sep 01 02:51:17 okay, is there a good way to populate it? Sep 01 02:51:32 my disk-on-module is showing up as hde and i'm not feeling right, to do some major/minor math Sep 01 02:51:57 openembedded/files/device_table-minimal.txt Sep 01 02:52:19 oh cool Sep 01 02:52:20 thanks Sep 01 02:52:25 hde: InnoDisk Corp. - EDC8000 2GB, ATA DISK drive Sep 01 02:52:26 btw :) Sep 01 02:52:32 finally get to try to bootstrap this Sep 01 02:53:23 /dev/sde b 664 0 0 8 64 - -- Sep 01 02:53:26 is what i needed Sep 01 02:53:34 image.bblcass looks at IMAGE_DEVICE_TABLES to find the right device table. I assume for a micro-image it would try to use the minimal table. Sep 01 02:54:11 just make your own device table file and put it in an overlay, then point IMAGE_DEVICE_TABLES at your newly created file. Sep 01 02:54:17 alright Sep 01 02:54:24 actually thats sde, i need hde Sep 01 02:56:05 that is good to know though Sep 01 02:56:11 i think this system will be running udev Sep 01 02:57:15 having a usable static /dev is useful if you want to defer starting udev until after your GUI loads too. Sep 01 02:57:34 on my system starting udev takes ~15seconds Sep 01 02:57:47 no idea why its so slow **** ENDING LOGGING AT Tue Sep 01 02:59:57 2009