**** BEGIN LOGGING AT Tue Jul 10 02:59:57 2007 Jul 10 04:55:00 * * OE Bug 2628 has been created by winthrowe(AT)gmail.com Jul 10 04:55:02 * * No audio driver built for h5000 Jul 10 04:55:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2628 Jul 10 05:07:00 * * OE Bug 2629 has been created by winthrowe(AT)gmail.com Jul 10 05:07:02 * * No wireless driver built for h5000 Jul 10 05:07:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2629 Jul 10 07:39:00 * * OE Bug 2630 has been created by xjqian(AT)gmail.com Jul 10 07:39:02 * * screen rotation applet broken in Angstrom-x11-image-glibc-test-20070605-tosa Jul 10 07:39:03 * * http://bugs.openembedded.org/show_bug.cgi?id=2630 Jul 10 08:07:59 * * OE Bug 2631 has been created by utx(AT)penguin.cz Jul 10 08:08:01 * * X server returns inconsistent RandR on spitz/akita Jul 10 08:08:03 * * http://bugs.openembedded.org/show_bug.cgi?id=2631 Jul 10 08:20:29 hi Jul 10 08:22:39 hey hrw Jul 10 08:24:48 ERROR: ------------------- Jul 10 08:24:48 ERROR: glib-2.0-dev is listed in PACKAGES mutliple times, this leads to packaging errors. Jul 10 08:24:48 ERROR: Please fix the metadata/report this as bug to OE bugtracker. Jul 10 08:24:48 ERROR: ------------------- Jul 10 08:24:50 WTF Jul 10 08:26:26 who messed with glib.inc Jul 10 08:26:42 PACKAGES =+ "${PN}-dev glib-2.0-utils " Jul 10 08:26:47 kinda obvious bug there Jul 10 08:27:16 I fixed such one in libxosd yesterday Jul 10 08:27:47 XorA: mtn annotate is your friend Jul 10 08:27:59 * * OE Bug 2632 has been created by utx(AT)penguin.cz Jul 10 08:28:01 * * icewm is unusable in the default configuration Jul 10 08:28:03 * * http://bugs.openembedded.org/show_bug.cgi?id=2632 Jul 10 08:28:05 koen|inverness: debian doesnt have a new enough version of mtn :-) Jul 10 08:28:19 | | | | | | | | | | | | | | | | | | | | | | | | | Revision: e25c5e38674da9988b268e6af89a90ecbbe1dac6 Jul 10 08:28:22 | | | | | | | | | | | | | | | | | | | | | | | | | Ancestor: adc9d2b07147baa9165b930a857e39f5f57331a3 Jul 10 08:28:23 e25c5e38.. by pfalcon 2007-07-04: PACKAGES =+ "${PN}-dev glib-2.0-utils " Jul 10 08:28:25 | | | | | | | | | | | | | | | | | | | | | | | | | Author: pfalcon@openembedded.org Jul 10 08:28:28 | | | | | | | | | | | | | | | | | | | | | | | | | Date: 2007-07-04T22:19:29 Jul 10 08:28:36 ~lart mtn log defaults Jul 10 08:28:36 * ibot steals mtn log defaults's mojo Jul 10 08:29:11 XorA: 0.33 is in experimental, 0.35 on monotone site Jul 10 08:29:52 XorA: http://rafb.net/p/OSWpJX45.html Jul 10 08:31:03 pushed fix anyway Jul 10 08:32:14 we should randomely add INHERIT += "insane" all over distro/*.conf to make people obey the warnings Jul 10 08:32:29 so they cant find all the occurances to remove it Jul 10 08:33:05 and few times in base.bbclass, bitbake.conf? Jul 10 08:34:34 hrw: :-D Jul 10 08:34:58 ~lart openmoko uboot Jul 10 08:34:59 * ibot beats openmoko uboot severely about the head and shoulders with a rubber chicken Jul 10 08:35:07 heh heh, whats it broke? Jul 10 08:35:27 | Patch uboot-dfu.patch does not apply (enforce with -f) Jul 10 08:35:43 and their do_quilt task is sick Jul 10 08:36:04 'fetch patches, apply patches, remove them' Jul 10 08:36:14 hrw: yeah I complained about it in kernel, but they are pretty dead set on it Jul 10 08:36:15 lack of 'check does they applied' Jul 10 08:36:45 they should use git tree for own branch of kernel and uboot Jul 10 08:36:50 hello all Jul 10 08:38:28 * XorA shudders at git Jul 10 08:39:41 03xora 07org.oe.dev * r2f67b52c... 10/ (1 packages/glib-2.0/glib.inc): packages/glib-2.0/glib.inc : remove bogus ${PN}-dev from packages. Jul 10 08:39:47 03koen 07org.oe.dev * r5675a11f... 10/ (4 files in 4 dirs): e-wm: fix startup script, fix u-a and install xdg stuff (per http://wiki.enlightenment.org/index.php/E17_and_Efreet) Jul 10 08:40:08 koen|inverness: I was thinking that you are on vacations... Jul 10 08:40:23 XorA: wasn't the -dev in front of the packages statement to get the headers inside /lib into the correct package? Jul 10 08:40:44 hrw: right, and I'm waiting for the train to arrive Jul 10 08:40:53 koen|inverness: -dev is already added to PACKAGES by default Jul 10 08:41:20 but after ${PN}, right? Jul 10 08:41:34 koen|inverness: well then += is the wrong operation and whoever wrote it is a tit Jul 10 08:42:45 * koen|inverness ponders about making an xdg package for applications.menu to stop reinventing the wheel Jul 10 08:42:52 anyway, /me heads out Jul 10 08:43:14 koen|inverness: there is no explanation of what was trying to do in that commit message therefore it is bogus Jul 10 09:13:44 hey guys, I pulled HEAD about a week ago and have 1380 Jul 10 09:14:03 the fix proposed in the bug repport does work, has someone forgotten to commit ? Jul 10 09:14:09 xaiki_: I think your talking to the wrong project Jul 10 09:14:21 xaiki_: mtn heads arent neat numbers like 1380 Jul 10 09:15:07 XorA: I'm talking about the bug repport. Jul 10 09:15:14 BR # 1380 Jul 10 09:16:23 xaiki_: you made sure to do mtn pull; mtn update? Jul 10 09:17:03 haha I might have forgotten to pull, but anyway, my first pull was only last week. Jul 10 09:17:12 and the bug is like 6months old =) Jul 10 09:24:45 cool, I'm back from vacation, and OpenMoko is available! Jul 10 09:24:54 xaiki_: I get feeling the fix wasnt what it says in bug report Jul 10 09:26:21 XorA: yeah, it looks more like a b0rk3d toolchain to me... Jul 10 09:26:33 djay-il hehe Jul 10 09:26:34 xaiki_: uclibc works for loads of people Jul 10 09:27:19 XorA: I don't doubt it, but still, I pulled the mtn tree like described in the wiki, and launched a compilation changing only minimal things and it fails... Jul 10 09:27:42 xaiki whtas the problem? Jul 10 09:28:06 woglinde: exactly the same failure as #1380 Jul 10 09:29:20 xaiki your uclibc is at 0.9.29? Jul 10 09:29:50 you did mtn pull and mtn update? Jul 10 09:30:05 woglinde: I'm doing that now, but I cloned the MTN tree about a week ago. Jul 10 09:30:52 woglinde: I was building 0.9.28 Jul 10 09:31:41 why? Jul 10 09:31:54 it's what it chose automagically Jul 10 09:31:57 0.9.28 is borked in many ways Jul 10 09:32:04 0.9.29 ypu should use Jul 10 09:32:17 woglinde: mayb should move 0.9.28 to non-working then Jul 10 09:32:24 xora yes Jul 10 09:32:26 ok, I'll try that. Jul 10 09:32:30 are you guys OE upstream ? Jul 10 09:32:36 xaiki_: yes Jul 10 09:32:44 I have 2 trivial patches that could have their use in bitbake: Jul 10 09:32:46 xaiki for which platform do you compile? Jul 10 09:32:46 http://cxhome.ath.cx/~xaiki/bitbake/ Jul 10 09:33:04 hm bitbake is seperated from oe Jul 10 09:33:09 and I've been working on packaging BB for debian, and am thinking about wether it would have any use to be in the main archive. Jul 10 09:34:21 hm there is bb package already Jul 10 09:34:48 http://packages.debian.org/unstable/devel/bitbake Jul 10 09:36:22 ok ... I missed it ... so weird ... Jul 10 09:36:38 no problem Jul 10 09:52:05 hi wrobbie Jul 10 09:52:48 I hate bugzilla Jul 10 09:52:57 where is the button that show me all open bugs Jul 10 10:02:30 nowhere Jul 10 10:02:57 advanced search, OPEN, search with no keywords Jul 10 10:03:48 or simple search and just press search button Jul 10 10:08:43 hi hrw Jul 10 10:09:09 hrw noe needs keyword Jul 10 10:09:12 simple search Jul 10 10:09:21 so its a ui desgin error Jul 10 10:09:59 OE bugzilla needs love Jul 10 10:24:50 hi henryk Jul 10 10:33:46 hi all Jul 10 10:34:38 henryk I overlook this yesterday in the phoneme forum Jul 10 10:34:44 http://forums.java.net/jive/thread.jspa?threadID=21591&tstart=0 Jul 10 10:35:00 woglinde: i have it running now Jul 10 10:35:19 i hacked tslib support in yesterday and everything works just fine Jul 10 10:35:35 i have opera mini on the full 480x640 running, for example :) Jul 10 10:36:10 uh nice Jul 10 10:36:35 can you provide the patches Jul 10 10:36:40 henryk: can you chek one app for me? Jul 10 10:36:46 hrw: sure Jul 10 10:36:48 I would like to test it on the simpad Jul 10 10:37:01 only limitation: there is no keyboard on the phone ... Jul 10 10:37:03 woglinde: opera miini on 800x600? Jul 10 10:37:16 hrw no phoneme with touchscreen Jul 10 10:38:17 henryk: http://download.pasjagsm.pl/files/ginger-3-2-3-poznan-1441.zip Jul 10 10:40:26 hrw: runs, and apparently pops up a number input field. and a menu with a funny foreign language Jul 10 10:40:53 henryk polski Jul 10 10:41:10 psokolovsky: feel free to dissaprove my commit, Ive put my point in an email, but its not that important a point for me Jul 10 10:41:28 henryk: enter 5, select any line and it will list bus/tram times on bus/tram stop Jul 10 10:41:38 XorA: I have fix for it Jul 10 10:41:53 XorA: ok, thanks Jul 10 10:42:05 03florian 07org.oe.dev * r84e965db... 10/ (1 conf/machine/fic-gta01.conf): fic-gta01.conf: Update included modules and enable padding of jffs2 images. Jul 10 10:42:07 hrw: no keyboard :( Jul 10 10:42:15 henryk: argh.. Jul 10 10:42:21 hmm, or I could try bluetooth hid ... hmm. Jul 10 10:42:25 henryk: but god to know that it works. Jul 10 10:45:53 hmm. kernel "Alignment trap:" is that a fatal error or just a warning? Jul 10 10:46:12 hm Jul 10 10:46:23 if the program works you can ignore it Jul 10 10:47:03 henryk: but it would be nice to fix it as they are total performance killers Jul 10 10:47:17 http://netwinder.osuosl.org/users/b/brianbr/public_html/alignment.html Jul 10 10:47:33 xora yes right Jul 10 10:48:26 took me 1/2 an hour once to boot into X when freetype was generating them :-) Jul 10 10:48:28 ok, thanks Jul 10 10:48:41 xora ouch Jul 10 10:48:55 woglinde: it did convince me to fix freetype .bb :-) Jul 10 10:49:18 henryk you can try out the programm pahole Jul 10 10:49:26 maybee it helps you find the spot Jul 10 10:49:33 xora what was the fix? Jul 10 10:50:59 * * OE Bug 2633 has been created by  Jul 10 10:51:01 * * autoconf-2.61 in OE doesn't handle OE site properly Jul 10 10:51:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2633 Jul 10 10:51:27 woglinde: upgrade to a recent version of freetype + fix some X stuff that didnt like new freetype Jul 10 10:51:36 psokolovsky: autoconf 2.61 in oe is not used Jul 10 10:51:50 hrw: and in world, it is Jul 10 10:51:58 so, welcome to real world Jul 10 10:52:09 Its good to fix it up anyway... Jul 10 10:52:20 Someday we might change... Jul 10 10:52:34 someday Jul 10 10:52:34 somedoay we need to change Jul 10 10:52:38 xora oh I meant patched file ;) Jul 10 10:52:45 wait for next bug. they have changed already Jul 10 10:52:56 as the longer we leave it, the more our metadata deteriorates Jul 10 10:53:07 haha, great. I can now use a Sony Ericsson W800i as a Bluetooth HID 'keyboard' for the Neo1973 Jul 10 10:53:08 when I added autoconf 2.61 too much soft did not built Jul 10 10:53:33 woglinde: this was years ago Jul 10 10:53:37 henryk nice Jul 10 10:54:01 XorA: We do need to change, I agree, its just going to cause pain :/ Jul 10 10:54:20 RP: I did start once, but life got in the way Jul 10 10:54:38 xora hm okay Jul 10 10:56:39 XorA: I know the feeling Jul 10 10:57:31 03florian 07org.oe.dev * r59a848ba... 10/ (1 packages/tasks/task-gpephone.bb): task-gpephone: Add some tools to debug section. Jul 10 10:58:00 * * OE Bug 2634 has been created by  Jul 10 10:58:02 * * Newer software requires autconf 2.61 (e.g. parted 1.8.7) Jul 10 10:58:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2634 Jul 10 10:58:21 hrw: ^ Jul 10 10:58:58 I'm going to DEF_PREF it -1 now, but who can be sure which version will build well now? Jul 10 10:59:03 hrw: Sometime when one of us has a free moment, we should see how much of poky breaks with the new autoconf... Jul 10 10:59:30 Poky will certainly be easier than OE :/ Jul 10 10:59:58 hi rp Jul 10 11:00:01 btw. Jul 10 11:00:12 hi woglinde Jul 10 11:00:37 RP: one day Jul 10 11:01:01 hrw: Maybe after the next poky release Jul 10 11:02:45 henryk stephwm starte moko javca side maybe you can add some points too Jul 10 11:03:04 woglinde: pardon? Jul 10 11:03:32 http://wiki.openmoko.org/wiki/Java Jul 10 11:04:04 woglinde: sure Jul 10 11:04:32 java is really chaos Jul 10 11:04:44 to many people did their on thing Jul 10 11:04:46 *g* Jul 10 11:05:12 hrw: ok, now I have at least a numeric keyboard Jul 10 11:09:34 btw, fresh thought: gnu crowd really should drop autoconf together with GPL3 coming into force. somebody should have set such an ultimatum for them ;-E Jul 10 11:10:09 psokolovsky what to use instead? Jul 10 11:11:07 woglinde: cmakes, scons'es, and lot of other stuff to choose from. Jul 10 11:12:47 oh no Jul 10 11:12:57 they all have problems with cross compiling Jul 10 11:13:16 psokolovsky: whats up with automake and GPL3? Jul 10 11:13:24 xora nothing Jul 10 11:13:35 XorA: see above ;-) Jul 10 11:13:39 psokolovsky ranted only a bit Jul 10 11:13:48 hrw: http://www.ploetzli.ch/forumtst/DSC00960.JPG Jul 10 11:14:17 henryk: great! Jul 10 11:15:02 henryk today I will start put phoneme in oe Jul 10 11:15:29 psokolovsky: you do know there is a workaround for autoconf 2.61 requiring stuff? Jul 10 11:16:09 XorA: I bet there's. I could start with killing that 2.61 req. but why? Jul 10 11:16:26 woglinde: and yes, I know now that cross-compiling without autconf is a PITA. PhoneME apparently is comprised of several distinct project with different Makefile styles and to get it to compile I think I have used four or five different ways to tell it where the cross-compiler lives. Jul 10 11:16:32 henryk: looks ok, will have to bump font sizes and add Polish chars Jul 10 11:16:34 do_configure() { Jul 10 11:16:34 gnu-configize Jul 10 11:16:34 libtoolize --force Jul 10 11:16:34 oe_runconf Jul 10 11:16:34 } Jul 10 11:16:46 henryk I know Jul 10 11:16:49 03pfalcon 07org.oe.dev * r82e025b5... 10/ (1 packages/parted/parted_1.8.7.bb): Jul 10 11:16:49 parted 1.8.7: Set DEFAULT_PREFERENCE = "-1", requires autoconf 2.61 to be built properly. Jul 10 11:16:49 * Requires autoconf 2.61. Without it, will build, but there Jul 10 11:16:49 will be link errors when some other package will link against Jul 10 11:16:49 libparted. This is actually due to fact that autoconf 2.61 doesn't want to Jul 10 11:16:51 source OE site files. See #2633, $2634. Jul 10 11:16:54 I build it serval times Jul 10 11:17:00 parted requires 2.61 since 1.8.3. and 1.8.2 doesn't build because it's incompatible with contemporary linux headers. LOL! Jul 10 11:17:06 I pachted the makefiles to yues qte-mt Jul 10 11:17:10 use Jul 10 11:17:23 psokolovsky: tried sed -e /2.61/2.59/ configure.ac? Jul 10 11:17:34 nope Jul 10 11:17:38 then try Jul 10 11:17:47 many times it works Jul 10 11:18:11 I'd prefer there was way to just build that sw, there must be more and more everyday Jul 10 11:18:20 now lunch Jul 10 11:18:41 psokolovsky: see the do_configure I posted, that should meet the requirement of just build Jul 10 11:19:29 hmmm, maybe we should needs-new-autoconf.bbclass with that hack Jul 10 11:19:50 XorA: you don't want to believe that autoconf of lower version won't touch aclocal.m4, etc of forced newer version? Jul 10 11:20:15 psokolovsky: works in other stuff that demands 2.61 Jul 10 11:20:30 XorA: just one q - are you sure gnu-configize should go before libtoolize? Jul 10 11:20:43 psokolovsky: no idea, I didnt write the workaround Jul 10 11:21:09 psokolovsky: I am puzzled by your you don't want to believe comment Jul 10 11:21:17 XorA: that's what parted 1.8.7 already has. it builds, but it's broken. and those commands just skip it apparently Jul 10 11:21:43 XorA: what do you think I was doing this night? ;-) Jul 10 11:21:59 psokolovsky: if you override do_configure you dont call autoconf Jul 10 11:23:12 psokolovsky: you built m4 1.4.8 for tests of autoconf 2.61? Jul 10 11:23:12 XorA: well, if you have time, just try and see what happens there please. I'm going to try s/2.61/2.59/ Jul 10 11:23:23 hrw: nope Jul 10 11:27:09 XorA: "aclocal.m4:17: error: this file was generated for autoconf 2.61. Jul 10 11:27:09 You have another version of autoconf. If you want to use that, Jul 10 11:27:09 you should regenerate the build system entirely. Jul 10 11:27:09 " Jul 10 11:27:34 what about autoreconf ? Jul 10 11:27:39 so there're no wonders at that snippet doesn't really work Jul 10 11:28:02 polyonymous: read again: autocrap 2.61 is *required* Jul 10 11:28:19 and autocrap 2.61 doesn't source site files, actually that's where problem is Jul 10 11:28:48 Hmm... "you have another version... if you want to use that, you should regenerate" ? Jul 10 11:29:19 NOTE: package parted-1.8.7-r1: task do_compile: started Jul 10 11:30:00 NOTE: package parted-1.8.7: completed Jul 10 11:30:03 XorA: read again: it build well. just libparted contains refs to non-existent symbols. Jul 10 11:30:23 what symbols? Jul 10 11:31:59 polyonymous: it references rpl_malloc and stuff. and of course they don't really use all that autoconf as it was designed too, so there's no rpl_malloc Jul 10 11:33:37 is this crap in the tree now? Jul 10 11:34:39 polyonymous: parted 1.8.7 in tree Jul 10 11:35:00 and autoconf too? Jul 10 11:35:35 polyonymous: yes, but def_pref -1 Jul 10 11:36:05 ok, then I'd rather see for myself - better than any explanations :) Jul 10 11:38:05 so to narrow this down, parted itself compiles and works, its just libparted that has problems? Jul 10 11:40:43 because parted compiles and works here Jul 10 11:42:32 henryk how do you run phoneme with qte on the X server of the moko? Jul 10 11:45:23 woglinde: i stop the x server and run it directly on the framebuffer Jul 10 11:45:43 henryk ah okay Jul 10 11:45:55 XorA: yes, exactly, linking another package (I hack on gparted) against libparted leads to unresolved symbols Jul 10 11:46:22 and no, s/2.61/2.59/ doesn't help much Jul 10 11:48:47 he stefan Jul 10 11:48:57 hi woglinde Jul 10 11:49:07 you are now where? Jul 10 11:49:12 in scottland? Jul 10 11:49:39 woglinde: Amiens, RMLL http://www.rmll.info/?lang=en Jul 10 11:49:47 woglinde: .fr Jul 10 11:49:58 woglinde: England is next week. ;) Jul 10 11:50:20 oh okay Jul 10 11:50:26 I just got the wifi enterprise setup ready to have real internet and not only http Jul 10 11:51:01 hihi Jul 10 11:51:19 Giving a Open{EZX,Moko} talk in two hours Jul 10 11:51:28 ;) Jul 10 11:55:30 psokolovsky: I think I can fix your problems with a one line patch to configure file Jul 10 11:55:59 XorA: heh, cool, what is it? Jul 10 11:56:22 psokolovsky: line 12559 change no to yes Jul 10 11:56:30 else Jul 10 11:56:30 if test "$cross_compiling" = yes; then Jul 10 11:56:30 ac_cv_func_malloc_0_nonnull=yes Jul 10 11:56:30 else Jul 10 11:56:53 XorA: LOL! You won the prize for today ;-) Jul 10 11:57:36 checking for GNU libc compatible malloc... yes Jul 10 11:57:37 XorA: but the problem is actually that 2.61 doesn't source site files, you noticed? ;-) Jul 10 11:58:10 psokolovsky: 2.61 is busted, but with one line patch we can use what we have with the recipe you posted I think Jul 10 11:58:20 it will take some time but I will check that problem too Jul 10 11:58:45 psokolovsky: but we arent actually calling autoconf 2.61 because we overrode do_configure Jul 10 11:58:46 XorA: no, sorry, that doesn't really qualify as a dirty hack, imho Jul 10 11:59:06 woglinde: ok, I've pulled and updated, but it still selects uclibc 0.9.28, any hint on how to trace why it's this version that gets selected ? Jul 10 11:59:28 XorA: we haev infra for such one-line patches top be reused to 0 lines, and it worked great. so, need to fix infra for 2.61, not sth else. Jul 10 12:01:16 xaiki for what machine do you compile? Jul 10 12:01:26 xaiki and which image? Jul 10 12:01:54 woglinde: nano for x86 Jul 10 12:02:00 (i486) Jul 10 12:02:26 psokolovsky: I personally have no time to fixup all the 2.61 problems, so we are probably reliant on OH doing the job and kindly donating it back Jul 10 12:02:34 ;) Jul 10 12:02:54 as least grepping for autoconf 2.61 hacks is easy :-) Jul 10 12:03:28 XorA: ok, np, I just wanted to submit those bugs to dump what I have found. Thanks for looking at that, it's much appreciated! Jul 10 12:03:48 psokolovsky: no probs, dodgy autoconf code is becoming a speciality Jul 10 12:04:02 xaiki can you provide your local.conf at pastebin? Jul 10 12:04:27 XorA: but how to convince mallum to pay for it? Jul 10 12:04:41 hrw: we buy him beers at guadec Jul 10 12:04:49 hrw: or find out what Scotch he likes to drink Jul 10 12:05:36 ;D Jul 10 12:15:37 hrw: just for fun (?!) I tried to make parted congizabel with 2.59: http://pastebin.ca/612429 . too much a bit, right? ;-) Jul 10 12:17:48 morning Jul 10 12:17:56 psokolovsky: packages/gnutls/*/onlyonce.m4 ;) Jul 10 12:18:18 is it possible to build angstrom-x11-image at the moment for h3900 ? Jul 10 12:18:33 jan_js: should be Jul 10 12:19:25 hmm... my build repeatedly fails at angstrom-version.bb Jul 10 12:19:34 hrw: such hacks will break again and again, as you know. maybe we should consider using multiple concurrent autconf versions if new indeed break too much. Jul 10 12:21:01 psokolovsky: onlyonce.m4 is not a hack. it is solution for 2.59 autoconf Jul 10 12:22:19 hrw: solution? you must be kidding. how that can be a solution if it requires 2.60 or 2.61 actually? Jul 10 12:26:14 psokolovsky: gnutls 1.6.3 req autoconf 2.60 - onceonly.m4 is set of macros for autoconf 2.54+ (2.57 recommended) and is distributed by default with autoconf 2.60+ Jul 10 12:27:24 ah, so nice of gnu eggheads... Jul 10 12:28:10 psokolovsky, I think multiple autoconfs is a good idea. Jul 10 12:29:01 hi mr_nice Jul 10 12:29:08 polyonymous: apparently yep, afaik, that's how "big distros" deal with such issues Jul 10 12:29:38 yes and I'd think for a reason. Jul 10 12:32:08 debian has autoconf 2.13 and 2.61 Jul 10 12:32:10 hi woglinde Jul 10 12:32:31 mr_nice: do you plan to push some of simpad/collie patches upstream? Jul 10 12:32:41 mr_nice: pcmcia_suspend one for example looks nice Jul 10 12:32:57 hrw you know the problem in person of rmk Jul 10 12:33:37 hrw: mrdata and I are discussing about push them upstream. the plan is to clean the code and try easy patches first. Jul 10 12:33:37 woglinde: sure, but if we do not send patches at all we will one day have 50M of kernel + 60M of patches Jul 10 12:33:39 hmm.. yes, in gentoo 2.59 and 2.61 belong to the same slot, so I think it's possible to make use of autoconf 2.61 instead of 2.5* Jul 10 12:34:25 what did the changelog says about 2.59->2.60->2.61 Jul 10 12:34:55 hrw: we also must ask thesing Jul 10 12:35:11 hrw: but I also would like to have the stuff upstream :) Jul 10 12:36:24 * woglinde too Jul 10 12:36:38 * hrw would like to have x2 in desktop Jul 10 12:36:52 will have to think about buying it one day Jul 10 12:42:44 who is the gcc-csl-arm expert? Jul 10 12:43:50 probably no one now Jul 10 12:45:39 how much ram is recommended for abiword? Jul 10 12:46:28 hrw: I checked in a patch for 3.4.4-csl-2005q3 - it allowed it to compile BE for me. I guess I can't break anything if no-one is using it :-) Jul 10 12:46:35 mr_nice: I started it on 64M device Jul 10 12:47:16 rwhitby: we decided to forget it existance when apps built with it started to break (were working when built with vanilla 3.4.4) Jul 10 12:47:59 fair enough - I'm only using it to temporarily build this freecom kernel until i can get sata-via to work on 2.6.22 Jul 10 12:48:39 woglinde: http://pastebin.com/m24f12826 Jul 10 12:55:52 XorA: FYI: "libparted.so: undefined reference to `rpl_memcmp'". that's all about oneliners - they are 1 lines, but you need dozens of them ;-) Jul 10 12:58:11 psokolovsky: I notice at the top of configure there is a site= line, I dont suppose if you feed our sitefile in that line it works? Jul 10 12:58:37 psokolovsky: or a --site= argument to calling it Jul 10 12:59:03 woglinde: DEBUG: adding /home/xaiki/Wrk/Sagem/OpenEmbedded/org.openembedded.dev/packages/uclibc/uclibc_0.9.28.bb to satisfy virtual/i486-linux-uclibc-libc-for-gcc Jul 10 12:59:34 XorA: well, the big issue is what's up with 2.61 that it doesn't want to interpret $CONFIG_SITE ... Did we patch anything regarding that in 2.59? Jul 10 13:00:11 psokolovsky: plase seperate 2.61 autoconf work from problems building parted, your starting to confuse me again :-) Jul 10 13:00:35 re Jul 10 13:00:43 jo florian Jul 10 13:01:45 XorA: well, I can build parted the way I need, forget about that. ;-) but leveraging 2.61 is what we at all should do, and apparently better earlier than later, as we already have bunch of hacks regarding it, it seems... Jul 10 13:02:03 psokolovsky: yeah, 2.61 is the way to go Jul 10 13:03:17 xaiki and it still build uclibc-0.9.28? Jul 10 13:03:26 woglinde: yes =( Jul 10 13:03:31 I removed my build/tmp dir Jul 10 13:03:58 woglinde: it does parse the 0.9.29 .bb file. Jul 10 13:04:07 but I dunno why it still chooses 0.9.28 Jul 10 13:04:10 xaiki okay please look at conf/distro/generic-uclibc.conf Jul 10 13:04:18 whats defined Jul 10 13:04:38 than type bitbake yourimagename -e and search for uclibc Jul 10 13:04:42 woglinde: 0.9.28 has: DEFAULT_PREFERENCE = "1" Jul 10 13:04:52 when .29 has no DEFAULT_PREFERENCE set Jul 10 13:04:58 xaiki yes but you ovveride it Jul 10 13:05:02 hm Jul 10 13:05:12 0.9.28 should be changed Jul 10 13:05:29 if no ones mind I will do it later this day Jul 10 13:05:37 ok, I'll try doing it locally and'll tell you. Jul 10 13:06:01 xaiki quick and diryt workaround set DEFAULT_PREFERENCE = "1" Jul 10 13:06:08 in uclibc-0.9.28 to -1 Jul 10 13:07:14 woglinde: so .28 shouldn't have DEFAULT_PREFERENCE ? should .29 have it ? Jul 10 13:07:23 (not really into quick and dirty =)) Jul 10 13:07:37 do either the -1 or edit both Jul 10 13:08:40 'tsbuilding Jul 10 13:09:01 woglinde: have you viewed my bb patches ? do you think I should submit them on the ML ? Jul 10 13:09:25 the pathes one is usefull to have a local conf, but it's not really elegant, the other one seems like a real bug. Jul 10 13:09:31 xaiki you can asked zecke if they are okay Jul 10 13:09:31 (and there might be others from the same kind) Jul 10 13:09:46 for the debian patch I think it will not go in Jul 10 13:10:13 damnit why I awalys put today the past form on verbs Jul 10 13:10:32 psokolovsky: configure in parted ignores CONFIG_SITE which is passed by siteinfo.bbclass Jul 10 13:11:44 woglinde: debian patch ? packaging you mean, yeah I rekon the current packages are good enough Jul 10 13:11:53 even tho I'm not a fan of python-support Jul 10 13:12:21 XorA: yes, and if I regenerate it with 2.61, it still ignores. and if I apply that hacky patch to downgrade it to 2.59, it does. Jul 10 13:12:33 XorA: that's why I say that problem is in autoconf 2.61 Jul 10 13:16:21 psokolovsky: you are right general.m4 in autoconf 2.61 looks broken to me Jul 10 13:17:53 psokolovsky: replace in tmp/staging/share/autoconf/general.m4 AC_SITE_LOAD with this# AC_SITE_LOAD Jul 10 13:17:53 # ------------ Jul 10 13:17:53 # Look for site or system specific initialization scripts. Jul 10 13:17:53 m4_define([AC_SITE_LOAD], Jul 10 13:17:53 [# Prefer explicitly selected file to automatically selected ones. Jul 10 13:17:54 if test -z "$CONFIG_SITE"; then Jul 10 13:17:56 if test "x$prefix" != xNONE; then Jul 10 13:17:58 CONFIG_SITE="$prefix/share/config.site $prefix/etc/config.site" Jul 10 13:18:02 else Jul 10 13:18:04 CONFIG_SITE="$ac_default_prefix/share/config.site $ac_default_prefix/etc/config.site" Jul 10 13:18:06 fi Jul 10 13:18:08 fi Jul 10 13:18:10 for ac_site_file in $CONFIG_SITE; do Jul 10 13:18:12 if test -r "$ac_site_file"; then Jul 10 13:18:14 AC_MSG_NOTICE([loading site script $ac_site_file]) Jul 10 13:18:16 sed 's/^/| /' "$ac_site_file" >&AS_MESSAGE_LOG_FD Jul 10 13:18:18 . "$ac_site_file" Jul 10 13:18:18 uh Jul 10 13:18:20 fi Jul 10 13:18:21 flush Jul 10 13:18:22 done Jul 10 13:18:24 ]) Jul 10 13:18:30 XorA, flooder! :) Jul 10 13:18:36 sorry, it looked smaller before I copied it Jul 10 13:18:46 It might have grown on the way. Jul 10 13:18:51 ;-) Jul 10 13:19:03 XorA: ok, hope to try in few Jul 10 13:19:15 psokolovsky: and I dont think that is a thing we patch Jul 10 13:19:40 psokolovsky: unless I didnt look in right patch Jul 10 13:22:43 XorA: "for ac_site_file" - so, does 2.61 expects site's on command line ?? Jul 10 13:24:26 psokolovsky, I think theoretically all the env vars you pass to autoconf are better to be passed via the command line. Jul 10 13:24:43 psokolovsky: maybe, but it looks just broken to me Jul 10 13:24:47 psokolovsky, this way makefiles can rerun autoconf preserving environment. Jul 10 13:25:43 psokolovsky: I cant see where it sets for ac_site_file but I didnt do a deep search Jul 10 13:25:45 polyonymous: well, that's not too grave issue, as bitbake runs all stuff with predefined env Jul 10 13:26:21 psokolovsky, right, but you see - they seem to rely on this behaviour now and the other way is somewhat deprecated for quite a while, I think. Jul 10 13:26:42 psokolovsky, if that actually fixes the problem I'd suggest that we do pass it in command line (we don't have to remove it from env, of course) Jul 10 13:27:21 I'm leaving soon so I don't even look at any part of the code atm... Jul 10 13:28:04 polyonymous: autoconf --help Jul 10 13:28:22 03rwhitby 07org.oe.dev * r30fd0291... 10/ (3 files in 3 dirs): gcc-csl-arm-2005q3: Added patch to support big-endian builds of 3.4.4 Jul 10 13:28:32 03rwhitby 07org.oe.dev * r7955ac42... 10/ (1 packages/linux/fsg3-kernel_2.6.18.bb): fsg3-kernel: Update to latest svn defconfig Jul 10 13:28:49 psokolovsky, and? Jul 10 13:29:14 polyonymous: see site files on command line in usage Jul 10 13:29:30 in autoconf?? Jul 10 13:29:31 so maybe 2.61 has that code in shell func? Jul 10 13:29:46 ah damn, ok Jul 10 13:30:21 To assign environment variables (e.g., CC, CFLAGS...), specify them as Jul 10 13:30:22 VAR=VALUE. See below for descriptions of some of the useful variables. Jul 10 13:30:28 that's a configure --help output Jul 10 13:31:10 yep, so no explicit sign either Jul 10 13:31:14 bye Jul 10 13:31:33 well, it explicitly points at the right way to pass env vars. Jul 10 13:31:48 once again, I don't t know if it's our problem :) Jul 10 13:32:27 btw, we can also place site file into staging which will source in the appropriate sitefiles Jul 10 13:41:48 * polyonymous is gone Jul 10 13:57:16 03tmbinc 07org.oe.dreambox * r05e6a4d6... 10/ (1 packages/linux/linux-dm7025.bb): linux-dm7025: build mksquashfs,unsquashfs Jul 10 13:57:20 03tmbinc 07org.oe.dreambox * r9891a214... 10/ (1 packages/file/file-native_4.16.bb): file-native: add zlib-native to depends Jul 10 14:16:41 Good afternoon Jul 10 14:16:56 hrw: fixed a autoconf 2.61 bug with site files, you might want to retest image making Jul 10 14:17:47 sirfred: I have you libqte file Jul 10 14:17:54 XorA: great! Jul 10 14:18:01 sirfred: where shall I put it Jul 10 14:18:12 XorA: Could you just mail it to me? Jul 10 14:18:20 sirfred: no probs, address Jul 10 14:18:22 XorA: How bit is it, compressed? Jul 10 14:18:30 manuel.teira at telefonica.net Jul 10 14:18:37 s/bit/big/ Jul 10 14:19:30 * djay-il got fired today Jul 10 14:19:34 hm Jul 10 14:20:04 djay-il: did you want to? Jul 10 14:20:21 sirfred: gzipping now Jul 10 14:20:39 XorA: my best day with this employer :-) Jul 10 14:20:50 djay-il: YAY then Jul 10 14:20:52 djay-il: Congrats, then Jul 10 14:21:22 I get 1 month pay and don't have to come here, in addition I get another payment because they fired me... Jul 10 14:21:39 thanks :-) Jul 10 14:21:46 which package holds the real killall , please ? Jul 10 14:23:30 Wow, I've just opened the browser to look for something, and I've forgot what it was. Jul 10 14:23:37 Terrible Jul 10 14:23:37 sirfred: 1.6M Jul 10 15:20 libqte.so.2.3.2.gz Jul 10 14:23:52 XorA: ok. Jul 10 14:25:06 sirfred: on its way Jul 10 14:25:06 * Zero_Chaos slaps slapin_ and XorA Jul 10 14:25:14 ~fish Zero_Chaos Jul 10 14:25:17 * ibot slaps Zero_Chaos around with a large trout Jul 10 14:25:34 XorA: double head, your fault by 23 seconds Jul 10 14:25:47 Zero_Chaos: already fixed Jul 10 14:25:53 XorA: no no, my fault, slapin_'s fault by 2 min Jul 10 14:25:56 XorA: Thanks a lot. I will take a look to see what's inside Jul 10 14:26:03 thanks Jul 10 14:26:16 Zero_Chaos: you can fix them yourself, just mtn merge Jul 10 14:26:47 XorA: I can fix a lot of things for me, I can't fix it for others. Jul 10 14:26:58 Zero_Chaos, what? Jul 10 14:27:07 psmisc , thanks ;-) Jul 10 14:27:10 Zero_Chaos: my script tells me when I do it Jul 10 14:27:16 Zero_Chaos: so I fix immediately Jul 10 14:27:22 slapin: you caused a double head in the monotone repo Jul 10 14:27:26 and it's not fixed... Jul 10 14:27:56 Zero_Chaos, I changed only 4 mine palm files and added one, how come? Jul 10 14:28:18 slapin: because you didn't update before you pushed I'm sure. Jul 10 14:28:28 but anyway, merge it ;-) Jul 10 14:32:22 Zero_Chaos, done Jul 10 14:32:31 slapin: thanks mate Jul 10 14:33:19 slapin: for future: "mtn pull&&mtn merge&&mtn update&&mtn push" if you have to push something Jul 10 14:33:27 XorA: pushed? Jul 10 14:33:36 hrw: yes Jul 10 14:34:22 night all Jul 10 14:34:39 hrw: bug meant none of our site files were acutally used in autoconf Jul 10 14:34:49 hrw: so all our nicely chosen values were ignored Jul 10 14:34:54 hrw, thanks! Jul 10 14:35:25 slapin: and first 3 parts if want to pull&update Jul 10 14:36:52 XorA: forgot to bump PR Jul 10 14:36:57 ~lart you XorA Jul 10 14:36:58 * ibot brandishes Excalibur! "With this sword, I vanquish thee, you XorA!" and lops off you XorA's head Jul 10 14:36:59 03xora 07org.oe.dev * r61fe56a7... 10/ (3 files in 3 dirs): Jul 10 14:36:59 autoconf_2.61.bb : fix a "bug" in autoconf where it treated CONFIG_SITE Jul 10 14:36:59 as a single filename instead of splitting it into individual filenames. Jul 10 14:36:59 This stopped autoconf 2.61 loading the site files. Jul 10 14:37:06 03slapin 07org.oe.dev * r25a1b52e... 10/ (6 files in 3 dirs): Palms: unification of configs Jul 10 14:39:36 03hrw 07org.oe.dev * r92f0750a... 10/ (3 files in 2 dirs): glib-2.0: fix packaging in 2.12.12 - should end discussion on OE list Jul 10 14:42:51 hrw: I didnt forget, it was marked as not default and didnt work so I figured no-one would have it built :-) Jul 10 14:44:13 hrw: can you quickly point me to package which resurrects kernel syscall macros which were gone in latest linux-headers? Jul 10 14:45:24 psokolovsky: no Jul 10 14:45:39 XorA: ok Jul 10 14:46:45 hrw: in hindsite I forgot you probably did have it built Jul 10 14:46:54 ~lart me for being a muppet Jul 10 14:46:55 * ibot throws xora's poor little doggy off a cliff for being a muppet Jul 10 14:47:44 mickeyl: hi Jul 10 14:48:29 hi Jul 10 14:48:44 XorA: Thinks like QGfxTC6393<1, 0>::blt(int, int, int, int, int, int), can be found inside. Jul 10 14:48:57 XorA: rebuilt and going to parted 1.8.7 Jul 10 14:49:06 XorA: Those are good news. It seems that it has an accelerated qte driver inside. Jul 10 14:59:00 * * OE Bug 2635 has been created by winthrowe(AT)gmail.com Jul 10 14:59:02 * * No sleeve driver built on h5000 Jul 10 14:59:03 NOTE: package parted-1.8.7-r1: task do_build: completed Jul 10 14:59:03 * * http://bugs.openembedded.org/show_bug.cgi?id=2635 Jul 10 14:59:36 XorA, psokolovsky: look at glib-2.0 fix Jul 10 15:00:20 what in OE use libparted at all? Jul 10 15:00:21 ML's slow ;-\ Jul 10 15:00:29 hrw: gparted will Jul 10 15:01:05 psokolovsky: will.. but do not use as it is not in OE yet Jul 10 15:03:13 hrw: and? Jul 10 15:03:38 nothing - wanted to know to test parted 1.8.7 with soemthing Jul 10 15:04:01 17:03 hrw@home:autoconf261$ tmp/staging/x86_64-linux/bin/autoconf --version Jul 10 15:04:01 autoconf (GNU Autoconf) 2.61 Jul 10 15:04:22 automake 1.9.6, m4 1.4.8 Jul 10 15:04:40 hrw: automake must be upgraded to 1.10 too Jul 10 15:04:56 psokolovsky: in debian 1.9 is default Jul 10 15:05:21 hrw: its majesty parted requires Jul 10 15:05:35 really? then why I built it with 1.9.6? Jul 10 15:06:06 ah. do_configure stuff Jul 10 15:06:15 re Jul 10 15:06:47 hrw: remember: it was building before. check that it autoreconfigured properly. Jul 10 15:07:38 ok. broke Jul 10 15:07:44 | configure.ac:47: require Automake 1.10, but have 1.9.6 Jul 10 15:08:38 hrw: (unrelated) will do_unpack_append() work? ;-) Jul 10 15:08:47 psokolovsky: no Jul 10 15:08:57 Zero_Chaos: what will? ;-) Jul 10 15:09:01 psokolovsky: nor do_patch_prepend in my experience Jul 10 15:09:11 psokolovsky: add a task before patch after unpack ;-) Jul 10 15:09:15 psokolovsky: should Jul 10 15:09:51 hrw: if it should then there is a bug because I did this dance with CoreDump last week and it doesn't. Jul 10 15:10:33 do_patch is a python task so it gets messy Jul 10 15:10:53 appending/prepending shell onto python doesn't work in general ;-) Jul 10 15:10:55 psokolovsky: NOTE: package parted-1.8.7-r2: task do_configure: completed Jul 10 15:11:04 psokolovsky: with automake-native 1.10 Jul 10 15:11:50 hrw: ok, now just check config.log that it sources site's ;-) Jul 10 15:12:27 hrw: Jul 10 15:12:28 ERROR: 0001:def do_unpack(): Jul 10 15:12:28 ERROR: 0002: bb.build.exec_func('base_do_unpack', d) Jul 10 15:12:28 ERROR: 0003: cp ${WORKDIR}/syscalls.h ${S}/libparted/arch/ Jul 10 15:12:37 that's why it doesn't work ;-) Jul 10 15:12:55 psokolovsky: it source site files Jul 10 15:13:02 hrw: cool! Jul 10 15:13:05 psokolovsky: that stuff is for do_configure_prepend Jul 10 15:13:33 psokolovsky: unpack/patch are python. configure/compile/install/stage are shell Jul 10 15:14:09 psokolovsky: like I said, you want to add a task before patch and after unpack and put things there. or if it should be after patch you can prepend configure or something Jul 10 15:14:39 psokolovsky: anyway it fails now during compile Jul 10 15:15:02 hrw: well, that bitabake bug, right? Jul 10 15:15:18 psokolovsky: bug, not really, just is. Jul 10 15:15:21 hrw: of course it does, remember that syscalls question? ;-) Jul 10 15:15:25 psokolovsky: tasks? it is just is Jul 10 15:15:29 hrw: I'll fix it and push Jul 10 15:15:29 psokolovsky: yes Jul 10 15:16:21 Zero_Chaos: it is bug, and needs to be fixed Jul 10 15:16:39 psokolovsky: it is not a bug Jul 10 15:16:50 psokolovsky: putting shell code into python should be expected to fail. Jul 10 15:16:59 psokolovsky: it should be documented though. Jul 10 15:17:01 hrw: what you mean? putthing code which is nor marked as python into python func? Jul 10 15:17:05 psokolovsky: it is like writing do_configure_append() in ruby and complaing that it does not work Jul 10 15:17:26 psokolovsky: you cant use shell in do_*_append/prepend to python tasks and viceversa Jul 10 15:17:31 hrw: python tasks/snippers are marked with "python" Jul 10 15:17:42 hrw: that's bug Jul 10 15:17:47 psokolovsky: I pushed automake 1.10 for you Jul 10 15:17:52 thanks Jul 10 15:19:17 np Jul 10 15:32:23 hrw: so, what do you think, memcmp() is working in glibc? ;-) Jul 10 15:40:07 should Jul 10 15:42:40 hrw: so I'll add that to common-glibc Jul 10 15:43:14 where do i get ldd ? Jul 10 15:44:04 03hrw 07org.oe.dev * rd1b0e1fc... 10/ (3 files in 2 dirs): automake: added 1.10 - not tested too much, needed by parted 1.8.7, not default Jul 10 15:44:12 03mickeyl 07org.oe.dev * r4ea1a0b7... 10/ (4 files in 3 dirs): fic-gta0[0|1].conf: move recommendation of alsa-state(s) from task-openmoko to machine configuration Jul 10 15:45:16 mickeyl: task-base recommends alsa-state for alsa machines Jul 10 15:45:49 hrw: ah good. i better add alsa-states to that and remove it from the machine confs then Jul 10 15:46:01 moin mickeyl Jul 10 15:46:06 hey woglinde Jul 10 15:50:50 heh, gparted at least starts ;-) Jul 10 15:58:18 RP: pign Jul 10 15:58:20 ping, even Jul 10 15:58:26 mickeyl: pong Jul 10 15:58:33 ah hey Jul 10 15:58:38 RP: two questions. 1st) did you see my botmail? Jul 10 15:58:50 2nd) how did your discussion with henryk_afc work out? Jul 10 15:58:51 mickeyl: Yes, I even replied Jul 10 15:58:55 oh Jul 10 15:58:58 ~botmail read Jul 10 15:59:22 make the module depend on the firmware or the firmware on the module? Jul 10 15:59:37 module on firmware Jul 10 15:59:42 thanks Jul 10 15:59:43 mickeyl: henryk_afc posted the patch and we've dicussed it. Basically it needs work Jul 10 15:59:44 ah, ok. I will add a bugzilla bug then w/ a testcase Jul 10 15:59:50 mickeyl: Its a good start though Jul 10 15:59:56 Zero_Chaos: you can get module into kernel. Jul 10 15:59:58 hrw: rdep right? Jul 10 15:59:59 mickeyl: testcases are good :) Jul 10 16:00:02 RP: ok, glad to hear. Jul 10 16:00:16 hrw: I was going to add the module to the kernel ;-) Jul 10 16:00:29 Zero_Chaos: module RDEPENDS on firmware Jul 10 16:00:38 hrw: that's what I thought, thanks Jul 10 16:00:43 mickeyl: I think we agreed it needs to be split into smaller chunks and merged piece by piece. I've rearranged the svn fetcher to help make it easier to work with as a start Jul 10 16:01:18 mickeyl: I ran out of time for anything more and will be out of time until after GUADEC now Jul 10 16:01:53 RP: ok. will we have time for a private chat on GUADEC? Jul 10 16:02:04 mickeyl: Yes, we should do Jul 10 16:02:10 great. looking forward to that. Jul 10 16:02:22 mickeyl: We'll have to make time :) Jul 10 16:02:25 Hi! Jul 10 16:02:45 Does anybody know why libxml2-native is configured with --without-schemas ? Jul 10 16:03:11 anttix: to be smaller Jul 10 16:03:28 hrw: native package :P Why? Jul 10 16:03:50 anttix: because it has same settings as target one Jul 10 16:03:59 hrw: the fun thing about this is, that libxml2 itself is not configured without schemas Jul 10 16:04:08 no, target has different options. Jul 10 16:04:15 then no idea Jul 10 16:04:24 03mickeyl 07org.oe.dev * r79a8be76... 10/ (4 files in 3 dirs): fic-gta0[1|2].conf: task-base handle alsa-states, no need to duplicate it here Jul 10 16:04:26 Yeah, it puzzles me too! Jul 10 16:04:46 Who maintains this libxml cruft? Jul 10 16:06:00 anttix: you can be the one Jul 10 16:07:06 RP: nod Jul 10 16:07:18 RP: I'll be there from Sunday to Thursday morning Jul 10 16:07:20 hrw: I have no commit access to OE mtn. Jul 10 16:07:31 anttix: but you have access to bugtracker Jul 10 16:07:44 But i will submit new versions of libxml2 and libxslt Jul 10 16:08:03 anttix: libxml in one, libxslt in second raport please Jul 10 16:08:06 mickeyl: I'm there Mon-Fri Jul 10 16:08:08 I need them since older xsltproc has a bug in namespace handling that affects me. Jul 10 16:08:13 anttix: both as 'mtn diff' please Jul 10 16:08:23 ok Jul 10 16:08:38 RP: ok, that'll work Jul 10 16:08:50 * hrw will be Mon evening -> Wed evening (leaving early Thur) Jul 10 16:09:02 Also I'm offline Wed-Fri this week FWIW Jul 10 16:09:23 hrw: any other things to keep in mind when submitting new .bb files? Jul 10 16:09:35 ok, so we will meet at the OE meeting Monday evening and then see from there Jul 10 16:12:58 mickeyl: sounds good. I think we all have mobile numbers from OEDEM? Jul 10 16:13:14 RP: yeah (if no one changed his) Jul 10 16:13:24 * RP hasn't Jul 10 16:13:30 * mickeyl neither Jul 10 16:13:44 hrw, florian ? Jul 10 16:13:55 woglinde: ? Jul 10 16:13:59 yes? Jul 10 16:14:07 woglinde: ah, you won't be there anyway Jul 10 16:14:10 or will you? Jul 10 16:14:12 (GUADEC?) Jul 10 16:14:15 mickeyl: do you have mine Jul 10 16:14:20 no Jul 10 16:14:21 XorA: let me check Jul 10 16:14:21 sorryx Jul 10 16:14:26 money time family Jul 10 16:14:32 yeah Jul 10 16:14:34 the three thingies Jul 10 16:14:49 hm Jul 10 16:14:56 XorA: i havent! Jul 10 16:15:02 why the hack classpath needs X Jul 10 16:15:06 let me check deps Jul 10 16:15:17 XorA: I haven't yours either Jul 10 16:15:30 mickeyl: My number changed... i'll sen an update mail. Jul 10 16:15:34 AH Jul 10 16:15:40 multiple providers Jul 10 16:20:45 florian: ok, thanks Jul 10 16:21:08 re Jul 10 16:21:33 hi henryk, glad to hear you and RP are coming forwards Jul 10 16:21:50 OE will soon require bb 1.8 and then we're screwed if we don't catch up ;) Jul 10 16:22:05 yep Jul 10 16:22:07 anyway, 90% of OM has now merged into OE Jul 10 16:22:13 the rest will come over the next days Jul 10 16:22:27 re henryk Jul 10 16:24:13 RP: The other day while walking home, I had an idea, let me see if I can get it together (I should have written it down right away) ... Jul 10 16:26:14 mickeyl: The 1st July deadline has passed so OE does require 1.8 now... Jul 10 16:26:15 RP: basically I think every recipe should provide its own functions to retrieve revision_identifier and revision_counter, though of course 99% of all packages will use a standard version, to be include in a srcrev.bbclass that is to be written. Jul 10 16:26:48 RP: right. which means OM can't sync. with OE anymore. but it doesn't matter anyway Jul 10 16:27:40 mickeyl: I don't think anything in the tree is 1.8 only yes, koen just bumped the version and you only hit that if you use sanity.bbclass Jul 10 16:27:41 RP: providers.py would then check whether the package currently processed contains these functions and behave accordingly. Jul 10 16:28:27 RP: sure. it's ok. 1.8 has been around for long enough. i'm glad you all respected my 2 months request. Jul 10 16:29:12 hi Jul 10 16:29:20 ewi died again Jul 10 16:30:18 mickeyl: I want decent svn "now" support in bitbake on a personal level and people using poky keep asking about it too Jul 10 16:30:26 mickeyl: We'll make it work somehow :) Jul 10 16:30:47 yay! Jul 10 16:30:50 thanks :) Jul 10 16:32:17 though not today. I have to hold a demonstration tomorrow. (That's why I invested the weekend in PhoneME on OpenMoko) Jul 10 16:32:49 henryk: Realistically, I won't get to this until after GUADEC now Jul 10 16:33:30 RP: ok Jul 10 16:33:41 hm is there an example how to compile a package with two source-archs? Jul 10 16:34:09 woglinde: for target with host binaries used during build? Jul 10 16:34:18 henryk: If you did find time to split the patch up a bit and work in stages or address any of the issues we discussed before then, that would be good Jul 10 16:34:26 hrw not quiet Jul 10 16:34:40 hrw some java sourceses which dont bee needed installed Jul 10 16:34:55 either in staging nor on devoce Jul 10 16:35:02 hi mrdata Jul 10 16:35:04 hi Jul 10 16:35:09 hi woglinde Jul 10 16:35:12 RP: yes, I will poke around. I want to integrate it more firmly. (Up to now I have tried to be minimally intrusive and only add) Jul 10 16:36:15 henryk: minimially intrustive is good in some ways. It you're not going to do that, it makes doing things in a logical patch series all the more important though Jul 10 16:49:58 hrw I will found myself what happend if I put two sources in _SRC Jul 10 16:50:18 happens Jul 10 16:50:24 *sigh* Jul 10 16:50:27 where should I put the kernel module rdepends? linux-rp_2.6.21.bb or linux-rp.inc? Jul 10 16:50:52 Zero_Chaos: which module it is? Jul 10 16:51:27 zd1211rw Jul 10 16:51:48 .inc I would use Jul 10 16:51:55 ooh, someone's gonna try to port busybox to mingw32, so people can get a shell without msys. thatd be nifty Jul 10 16:52:21 Zero_Chaos: or add zd1211rw-firmware into proper place in task-base even Jul 10 16:53:24 hrw: I don't think anyone wants me to add usb wifi drviers to task-base ;-) Jul 10 16:53:55 ;) Jul 10 16:53:57 RP: ping Jul 10 16:55:20 http://pastebin.ca/612792 - how funny output bitbake can show Jul 10 16:58:05 bye Jul 10 17:01:24 well, gparted even works a bit, and even barely usable on qvga ;-D Jul 10 17:03:43 Zero_Chaos: pong Jul 10 17:04:28 RP: i'm submitting a patch to you for your kernel, but where would you like KERNEL_MODULE_zd1211rw_rdepends? in which file? Jul 10 17:05:28 Zero_Chaos: Why does it need RDEPENDS? Jul 10 17:05:35 RP: firmware Jul 10 17:06:01 Zero_Chaos: linux-rp.inc I guess Jul 10 17:06:20 RP: thanks Jul 10 17:06:34 and it was "KERNEL_MODULE_zd1211rw_rdepends" right? Jul 10 17:06:47 Zero_Chaos: Probably not Jul 10 17:06:55 heh Jul 10 17:07:11 Zero_Chaos: RDEPENDS_kernel-module-zd.... = at a guess Jul 10 17:19:34 RP: So, did you hear that (R)DEPENDS overriden with PACKAGES_DYNAMIC thing doesn't really work? Jul 10 17:19:57 psokolovsky: Not really Jul 10 17:20:42 RP: look at kernel.bbclass, grep for dtl1 - all that doesn't really work, I had to add RDEPENDS to task-base after all Jul 10 17:21:55 psokolovsky: no match found? Jul 10 17:22:06 RP: another issue which came out today - if a task's body originally was defined as python, if you try to append/prepend to it, it's blindly stuffed in it, not paying attentioon that now it's another language Jul 10 17:22:44 psokolovsky: Is that user error or bitbake's problem though? Jul 10 17:22:47 RP: simply didn't work - *dtl1-workaround* wasn't build (but kernel module has Depends: in control of course) Jul 10 17:23:12 psokolovsky: What simply didn't work? :) Jul 10 17:23:15 RP: it's bitbake problem. for what reason it puts shell code into python? Jul 10 17:23:42 psokolovsky: You mean RDEPENDS_kernel-module-something = "something" didn't set the module's RDEPENDS correctly? Jul 10 17:23:48 RP: what written after dash Jul 10 17:24:07 psokolovsky: How is bitbake supposed to know if an append/prepend is shell or python? Jul 10 17:24:08 RP: "*dtl1-workaround* wasn't build" Jul 10 17:24:29 RP: the same way it nows about not-append/prepend? Jul 10 17:24:36 psokolovsky: Ah, it wasn't built. Right, that is expected :-( Jul 10 17:25:07 AFAIK, python code is explicitly marked as such, default is shell. pretty easy to know. Jul 10 17:25:15 psokolovsky: You tell bitbake whether a function is shell or python. You don't do that with append/prepend, its simply not part of the syntax Jul 10 17:25:48 RP: so, that's bug with the cause "bad spec". it's still a bug. Jul 10 17:26:00 * * OE Bug 2636 has been created by r.goyet(AT)gmail.com Jul 10 17:26:02 * * The corgi_ts driver has a small bug : it doesn' t trigger the KEY event it advertises Jul 10 17:26:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2636 Jul 10 17:26:14 psokolovsky: File a bug report then, I don't have time to think about that atm :-( Jul 10 17:26:18 sorry :-( Jul 10 17:27:10 psokolovsky: The PACKAGES_DYNAMIC problem is a really nasty one. All I can suggest is adding the module name to PACKAGES manually and teaching OE to get along with duplicate data in PACKAGES and PACKAGES_DYNAMIC Jul 10 17:27:27 ok ;-) this append is really not much important, just wanted to "let know" ;-) Jul 10 17:27:46 psokolovsky: File a bug so it doesn't get forgotten about, its a long standing problem Jul 10 17:27:54 RP: well, I just added it at higher level, but yes, this one is more annoying... Jul 10 17:28:00 RP: ok Jul 10 17:29:08 psokolovsky: There is no realistic way bitbake can inspect PACKAGES_DYNAMIC and find any RDEPENDS which might or might not match since it would would have to trawl the entire keyspace :/ Jul 10 17:30:14 Zero_Chaos: You might want to pay attention to the above as I think this affects what you're trying to do as well Jul 10 17:30:25 RP: well, ok, but worth describing in man then (if not yet ;-) ) Jul 10 17:30:36 RP: I add "RDEPENDS_kernel-module-zd1211rw = zd1211rw-firmware" and it doesn't work, adds a DEPENDS but not RDEPENDS Jul 10 17:31:21 Zero_Chaos: I'm surprised it even added the DEPENDS :/ Jul 10 17:32:04 RP: well that sucks. so the only way to add dep/rdep to the kernel-module doesn't work right now? great. Jul 10 17:32:28 Zero_Chaos: That's life. A patch to fix the situation would be most welcome ;-) Jul 10 17:33:27 RP: yeah, I can barely use it and you want me to patch it ;-) Jul 10 17:50:10 ~oemirrors Jul 10 17:50:12 from memory, oemirrors is extra, extra, read all about it, oemirrors is OpenEmbedded monotone mirrors are ewi546.ewi.utwente.nl, monotone.nslu2-linux.org, opensource.wolfsonmicro.com also see http://www.openembedded.org/wiki/GettingStarted Jul 10 18:13:53 RP: hrw|gone Angstrom-image built with autoconf 2.61 Jul 10 18:13:59 psokolovsky: as well for you Jul 10 18:14:15 hm linxutogo server is overloaded Jul 10 18:16:34 XorA|gone: from scratch? Jul 10 18:16:49 damnit just now where I need it Jul 10 18:18:30 florian please help! Jul 10 18:19:29 * florian takes a look Jul 10 18:19:46 :) Jul 10 18:19:50 ~hail florian Jul 10 18:19:52 * ibot bows down to florian and chants, "I'M NOT WORTHY!!" Jul 10 18:22:02 psokolovsky: yes Jul 10 18:22:41 XorA|gone: cool! I guess I'll try that too, and that will add to evidence Jul 10 18:24:20 woglinde: massive web access and the monotone server... Jul 10 18:24:26 florian okay Jul 10 18:24:34 maybay a om attacke Jul 10 18:25:20 florian thanks Jul 10 18:25:29 florian: maybe we should round robin mtn.openembeded.org Jul 10 18:25:54 unluckily I have no idea how to forward to the ewi mtn server Jul 10 18:26:47 XorA|gone: tat doesn't help i fear... its the memory usage the server always needs. but maybe we can add one here at kc if ewi fails Jul 10 18:26:59 florian: I will have to check with Liam, but I think wolfson server can take the bandwidth Jul 10 18:28:20 XorA|gone: that's our only problem here too, but koen mentioned a quite low bandwith usage. If someone gives me a number its quite likely that I only have to turn on a machine. Jul 10 18:29:23 florian: how much memory does the mtn server burn? Jul 10 18:31:04 XorA|gone: ~300mb the last time I checked. Jul 10 18:31:17 florian: ouch Jul 10 18:31:22 oops Jul 10 18:32:44 XorA|gone: right, without this we can live with 1gb of memory easily. Jul 10 18:33:23 florian: opensource.wolfsonmicro.com has 1G in it and all it runs it apache and git-server, it might be an idea for us to share some load for a while Jul 10 18:33:45 florian: lrg and I are admins of that machine Jul 10 18:35:31 hm which variable is for the -prX dir instead of the real source dir of a package? Jul 10 18:36:01 hm AH PF Jul 10 18:36:03 okay Jul 10 18:36:05 XorA|gone: sounds good :-) Jul 10 18:36:17 I am slow today to understand Jul 10 18:36:28 florian: I shall try and talk to lrg this week then, but we both be at guadec Jul 10 18:36:58 XorA|gone: maybe we can du round robin with this, ewi and maybe a third one here then. Jul 10 18:37:14 florian: round robin on ewi is hard as your not allowed to cname it Jul 10 18:37:36 florian: or nslu-linux.org if they are happy Jul 10 18:37:39 XorA|gone: ah too bad... well, I guess our machine is :-) Jul 10 18:39:10 florian: are you at guadec Jul 10 18:39:16 I assume you are Jul 10 18:41:09 XorA|gone: yep Jul 10 18:41:40 florian: then I guess at worst all admins will be there to talk about it Jul 10 18:44:14 XorA|gone: perfect :-) Jul 10 18:44:16 bbl Jul 10 18:46:57 bye florian Jul 10 18:47:03 hm too late Jul 10 19:14:56 * polyonymous is back Jul 10 19:31:28 do mtn mirrors allow pushing? Jul 10 19:32:05 yes Jul 10 19:35:25 hmm, the site working or it just me? Jul 10 19:36:51 timtimred: site is under mega load Jul 10 19:37:30 * woglinde is working on classpath Jul 10 19:38:42 heh, i'll stop hammering it then... doesnt work at all here (timeout) Jul 10 19:45:58 psokolovsky: just pushing m4-native 1.4.8 which is needed by some packages when using autoconf 2.61 Jul 10 19:46:22 just about time. Jul 10 19:47:15 synced to a mirror so bugger knows when it will hit OE main Jul 10 19:47:50 I'm having troubles including a new package into an openmoko build. I added DEPENDS += "mypackage" to local.conf. I'm getting a circular dependency with quilt-native Jul 10 19:49:18 upwards chain is: openmoko-devel-image -> ipkg-native -> pkgconfig-native -> autoconf-native -> quilt-native -> mypackage -> ipkg-utils-native Jul 10 19:50:03 I have no dependencies specified in mypackage.bb, nor does it include or inherit anything. Jul 10 19:50:11 Any idea what's going on ? Jul 10 19:51:07 jseghers: you added all packages depend on your package Jul 10 19:51:13 jseghers: including your package Jul 10 19:52:00 How do I include it in the build? Jul 10 19:52:41 without saying everything depends on it? Jul 10 19:52:44 add it as an RDPENDS to openmoko-devel-image or any of the tasks that openmoko-devel-image calls Jul 10 19:53:33 I'd tried that and it just ignored it in the build. Jul 10 19:54:36 add it to task-openmoko.bb in one of the tasks there Jul 10 19:54:40 I'd added it to one of the RDEPENDS_ variables already defined. Jul 10 19:54:43 cu Jul 10 19:54:47 yes, added it in there. Jul 10 19:55:23 jseghers: then you got it wrong, pastebin what you put Jul 10 19:55:27 that should work (tm) promise :) Jul 10 19:55:38 did you rebuild that package properly? :) Jul 10 19:55:39 OK... I'll try again... Jul 10 19:56:05 I did a completely clean build, after creating a patch to add it to the task Jul 10 19:56:16 my package didn't get built Jul 10 19:56:44 I'd been directly building it to develop the .bb file, and then was scratching my head as to why it didn't get into the image Jul 10 19:57:08 Then did a complete rebuild and, lo and behold, it was being ignored... Jul 10 19:57:23 But I'll try again and if it doesn't work I'll post back. Jul 10 19:57:45 jseghers try bitbake myimage -g to see dependencies in depends.dot Jul 10 19:58:49 OK... just a moment... it's in the middle of a build where I'd changed local.conf to have RDEPENDS instead of DEPENDS... 99% through the 4400+ .bb files. Jul 10 19:59:18 puh finally classpath-minimal configures Jul 10 20:01:31 Changing it to RDEPENDS in local config is at least building my package (that being phoneME-Feature)... we'll see if it also makes it in to the image (crossing fingers) Jul 10 20:02:09 (this will take a while, phoneME doesn't build quickly--though a bit more quickly than all of openmoko :P ) Jul 10 20:02:12 XorA|gone: here it failed on alsa: | /home/hrw/devel/build/autoconf261/tmp/cross/lib/gcc/arm-angstrom-linux-gnueabi/4.1.2/../../../../arm-angstrom-linux-gnueabi/bin/ld: .libs/libasound.so.2.0.0: undefined versioned symbol name snd_pcm_sw_params_get_start_threshold@ALSA_0.9 Jul 10 20:02:16 | /home/hrw/devel/build/autoconf261/tmp/cross/lib/gcc/arm-angstrom-linux-gnueabi/4.1.2/../../../../arm-angstrom-linux-gnueabi/bin/ld: failed to set dynamic section sizes: Bad value Jul 10 20:02:51 * hrw|gone back to tv Jul 10 20:03:02 bye hrw Jul 10 20:03:17 hrw|gone: alsa-lib built with no errors here Jul 10 20:03:42 hrw|gone: but that error doesnt sound very autoconfy Jul 10 20:05:43 ~seen ljp Jul 10 20:05:46 ljp was last seen on IRC in channel #openmoko, 9h 28m 44s ago, saying: 'jgm: I am getting the same error'. Jul 10 20:06:25 ~oemirrors Jul 10 20:06:27 somebody said oemirrors was extra, extra, read all about it, oemirrors is OpenEmbedded monotone mirrors are ewi546.ewi.utwente.nl, monotone.nslu2-linux.org, opensource.wolfsonmicro.com also see http://www.openembedded.org/wiki/GettingStarted Jul 10 20:09:19 jseghers??? now 3 people working on phoneme.bb? Jul 10 20:10:06 fine I started with cacao Jul 10 20:10:09 I couldn't find anyone else working on it Jul 10 20:10:15 *g* Jul 10 20:10:40 but you know you have to patch it that it works with qte-mt? Jul 10 20:11:14 I'm trying it just bsd/generic Jul 10 20:11:51 Are you working on Advanced or Feature? Jul 10 20:11:57 I specifically need Feature. Jul 10 20:12:43 jseghers are you in contact with henryk? Jul 10 20:12:53 no, I'm not Jul 10 20:13:07 he spent thewhole weekend on phoneme with feature Jul 10 20:13:14 hehe Jul 10 20:13:16 including tslib support Jul 10 20:13:46 I posted questions on the OpenMoko Devel board several times asking if anyone was working on it :) Jul 10 20:13:59 jseghers do the following Jul 10 20:14:08 made the bb until it works Jul 10 20:14:13 provide it to me Jul 10 20:14:29 and will shape it that it fits in oe and push it Jul 10 20:15:10 OK... I've got the .bb working (at least for the ARM target, not guaranteeing any other target at the moment) Jul 10 20:15:15 I didnt ask henryk if he made a .bb too Jul 10 20:15:28 And it AGAIN didn't get into the image...just into an ipkg. Jul 10 20:15:31 woglinde: i did Jul 10 20:15:37 hehe Jul 10 20:15:52 i'm currently in a meeting, back in 30min or so Jul 10 20:15:55 good to have all people for looking at java around Jul 10 20:15:59 hehe :) Jul 10 20:16:07 jseghers seems you doubled the work Jul 10 20:16:17 It would seem so. Jul 10 20:16:18 *sigh* Jul 10 20:16:43 hm but why the hell you and henryk only needs feature Jul 10 20:16:54 I guess I'll wait for henry to get out of his meeting :) Jul 10 20:16:54 thats what I am puzzling about Jul 10 20:16:59 sure Jul 10 20:17:26 I need Feature because I'm writing a low-level service that need to be exported to J2ME via a private class Jul 10 20:17:53 * timtimred beats openembedded.org with a stick Jul 10 20:17:53 The product gets build onto the phone by OEMs who also need to call the low-level C interface from their UI Jul 10 20:18:28 So OpenMoko provides me with a great development environment to get as close to what an OEM might normally work within. Jul 10 20:19:46 ~lart timtimred Jul 10 20:19:47 * ibot blasts timtimred to oblivion with a kamehameha wave Jul 10 20:19:48 So focusing on CLDC/MIDP 2 provides me with something much closer to what would be on a closed phone. Jul 10 20:20:11 hm does cldc really work with midp? Jul 10 20:20:23 I only tested advance Jul 10 20:20:33 Yep, that's the basis of MIDP Jul 10 20:20:42 CDC is the basis of Advanced Jul 10 20:20:58 yeh I know Jul 10 20:21:07 but midp in phoneme is extra feature Jul 10 20:21:48 phoneme-feature is actually PCSL + CLDC + MIDP + whatever JSRs Jul 10 20:22:18 The CLDC/MIDP pairing is basically a requirement for a feature phone. Jul 10 20:22:30 feature phones only run MIDLets Jul 10 20:22:48 yeah but default cdc is build without midp Jul 10 20:22:54 thats why I am aksing Jul 10 20:23:21 hm Jul 10 20:23:30 Right... That's the main difference between feature and Advanced. MIDP is the only applet-type interface for feature phones Jul 10 20:23:45 CDC gives you other, more capable, models for applets Jul 10 20:23:53 yes Jul 10 20:23:57 xlet Jul 10 20:24:07 okay now I got it Jul 10 20:27:34 I've been working on this integration attempt for over a week now... Jul 10 20:27:50 ;) Jul 10 20:28:01 finally gave up trying to use oe_make and just directly call make Jul 10 20:28:07 witin the .bb Jul 10 20:28:12 jseghers are you wokring on phoneme too? Jul 10 20:28:18 gtk support? Jul 10 20:28:24 not gtk... Jul 10 20:28:32 planning to run qvfb Jul 10 20:28:47 okay Jul 10 20:29:13 I've manually integrated and run it on the desktop inside of OpenMoko inside of Xephyr Jul 10 20:29:44 Ran jbenchmark on it and got a score of 40,000+...but that's on the desktop, not a phone :) Jul 10 20:30:25 because there there this project is too http://projects.openmoko.org/projects/java-pkg/ Jul 10 20:30:53 openmoko gonna end up with so much java it'll need a 1Gig flash Jul 10 20:31:02 xora no Jul 10 20:31:16 phoneme feature with all qtlibs is 3,8 megs Jul 10 20:32:31 woglinde: yeah, but kaffe, sable etc etc as well :-) Jul 10 20:32:48 hrms why one would install 6 java-vms? Jul 10 20:32:48 woglinde: and then there is OpenJDK Jul 10 20:32:58 woglinde: have you been in #openmoko? Jul 10 20:33:05 xora yes Jul 10 20:33:13 more crazy ideas than sense Jul 10 20:33:13 xora: for phone environment, J2ME is king :) Jul 10 20:33:21 xora because they dont have a clue Jul 10 20:33:29 hear hear Jul 10 20:33:39 jseghers: I mainten for 5 years a Javacard VM :-) Jul 10 20:33:42 it will be phoneme or cacao-cldc with midpath Jul 10 20:33:45 jseghers: Im sick of java Jul 10 20:34:05 oe needs gcj-native Jul 10 20:34:24 xora: I wish it would go away... BREW is a much better environment...but it just won't die Jul 10 20:34:25 woglinde: I think we have virtual/java-native? Jul 10 20:34:40 zecke I would look this Jul 10 20:34:41 * jseghers worked on J2ME and BREW games last 4 years Jul 10 20:34:49 but if its kaffe or sable Jul 10 20:34:51 woglinde: that needs some toolchain voodoo if I remeber Jul 10 20:34:56 that will not work Jul 10 20:35:02 woglinde: it wouldnt build so it was disabled Jul 10 20:35:12 *sigh* Jul 10 20:35:37 everyone hide, the zecke meister is around Jul 10 20:36:01 okay Jul 10 20:36:04 we have ecj Jul 10 20:36:13 thats enough Jul 10 20:39:32 Where in the linux hierarchy would you put the phoneme runtimes? It builds to three directories (pcsl, cldc, and midp) with various combinations of bin/inc/lib directories. I am trying to put it at /usr/java/... in my current attempts, and in the package, that's where they go...but my main stumbling block is getting them to be included in the Image. Jul 10 20:40:57 jsegehrs hm please look at lsb Jul 10 20:41:07 linux standard base if it says something Jul 10 20:41:44 where is lsb? Jul 10 20:42:35 never mind... found it Jul 10 20:42:38 mhm, if I load a kernel image via JTAG and tell uboot to boot it - "Uncompressing Kernel Image ... Error: inflate() returned -3" does not really make sense, right? I first thought my flash was corrupt Jul 10 20:42:48 any ideas what I may be missing? Jul 10 20:43:13 ...so I tried directly from memory but still same problem Jul 10 20:43:35 night all Jul 10 20:43:51 night.. Jul 10 20:44:22 Jin^eLD: loaded in uImage format? Jul 10 20:44:37 Jin^eLD: scratch that, just re-read what you put Jul 10 20:44:37 XorA|gone: yes.. I generated the uImage with mkimage in my kernel .bb Jul 10 20:45:05 uhm.. how? :) Jul 10 20:45:24 the device has no network, I guess that's the main part of the struggle Jul 10 20:45:51 Jin^eLD: you can send serial to uboot Jul 10 20:45:58 Jin^eLD: takes a while, but works Jul 10 20:46:02 jseghers google linux standard base could help Jul 10 20:46:09 Jin^eLD: first time i booted a neo was that way Jul 10 20:46:29 woglinde: Yep, found that. Jul 10 20:46:59 woglinde: It might say something about it...but not that I could find in a quick perusal. Jul 10 20:47:01 XorA|gone: so that feature is not some part that needs to be ported/adapted to the board, but this is something uboot'ish that will always work? Jul 10 20:47:07 let me try.. Jul 10 20:47:40 Jin^eLD: I think that is in core uboot, I thought I saw it on my smdk2443 board as well Jul 10 20:47:48 Jin^eLD: but I dont have uboot devices at home Jul 10 20:47:55 jeseghers cvm should go under bin Jul 10 20:48:32 jar's mabey under /usr/share/java Jul 10 20:48:34 indeed, load binary file over serial line Jul 10 20:48:44 jineld hehe Jul 10 20:48:54 :> Jul 10 20:49:01 simpad can this too Jul 10 20:49:12 XorA|gone: well, moment, how will that help me again? I want to dump the stuff back, not read it in, right? Jul 10 20:49:22 because I can load the image via JTAG as well and thats kind of faster in this case Jul 10 20:49:48 Jin^eLD: but not working, I thought you might want to try a differnt approach Jul 10 20:49:54 aah Jul 10 20:50:03 ok, now I get your idea :) Jul 10 20:50:27 Jin^eLD: if it still doesnt work then the image is at fault, if it does work, jtag download is screwed Jul 10 20:50:36 good point Jul 10 20:51:25 if I remember right, to serial download I issued the commands in terminal left terminal open then just cat uImage >/dev/ttyS0 Jul 10 20:51:43 you get lots of dots drawn while it downloads Jul 10 20:52:21 oh, so you did not even use kermit or alike to send the file Jul 10 20:52:27 woglinde: I'm trying to keep it somewhat localized :) trying not to have fingers spreading out all over the place...and a bunch of the scripts tend to reference things in the built hierarchy... Jul 10 20:52:50 woglinde: which was why I was thinking about perhaps /usr/java.... Jul 10 20:53:12 jsehers okay than lets try the debian approach Jul 10 20:53:17 But I'm relatively new to Linux :) Jul 10 20:53:31 and really don't know the conventions. Jul 10 20:53:37 all under /usr/lib/jvm/ Jul 10 20:53:56 OK... That would work. Jul 10 20:54:07 and than symlink bin/cvm to /usr/lib/jvm/phonemewhatever Jul 10 20:54:19 with update-altervanties Jul 10 20:54:23 args Jul 10 20:54:29 updates-alternatives Jul 10 20:55:41 Jin^eLD: from memory I think it was just raw binary Jul 10 20:55:51 Jin^eLD: but it was back in Nov Jul 10 20:56:07 the cat approach did not quite work for me, trying to figure this out with kermit Jul 10 20:57:24 woglinde: So, while referencing the package from RDEPENDS in local.conf go the package to build, it didn't include it in the rootfs. I'm now trying it not being referenced in local.conf and added it to PACKAGES in task-openmoko.bb... it's parsing .bb files now... we'll see if it a) builds the package and b) adds it to the image. Jul 10 20:58:21 seems to be uploading.. Jul 10 21:00:13 ltg is so bad monotone mirror was trying to connect from 19:44 to 22:01 Jul 10 21:00:52 * XorA|gone just killed that process Jul 10 21:02:29 hi laibsch Jul 10 21:02:49 * RP wonders what the bottleneck with the ltg server is :/ Jul 10 21:03:25 RP: mtn is eating all the RAM Jul 10 21:03:43 XorA|gone: ah :/ Jul 10 21:03:52 Its another bitbake Jul 10 21:04:08 RP: I was chatting to florian about round robinning with wolfson server which sits pretty idle all the time Jul 10 21:04:12 RP: to take some load off Jul 10 21:04:20 XorA|gone: That would be handy... Jul 10 21:04:23 RP: I think we organise this at guadec Jul 10 21:04:41 XorA|gone: I think you just volunteered to write an agenda ;-) Jul 10 21:04:49 XorA|gone: well done :) Jul 10 21:05:14 RP: ok, I will post an email tomorrow then Jul 10 21:05:50 RP: btw, another bit in the autoconf puzzled, I added m4-native 1.4.8 (DEF_PREF = -1) as its needed by some macros Jul 10 21:06:04 RP: if the servers ever sync you'll see it Jul 10 21:06:13 XorA|gone: ah, right, handy to know Jul 10 21:09:07 good evening Jul 10 21:09:13 so, back Jul 10 21:09:39 Hi Henry... sounds like we're trying to do similar things :) Jul 10 21:10:52 WOW, load of 125! I never knew linux could go that high Jul 10 21:11:44 XorA|gone, RP: top shows a bunch of apache2 processes as consuming the RAM Jul 10 21:12:47 Laibsch, it can go higher :) Jul 10 21:12:56 jseghers: yeah, over the course of the last weekend i massaged phoneme feature to compile for openmoko (i need it for a demo tomorrow) Jul 10 21:13:04 Laibsch: i`ve seen near 200 Jul 10 21:13:15 jseghers: I run it with directfb and tslib for touchscreen-input Jul 10 21:13:20 i have a screenie of >300 right here! Jul 10 21:13:29 but LA is somewhat unrelated to real load :] Jul 10 21:13:35 my 200 was caused by stuck hdd Jul 10 21:13:39 indeed :) Jul 10 21:14:29 henryk: that sounds great! Is it running on ARM hardware and/or emulation? or on Desktop? Jul 10 21:14:44 jseghers: on the openmoko, arm hardware Jul 10 21:14:59 (runs in neo1973 qemu too) Jul 10 21:15:14 * Laibsch restarts the web server Jul 10 21:15:21 Let's see if that helps at all Jul 10 21:15:29 you lrg Jul 10 21:15:35 -u Jul 10 21:15:47 henryk: I'm trying to get a build running for qemu since I don't have a neo yet (ordered three yesterday :) Jul 10 21:15:53 hey XorA|gone Jul 10 21:16:14 XorA|gone: Pizza Express Jul 10 21:16:15 jseghers: since qemu is designed to run the same image it just works Jul 10 21:16:32 XorA|gone: same result over serial, so I guess my image is borked Jul 10 21:16:32 Could I get your .bb and patches? I've been unable to get my build to be included in the image :( Jul 10 21:16:49 XorA|gone: but thanks for the hints! Jul 10 21:16:55 jseghers: image integration is another point, yes Jul 10 21:17:08 especially the serial thing is cool don't know why I did not notice it in the first place Jul 10 21:17:21 henryk: you haven't been able to do that either? Jul 10 21:17:51 jseghers: basically I introduced a new image target that depends on the packages i want Jul 10 21:19:18 jseghers: for some reason we documented it in german: http://sarwiki.informatik.hu-berlin.de/Eigene_Pakete_einbinden Jul 10 21:19:52 henryk: I *may* be having luck at the moment... I added the phoneme package to RDEPENDS_task-openmoko-linux in task-openmoko.bb... it's building the package... but I don't know if it will make it into the image or not. Jul 10 21:20:34 night everyone Jul 10 21:21:34 jseghers: image installation seems to be controlled by PACKAGE_INSTALL Jul 10 21:26:22 jseghers: I don't have a svn in place, so for now i only have a tarball: http://people.openmoko.org/henryk/openmoko-sar-20070710.tar.bz2 Jul 10 21:27:03 (We want to have this as a project here at the university, seems like it would be easiest for us to just have a project on top of openmoko) Jul 10 21:35:39 ltg and everything associated with it should be back to normal for now. Jul 10 21:35:48 Let's hope it stays that way Jul 10 21:36:14 henryk: sounds good... I'll pull yours down. If I add my package to PACKAGE_INSTALL in openmoko-devel-image, it just seems to ignore it during the build. Jul 10 21:37:37 ljp: hi Jul 10 21:38:38 jseghers: I just did a quick translation: https://sarwiki.informatik.hu-berlin.de/Embed_Custom_Packages Jul 10 21:38:44 * XorA|gone wonders if we should organise a GUADEC handbag tournament Jul 10 21:42:23 evening Jul 10 21:44:57 henryk: thank you! I was getting the gist of it from Google translation, but that will be useful. Jul 10 21:45:46 henryk: Whey do you get the appropriate Qtopia SDK? I can never seem to find anything but their latest which doesn't seem to be what openmoko wants. Jul 10 21:46:16 er.. "Where do you..." Jul 10 21:47:21 jseghers: qte is in openembedded Jul 10 21:47:32 i'm not even sure i need that dependency anymore. Jul 10 21:47:39 henryk: thanks :) Jul 10 21:47:49 as i said, i don't build against qvfb. Jul 10 21:49:11 henryk: I had also opted for building the generic network stuff instead of qt... I didn't know if it was going to work correctly, but thought I'd try that :) Jul 10 21:49:17 I see you chose the same. Jul 10 21:49:40 jseghers: oh, you'll need jdk 1.4.* to build and maybe change the path in the .bb. I didn't bother to do this correctly for now. Jul 10 21:50:00 I do have 1.4.*... I'll check the path Jul 10 21:52:00 jseghers: there was not much 'choosing' involved. At first I built pcsl with bsd/qte for network (following the instructions on https://phoneme.dev.java.net/content/mr2/pcsl_feature.html but then later on (can't remember if it was cldc or midp) I got an error message that said that I was supposed to rebuild with bsd/generic networking Jul 10 21:52:30 jseghers: I have to relocate now, will be back online in 1h Jul 10 21:53:21 hehe... I tried to build with qte and found that my Qtopia installation was 4.2.3... and things like QSocket didn't exist any more. Jul 10 21:53:46 jseghers yes Jul 10 21:54:02 only qt3 or qte2 is supported Jul 10 21:54:02 If qte is part of openembedded...I suppose I could have found and pointed it there... Jul 10 21:54:11 qte-mt Jul 10 21:54:18 but you have to patch the Makefile Jul 10 21:54:27 woglinde: never could find out how/where to install qt3 Jul 10 21:57:00 * * OE Bug 2637 has been created by utx(AT)penguin.cz Jul 10 21:57:02 * * fixes to make Sharp CE-RH2 remote working in X Jul 10 21:57:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2637 Jul 10 22:02:35 henryk: I followed the instructions, but I'm getting an error that Nothing provides dependency openmoko-sar-devel-image. Jul 10 22:03:18 henryk: oops.. hadn't flushed my edit buffer for local.conf Jul 10 22:03:34 * jseghers is off to a meeting Jul 10 22:17:42 Laibsch: Probably webmtn ones :/ Jul 10 22:18:30 I wonder if its something silly like a robot idexing webmtn :-( Jul 10 22:23:35 hi Jul 10 22:24:38 hi Jul 10 22:34:27 gnumeric, abiword, minimo all compile with autoconf 2.61 + all their deps of course Jul 10 22:36:15 XorA|gone: not bad Jul 10 22:39:46 epdfview so thats poppler chain Jul 10 22:44:19 morning Jul 10 22:46:08 rwhitby: morning Jul 10 22:50:59 re Jul 10 22:51:29 oh, morning rwhitby Jul 10 22:56:16 henryk: I'm getting errors that it can't find packages rxvt-unicode and mono Jul 10 22:57:02 jseghers: yes, just comment them out from openmoko-sar-devel-image. I like to have these packages installed in my image, but somehow they don't get built as dependency Jul 10 22:57:40 they're completely non-related to the j2me stuff, just some test for me whether building my own image works Jul 10 22:59:23 henryk: Ok... removed and trying the build again :) Jul 10 23:50:15 jseghers: does it build? Jul 11 00:13:40 henryk: I'm running into a compiler error in qte Jul 11 00:15:53 henryk: I seem to remember I'd had to make sure I had gcc 3.x Jul 11 00:16:10 but I'm forgeting where I tell it to *use* gcc 3.x Jul 11 00:18:11 or perhaps I'm thinking of phoneme itself... Jul 11 00:20:48 henryk: the error actually seems to be that it's trying to run "moc" from the target toolchain Jul 11 00:22:31 I'm getting /home/moko/build/tmp/work/armv4-linux/qte-2.3.10-r49/qt-2.3.10/bin/moc: Command not found Jul 11 00:22:58 When trying to run moc on kernel/qthread_unix.cpp Jul 11 00:25:17 actually... there's nothing in that bin directory... Jul 11 00:25:35 cd \ Jul 11 00:25:48 (wrong window...sorry) Jul 11 00:54:55 henryk: Yes!!! it built. I just had to remove the DEPENDS for qte. Jul 11 00:55:13 henryk: Thank You! Jul 11 01:24:37 henryk: It's running. It is taking over the entire screen when I run "usertest" but I saw in your build where you set the frame-buffer size. I don't suppose it's easy to make the frame buffer resize, or not be on top of other windows.... Jul 11 01:53:59 * * OE Bug 2638 has been created by r.farina(AT)adelphia.net Jul 11 01:54:02 * * ZD1211RW kernel drivers Jul 11 01:54:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2638 Jul 11 01:54:36 :-) Jul 11 02:21:09 Hello, im attempting to install angstrom on my zaurus but i want to install it to my sd card and boot directly to it using a new bootstrap image Jul 11 02:21:15 i've not quite found any clear instructions on how to do this so would i just get the bootstrap-image using the bitbake recipe and then make some modifications? **** ENDING LOGGING AT Wed Jul 11 02:59:57 2007