**** BEGIN LOGGING AT Mon Jun 02 02:59:56 2008 Jun 02 03:02:17 * rwhitby wonders who mickeyl was referring to in his latest blog post .... Jun 02 05:51:44 Crofton, Crofton|work, any chance you submitted that virtual/mta patch? Jun 02 08:09:17 bonjour :) Jun 02 08:26:35 Noobean: Jun 02 08:26:38 ops Jun 02 08:40:07 morning Jun 02 08:44:42 morning all Jun 02 08:45:46 03tmbinc 07org.oe.dreambox * r30179561... 10/ (3 files in 3 dirs): linux-dm7025: put firmware loader into kernel Jun 02 08:45:50 03tmbinc 07org.oe.dreambox * r8af6e62a... 10/ (1 packages/linux/linux-dm7025/dm7025_defconfig): linux-dm7025: enable autofs modules, build msdos,fat,vfat as modules Jun 02 08:45:53 03tmbinc 07org.oe.dreambox * r4cde9535... 10/ (4 files in 4 dirs): initscripts-opendreambox: load zd1211b module Jun 02 08:45:57 03tmbinc 07org.oe.dreambox * r77f8deb1... 10/ (3 files in 3 dirs): netbase: shut down wpa_supplicant after deconfiguring network interfaces Jun 02 08:46:00 03tmbinc 07org.oe.dreambox * r5a4c98ec... 10/ (4 files in 3 dirs): zd1211b: add zd1211b driver Jun 02 08:46:04 03tmbinc 07org.oe.dreambox * r98a7faa6... 10/ (1 conf/machine/dm7025.conf): conf/machines/dm7025.conf: include autofs and vfat Jun 02 08:51:34 03trini 07org.oe.stable * r889b3b71... 10/ (1 packages/mtd/mtd-utils_1.0.0+git.bb): Jun 02 08:51:34 mtd-utils: install and package headers Jun 02 08:51:34 Acked-by: Koen Kooi Jun 02 08:51:34 Acked-By: Marcin Juszkiewicz Jun 02 09:05:37 http://tinderbox.openembedded.net/builders/hrw-angstrom/ looks UGLY Jun 02 09:06:57 that's what SHE said! Jun 02 09:06:59 * Kryczek hides Jun 02 09:10:41 03rpurdie 07org.oe.dev * rad7898de... 10/ (31 files in 3 dirs): linux-rp: Add 2.6.25+2.6.26-rc4 version, making it the default for qemuarm, qemux86 and spitz so far (from poky) Jun 02 09:10:45 03rpurdie 07org.oe.dev * r57b770a7... 10/ (1 packages/v86d/v86d_0.1.3.bb packages/v86d/v86d_0.1.5.bb): v86d: 0.1.3 -> 0.1.5 Jun 02 09:10:50 03koen 07org.oe.dev * ra6a18f67... 10/ (8 files in 3 dirs): Jun 02 09:10:50 pkgconfig: sync with poky Jun 02 09:10:50 * pkgconfig 0.23 added, disabled by default Jun 02 09:10:50 * fiddled with 0.22 SRC_URI to to keep the .inc sane Jun 02 09:11:52 hi Jun 02 09:13:06 03tmbinc 07org.oe.dreambox * rcca70242... 10/ (1 packages/images/dreambox-image.bb): dreambox-image: add wlan support to dm7025, add zd1211b support Jun 02 09:17:15 03koen 07org.oe.dev * r6d1714dd... 10/ (1 packages/angstrom/angstrom-zeroconf-audio.bb): angstrom zeroconf audio: add more RDEPENDS Jun 02 09:25:49 morning Jun 02 09:26:02 hi methril Jun 02 09:28:34 angstrom 2008.1 is not buildable :( Jun 02 09:30:24 hrw whats wrong? Jun 02 09:31:07 http://tinderbox.openembedded.net/builders/hrw-angstrom/ Jun 02 09:32:29 hm why is dbus failing Jun 02 09:32:39 maybe old expat Jun 02 09:32:42 no build logs? Jun 02 09:33:09 I will look - that mega build was started few days ago Jun 02 09:33:21 I suspect ~180GB of disk space used Jun 02 09:33:30 ah there Jun 02 09:33:43 Explicitly requested expat but expat not found Jun 02 09:35:23 something broke expat as 1.2.1-r1 was built before Jun 02 09:36:07 * * OE Bug 4352 has been created by meduzapat(AT)hotmail.com Jun 02 09:36:09 * * Ipaq 2215 99% CPU all the time Jun 02 09:36:11 * * http://bugs.openembedded.net/show_bug.cgi?id=4352 Jun 02 09:37:48 /home/hrw/devel/build/angstrom/tmp/cross/lib/gcc/arm-angstrom-linux-gnueabi/4.2.2/../../../../arm-angstrom-linux-gnueabi/bin/ld: warning: libgcc_s.so.1, needed by /home/hrw/devel/build/angstrom/tmp/staging/arm-angstrom-linux-gnueabi/usr/lib/libexpat.so, not found (try using -rpath or -rpath-link) /home/hrw/devel/build/angstrom/tmp/staging/arm-angstrom-linux-gnueabi/usr/lib/libexpat.so: undefined ... Jun 02 09:37:53 ... reference to `__aeabi_unwind_cpp_pr0@GCC_3.5' /home/hrw/devel/build/angstrom/tmp/staging/arm-angstrom-linux-gnueabi/usr/lib/libexpat.so: undefined reference to `__aeabi_unwind_cpp_pr1@GCC_3.5' Jun 02 09:37:58 thats why it fails Jun 02 09:38:02 yes Jun 02 09:38:20 I saw that on the gta01 build in your logs too Jun 02 09:38:23 hm Jun 02 09:38:27 gcc-3.4 error? Jun 02 09:38:32 ups Jun 02 09:38:34 4.3 Jun 02 09:39:21 arm-angstrom-linux-gnueabi-gcc (GCC) 4.2.2 Jun 02 09:40:01 hm Jun 02 09:40:16 ./armv5te-angstrom-linux-gnueabi/gcc-cross-4.2.2-r9.do_build Jun 02 09:40:19 I will test gcc-4.3 and uclibc and then gcc 4.2.4 Jun 02 09:40:23 no other gcc-cross built Jun 02 09:44:51 moin! all back home well from LinuxTag ? Jun 02 09:45:01 uh hi Jun 02 09:45:19 which booth did you maned? Jun 02 09:45:38 maned ? Jun 02 09:45:50 run? Jun 02 09:46:02 * hrw did not maned any booth this time Jun 02 09:46:06 or was you only a visitor Jun 02 09:46:31 woglinde: I am rebuilding expat now Jun 02 09:46:36 hrw hm Jun 02 09:46:40 helo. Jun 02 09:46:41 ops Jun 02 09:46:42 thats not the error I guess Jun 02 09:46:50 hrw its in gcc somewhere Jun 02 09:46:51 woglinde: i was part of the openSUSE-community crew Jun 02 09:47:07 * * OE Bug 4353 has been created by meduzapat(AT)hotmail.com Jun 02 09:47:09 * * sysinfo hang Jun 02 09:47:11 * * http://bugs.openembedded.net/show_bug.cgi?id=4353 Jun 02 09:47:25 *g* I am not a big fan of suse Jun 02 09:47:47 neither am I Jun 02 09:47:47 but I am tolerant Jun 02 09:47:50 woglinde: thats ok - there are plenty distros out there ;) Jun 02 09:47:58 but I had opensuse install in vbox one time Jun 02 09:48:22 now i'm back working on/with OE ... Jun 02 09:51:39 I'm having troubly building bootstrap-image of angstrom-2008.1 with glibc. fails, anybody with the same experience? NOTE: package glibc-2.6.1-r6: task do_install: failed Jun 02 09:52:05 markus please use pastbin Jun 02 09:52:08 ~pastebin Jun 02 09:52:09 [~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 Jun 02 09:52:49 markus: last week before heading to LT i had this, too - just no time to look at it, yet Jun 02 09:55:36 woglinde: yep - with rebuilt expat other parts breaks Jun 02 09:59:30 http://pastebin.com/m17e340b2 <-- here is local.conf, output of bitbake bootstrap-image and error log Jun 02 10:01:52 markus hm uh Jun 02 10:01:55 never saw this Jun 02 10:02:38 but why the hack this happens cannot find /lib/libc.so.6 inside /home/markus/openembedded/build/tmp/staging/arm-angstrom-linux-gnueabi Jun 02 10:03:19 maybe i need to "make clean" and then build again or something Jun 02 10:05:40 hm maybee Jun 02 10:05:41 du in angstrom build is still going... Jun 02 10:05:51 started over half hour ago Jun 02 10:08:28 du in angstrom build? Jun 02 10:10:11 builddir Jun 02 10:10:39 so Jun 02 10:10:52 angstroem uclibc with gcc and libtool 2.4.2 Jun 02 10:10:57 aeh 2.2.4 Jun 02 10:11:12 gcc 4.3.0 Jun 02 10:11:27 i don't get it.. Jun 02 10:13:35 ??? Jun 02 10:13:59 hrw: koen has reported the same gcc 3.5 symbol issues Jun 02 10:14:13 RP: good to know Jun 02 10:14:16 hrw: Its due to mixing armv4567 in staging Jun 02 10:14:28 urgs Jun 02 10:14:41 rp only glibc issue? Jun 02 10:16:01 so Jun 02 10:16:13 now I will make me some steaks Jun 02 10:16:17 for lunch Jun 02 10:16:20 till later Jun 02 10:19:42 RP: just like I thought Jun 02 10:20:05 RP: so it looks like arm456 works (Poky) but arm4567 does not Jun 02 10:20:39 03koen 07org.oe.dev * r08bc1e87... 10/ (1 packages/libtool/libtool-cross_2.2.4.bb): libtool cross 2.2.4: check for dolt.m4 before trying to install it Jun 02 10:22:15 rp libtool-native dont compile here, with the following error http://de.pastebin.ca/1036783 Jun 02 10:22:27 2.2.4 Jun 02 10:23:52 woglinde: see the checkin just 2 min ago ... probably related ?! Jun 02 10:25:28 dl9pf maybee but seems install for m4 macros Jun 02 10:25:39 hmmm, true Jun 02 10:26:59 woglinde: you need automake 1.10 Jun 02 10:27:12 rp oh okay Jun 02 10:27:20 koen hit that too... Jun 02 10:27:33 hm than versione dependency for automake Jun 02 10:27:43 aeh libtool-2.2.4 Jun 02 10:27:55 I will fix that Jun 02 10:37:34 03koen 07org.oe.dev * r7feb68cc... 10/ (4 files in 3 dirs): glib-2.0: add 2.16.3 from poky, patched to work with libtool 2.2.4, disabled by default for that reason Jun 02 10:40:26 woglinde: http://rafb.net/p/HB1Sn757.txt Jun 02 10:40:42 woglinde: The versions he's needed to change so far... Jun 02 10:44:43 RP: I use OE to build only for one target, and so I usually disabled multimachine to get shorter paths below tmp/ Jun 02 10:44:54 hm Jun 02 10:45:05 versions dep arent functional Jun 02 10:45:29 schurig: ah :/ Jun 02 10:45:55 schurig: I was hoping to simplify things a bit... Jun 02 10:46:22 RP: I'm not really against it, just that you consider this in your decision Jun 02 10:46:38 schurig: One way forward might be the creation of a singletarget.bbclass ? Jun 02 10:46:39 RP: you're surely a multimachine man, but I guess that many people just have only one Jun 02 10:47:05 hm in which .bbclass DEPENDS is handeld? Jun 02 10:47:19 RP: what would be an option Jun 02 10:47:38 schurig: I use multimachine heavily, yes Jun 02 10:47:42 woglinde: bitbake Jun 02 10:47:48 RP: but what buys you this? As soon as someone sets DISTRO = angstrom, he is on multimachine (except if he twiddles the angstrom conf files) Jun 02 10:49:03 RP: so I think that multimachine is already the default :-) Jun 02 10:49:20 schurig: We're getting more and more subtle assumptions about multimachine being used in the metadata like the SDK stuff I mentioned Jun 02 10:49:49 schurig: meta-toolchain won't work properly unless multimachine is in use for example Jun 02 10:49:58 * Crofton|work is working on three machines atm, multimachine saves my but Jun 02 10:50:00 can you fix that with metaclass.oeclass: INHERIT += "multimachine" and a comment why multimachine is needed? Jun 02 10:50:10 schurig: Its the default for Angstrom and Poky but not everyone uses those distros Jun 02 10:50:42 schurig: multimachine has to be applied globally, not just in some classes Jun 02 10:50:45 I do not count amount of machines Jun 02 10:51:00 yeah, Angstrom and Poky are meant to be run on a wide range of machines. My "distro" is meant to run on exactly one machine. And I guess this is true for some other specialized uses. Jun 02 10:51:38 anyway, I could live with long filesnames ... as long as OE stays on monotone, I'll have my own private fork anyway. So I don't care that much ... :-) Jun 02 10:51:49 12:51 hrw@home:angstrom$ ls tmp/stamps/|wc Jun 02 10:51:53 194 194 5695 Jun 02 10:52:14 schurig: Its much easier to collapse OE from multimachine to single machine than the other way around. Creating a singlemachine.bbclass shouldn't be hard Jun 02 10:52:37 RP: okay, the go for multimachine by default Jun 02 10:53:02 schurig: I do understand what you mean about the pathnames though Jun 02 10:53:06 RP: Ack from me too Jun 02 10:53:44 ~curse big builds to take lot of space Jun 02 10:53:45 May you be reincarnated as a Windows XP administrator, big builds to take lot of space ! Jun 02 10:53:54 RP: yeah, if your paths end up longer than > 120 characters, they look clumsy in Konsole :-) Jun 02 10:53:54 11:28 hrw@home:~$ time du -hs devel/build/angstrom/ Jun 02 10:54:03 its 12:53 now and it is still working.. Jun 02 10:54:17 hrw: get a faster disk then :-) Jun 02 10:54:21 schurig: export PS1="\[\e[36m\]\A \[\e[31m\]\u\[\e[34m\]@\[\e[33m\]\h\[\e[36m\]:\[\e[32m\]\W\[\e[35m\]\$\[\e[0m\] Jun 02 10:54:32 schurig: 75MB/s is not enough? Jun 02 10:55:05 hrw: then use a better filesystem :-) Jun 02 10:55:13 I am used to ext3 Jun 02 10:55:16 hrw: or a ramdisk (cought, cought) Jun 02 10:55:24 used reiserfs, xfs and do not want to go back to them Jun 02 10:55:39 hrw: no, I also didn't really like reiserfs Jun 02 10:55:47 schurig: donate me 512GB RAM + hosting for it Jun 02 10:57:29 hm yes version dependcies isnt supported right Jun 02 10:57:38 **sigh* Jun 02 10:57:59 woglinde: Not yet. You can add it to the recipe though Jun 02 10:58:07 woglinde: bitbake just doesn't parse it :/ Jun 02 10:58:11 rp yes Jun 02 10:58:24 Adding parsing to at least error about incompatibilities shouldn't be too hard Jun 02 10:58:25 rp but bitbake should support it Jun 02 10:58:38 maybee I will hack it Jun 02 10:58:42 woglinde: totally agreed, patches welcome Jun 02 10:58:58 I've not written those patches for good reason :/ Jun 02 10:59:02 (its hard) Jun 02 10:59:18 as far as I see its done in cache.py Jun 02 10:59:29 hm Jun 02 11:01:07 woglinde: Thats where its parsed. You then need to act on the info somewhere Jun 02 11:01:29 You then see the mess that is providers.py and run away screaming Jun 02 11:01:51 I'm slowly improving that code hopefully Jun 02 11:25:36 okay ncurses Jun 02 11:25:45 needs love Jun 02 11:33:31 hm maybee sed error Jun 02 11:35:01 03koen 07org.oe.dev * rd19ecf6f... 10/ (4 files in 3 dirs): libusb: add libtool patch from poky, safe for general consumption Jun 02 11:40:49 2h for "du -hs devel/build/angstrom/" and still counting... Jun 02 11:47:27 hrw: uh ... what ? Jun 02 11:48:09 dl9pf: lot of small files... and probably over 200GB of them Jun 02 11:54:20 03koen 07org.oe.dev * rc0b4eb64... 10/ (1 packages/gstreamer/gst-plugins.inc): gst-plugins.inc: remove esound dependency Jun 02 11:54:20 03koen 07org.oe.dev * rc0b4eb64... 10/ (1 packages/gstreamer/gst-plugins.inc): gst-plugins.inc: remove esound dependency Jun 02 11:54:20 03koen 07org.oe.dev * rc0b4eb64... 10/ (1 packages/gstreamer/gst-plugins.inc): gst-plugins.inc: remove esound dependency Jun 02 11:54:30 03koen 07org.oe.dev * rc0b4eb64... 10/ (1 packages/gstreamer/gst-plugins.inc): gst-plugins.inc: remove esound dependency Jun 02 11:56:51 03crofton 07org.oe.dev * rd05a5685... 10/ (1 packages/linux/linux-omap1-2.6.25-omap1/defconfig): Jun 02 11:56:51 linux-omap1 : Use defconfig from kernel source. This boots in Angstrom, but Jun 02 11:56:51 does not have tmpfs or sound enabled. Attempts to add these Jun 02 11:56:51 lead to hanging during boot. Work continues ... Jun 02 12:33:40 03koen 07org.oe.dev * recb1d938... 10/ (10 files in 3 dirs): linux-omap2 git: enable smart reflex driver for beagleboard Jun 02 12:36:56 ok, angstrom took only 156GB of space Jun 02 12:37:09 and 174 minutes for "du -hs" to measure it Jun 02 12:48:16 heh Jun 02 12:52:09 03koen 07org.oe.dev * r221e5d30... 10/ (1 packages/freetype/freetype_2.3.5.bb): freetype: sanitize do_configure Jun 02 12:52:11 and result was broken totally Jun 02 13:27:36 hehe koen is in good mood Jun 02 13:27:53 hi Jun 02 13:27:54 re Jun 02 13:28:00 hi likewise Jun 02 13:28:17 hi woglinde, florian, et al! Jun 02 13:28:35 florian: thanks for the LinuxTag presence (and all who joined in)! Jun 02 13:29:47 how it was ? Jun 02 13:29:54 gremlin hm okay Jun 02 13:30:19 seems the day as I was there, there arenot many specific oe questions Jun 02 13:30:32 only normal thinks like what we do and so on Jun 02 13:31:17 as usual :) Jun 02 13:32:02 mhhh news about OE foundation ? Jun 02 13:32:16 yes Jun 02 13:32:40 all done ... we are officially a foundation ? Jun 02 13:33:04 hm in germany that isnt so fast Jun 02 13:33:12 we have many burocreaatics Jun 02 13:33:55 mhhh ok at least italy isn't the slowest bureaucratics :P Jun 02 13:34:15 hi likewise Jun 02 13:34:18 hehe but you have siesta or something Jun 02 13:35:06 likewise: yw... yes I have to share the credit with woglinde and pH5 Jun 02 13:35:51 and thesing Jun 02 13:35:54 gremlin[it]: yep, from what I know France and Germany are worse Jun 02 13:36:35 damnit where is the blogpost with the laywer meeting Jun 02 13:36:36 i really thought we (italy) what the worse .... Jun 02 13:37:05 * florian thinks LinuxTag was quite some success - quite some people with valid interest or knowing OE already. Jun 02 13:37:20 ah right Jun 02 13:37:28 that was only a mail from mickeyl Jun 02 13:38:05 03koen 07org.oe.dev * rdec1e557... 10/ (4 files in 3 dirs): cairo 1.6.4: add libtool patch Jun 02 13:38:23 gremlin -> http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-June/005472.html Jun 02 13:38:57 thanks woglinde Jun 02 13:46:03 hi kgilmer Jun 02 13:46:07 morning Jun 02 13:46:35 hi mickeyl Jun 02 13:46:40 gm Jun 02 13:46:47 yo florian Jun 02 13:46:49 hey crofton Jun 02 13:47:16 mickeyl: How many people asked if they are affected by your last blog post? ;) Jun 02 13:47:28 hehehe Jun 02 13:47:32 actually, not that many ;) Jun 02 13:47:38 or how many just assumed mickeyl hated them :-) Jun 02 13:47:51 i was careful not to have it included on planet.openmoko.org ;) Jun 02 13:47:59 hi mickeyl Jun 02 13:48:05 mickeyl, I find that after a certain age, patience increases again Jun 02 13:48:06 hey gremlin[it] Jun 02 13:48:12 er Jun 02 13:48:16 Chemistry? Jun 02 13:48:17 Crofton|work: haha, ok, i hope I'll soon be there ;) Jun 02 13:48:24 ya Jun 02 13:48:30 * hrw connected 16 devices to n810 Jun 02 13:48:34 insane a bit Jun 02 13:48:41 oh, I thought you meant linuxtogo :) Jun 02 13:49:11 mickeyl: one unversal in life is you will never be old enough :-) Jun 02 13:49:20 or young enough Jun 02 13:49:41 hehe Jun 02 13:49:57 best you can do is get old and date young Jun 02 13:50:12 mmh, ya ;) Jun 02 13:54:56 mickeyl, hiya Jun 02 13:55:21 hello Jun 02 13:55:28 how can I gather more info for bug http://bugs.openembedded.net/show_bug.cgi?id=4350 Jun 02 13:55:28 ? Jun 02 13:55:36 hey Sup3rkiddo Jun 02 13:55:45 that reminds me Jun 02 13:56:03 * florian too Jun 02 13:56:22 kgilmer: its really nice to see that your company goes 'the right way' ;) Jun 02 13:56:46 kgilmer: I linked http://buglabs.net/license from https://wiki.maemo.org/index.php/100Days#Get_official_statement_about_Nokia.27s_stance_towards_software_freedom Jun 02 13:57:11 in the hope that this will help them to provide a similar document Jun 02 13:57:16 hi robert Jun 02 13:57:27 hi henning. sorry we did not meet a LT Jun 02 13:57:37 I was not there on friday & sunday Jun 02 13:57:39 robert yes Jun 02 13:58:06 Did you guys all check out the beagle both? Jun 02 13:58:13 Did anyone go to the beagle talk? Jun 02 13:58:24 woglinde: partying with my girlfriend at immergut festival was more important ;) Jun 02 13:58:48 what kinda geek are you :-D Jun 02 13:58:53 rschuster yes sebastian told me that Jun 02 13:59:05 Crofton|work: no but I was at the beagleboard booth and asked them all kinds of question Jun 02 13:59:09 Crofton|work: I mised the talk, but had a nice chat with the TI people at the booth... Jun 02 13:59:38 Crofton|work: I was a bit surprised ... Jun 02 13:59:45 by? Jun 02 14:00:09 Crofton|work: wasn't TI supposed to be another 'we lock everything down and are happy with this' company? Jun 02 14:00:25 Crofton|work: and took the direct route and asked about NDAs, specs and all that Jun 02 14:00:32 it looks like they are trying to change that Jun 02 14:01:09 Crofton|work: they pointed me to the CDs lying on their. +3000 pages worth of specification ready to be taken home Jun 02 14:01:28 yeah Jun 02 14:01:42 * Crofton|work wishes they had paper copies Jun 02 14:01:52 Things are changing as the OMAP TRMs are open for OMAP2/3 now Jun 02 14:01:56 Crofton|work: that would look more impressive Jun 02 14:02:01 I really wish I hadn't chucked my Motorla 6800 docs Jun 02 14:02:11 that was a managebale amount of docs Jun 02 14:02:16 RP: what does 'TRM' mean? Jun 02 14:02:21 RP, agreed Jun 02 14:02:27 Techniical reference manual Jun 02 14:02:42 The useful one :) Jun 02 14:02:53 not the marketing speak Jun 02 14:03:14 RP: for omap3 they opened. for omap2 iirc not quite Jun 02 14:03:15 anyway, I need to go look for a connector to make the OTG port into a host port on my beagle Jun 02 14:03:23 some stuff will require money/is of non-free nature. like the spec for the DSP capable of video decoding stuff Jun 02 14:03:28 hrw: omap2 TRM is public Jun 02 14:03:32 ah Jun 02 14:03:39 but there showed willingness to fix this Jun 02 14:03:41 rschuster: omap3 video stuff will be open Jun 02 14:03:46 s/there/they/ Jun 02 14:03:54 dsp tools are a headache atm Jun 02 14:03:55 rschuster: omap2 3d video stuff is closed Jun 02 14:03:59 no free compiler Jun 02 14:04:09 hrw: can live with that Jun 02 14:04:17 Crofton|work: :( Jun 02 14:04:28 Lets hope things continue to improve Jun 02 14:04:31 hrw: it is bad but not as bad as having no specification for battery mgmt Jun 02 14:04:35 they know they need some solution to the porblem though Jun 02 14:05:17 RP: yes, we are starting with a say 80% open device. that is a good start Jun 02 14:05:27 http://mailman.tu-chemnitz.de/pipermail/gcc-c6x/2007-March/000031.html Jun 02 14:05:43 is as close as I have found to c64 support in gcc Jun 02 14:05:57 apart from that I don't know any available armv7 device that is such open at all Jun 02 14:07:02 hello woglinde and rschuster. I've been out of the office with xchat running... Jun 02 14:07:20 kgilmer: hi ken Jun 02 14:09:15 rschuster: thanks! I'll send this around internally. It's always important to show the results/response of trying to do things 'the right way' :) Jun 02 14:10:43 03koen 07org.oe.dev * r74076120... 10/ (1 packages/libtiff/tiff_3.8.2.bb): tiff: update to 4.0.0beta2 for libtool fixes Jun 02 14:11:08 kgilmer: I cannot open the link you sent me containing the host newkeds.buglabs.net Jun 02 14:11:21 kgilmer: perhaps it is behind the firewall? Jun 02 14:11:24 rschuster: hmm Jun 02 14:11:38 odd Jun 02 14:11:44 works for me...what do you get? Jun 02 14:11:56 rschuste@hagen:~$ ping newkeds.buglabs.net Jun 02 14:11:56 ping: unknown host newkeds.buglabs.net Jun 02 14:12:27 Host newkeds.buglabs.net not found: 3(NXDOMAIN) Jun 02 14:12:58 this is from university (not through Berlin's public meshnet which is sometimes flaky with those things) Jun 02 14:13:49 rschuster: ok Jun 02 14:13:57 maybe it's an internal DNS thing Jun 02 14:14:07 kgilmer: have an IP for me? Jun 02 14:14:37 rschuster: 64.145.49.107 Jun 02 14:15:08 rschuster: I'll check with our sysadmin about newkeds. does that work? Jun 02 14:15:09 regarding the beagleboard. I assumed koen is doing paid work for this but he is doing that voluntarily. crazy dude! Jun 02 14:15:34 'I will work for board' is common for students Jun 02 14:15:51 but board is 149usd... Jun 02 14:16:08 kgilmer: works perfect Jun 02 14:16:14 kgilmer: may I post the link here Jun 02 14:16:17 ?= Jun 02 14:16:34 kgilmer: woglinde might want to have a look at the recipes too Jun 02 14:16:44 kgilmer: absolutely :) Jun 02 14:16:48 oops Jun 02 14:16:50 I'm kgilmer Jun 02 14:16:51 sheesh Jun 02 14:17:14 ken gillmer of bug labs fame is bringing sun's phoneme to openembedded: Jun 02 14:17:14 http://64.145.49.107/cgi-bin/viewvc.cgi/com.buglabs.build.oe/meta-bug/packages/phoneme-advanced/?root=cvs Jun 02 14:17:32 (and I will help him with it) Jun 02 14:18:39 kgilmer: no problem. everyone talks to himself sometimes :) Jun 02 14:18:41 rschuster: awesome! Yes I wasn't sure if I should submit it to OE as I couldn't get it to build...but I think it was more of a config issue with OE rather than a problem w/ the recipie. Jun 02 14:18:53 * RP is pleased to see this coming into OE Jun 02 14:19:16 rschuster: I flubbed the URL. newkeds is only for us internally. the proper URL is cvs.buglabs.net Jun 02 14:19:45 btw. I did a talk on OE at LinuxTag. If someone wants the slides: http://www.fsfe.org/fellows/robertschuster/data/cross_compilation_with_openembedded Jun 02 14:19:59 btw speaking about beagleboard. it was nice surprise to hear 'it use Angström distribution Jun 02 14:20:07 btw speaking about beagleboard. it was nice surprise to hear 'it use Angström distribution' on big vendor stand Jun 02 14:20:11 rschuster: we are also working on GTK support to pMEA. There is already some work in that direction. Jun 02 14:20:30 I like to stress the fact that OE is very flexible and surpasses all other projects in this regard Jun 02 14:21:00 kgilmer: oh, that sounds great. Jun 02 14:23:01 * RP would love to see gtk support since that would open up its use in poky Jun 02 14:23:56 RP: yep. on the phoneme forums a guy posted patches against gtk 1.x that compiles and partly works. Jun 02 14:24:30 kgilmer: I hope these patches are several years old :-) Jun 02 14:24:47 kgilmer: You are aiming for gtk+ 2.x I hope? Jun 02 14:25:39 florian, RP: yep. honestly I know next to nothing about GTK. But 2.x is the goal. Support in Poky is probably 'the deliverable' Jun 02 14:26:08 kgilmer: Trust us, you want 2.x ;-) Jun 02 14:26:17 RP: I trust you :) Jun 02 14:26:46 btw: plan to blog about moving to OE/Poky this week. will post back here when it's done. Jun 02 14:27:07 www.bugblogger.com Jun 02 14:27:12 :-) Jun 02 14:27:25 kgilmer: cool :) Jun 02 14:27:38 nice ;) Jun 02 14:28:06 * mwester agrees with multimachine being the default. Jun 02 14:28:10 Mornin' Jun 02 14:28:46 hi mwester Jun 02 14:29:50 morning Jun 02 14:31:12 hi chouimat|work Jun 02 14:37:05 03koen 07org.oe.dev * rd4f98623... 10/ (10 files in 4 dirs): hal: sync with poky Jun 02 14:41:16 kgilmer: what is phoneme's latest released version? your usage of 0.0 they did not have any Jun 02 14:43:30 rschuster: yeah, good question Jun 02 14:43:37 they have a strange versioning scheme Jun 02 14:43:45 i was goign ot ask on the forums but forgot Jun 02 14:43:52 they use build numbers and milestone releases Jun 02 14:44:02 so MR80 build 72 Jun 02 14:44:05 etc Jun 02 14:44:18 not sure how to map that into the OE/Ubuntu/Gentoo scheme. Jun 02 14:45:43 rschuster: I'll ask the phomeme guys how to map that in and get back to you. Jun 02 14:45:46 and would MR80 build 73 be bigger? Jun 02 14:45:55 rschuster: AFAIK Jun 02 14:46:27 rschuster: but I also think they have a notion of "1.0" and they are not there yet. but that is not expressed in the "global" versioning system. Jun 02 14:46:31 mmh Jun 02 14:46:37 * mickeyl seens poky sync breaking hal and dbus again Jun 02 14:46:42 ~lart koen Jun 02 14:46:43 * ibot teaches koen the basics, including how to RTM Jun 02 14:46:43 kgilmer: rschuster: rember if you get it wrong you can always use PE to recover when a release happens Jun 02 14:47:09 XorA: PE? Jun 02 14:47:17 Package Epoch Jun 02 14:47:33 XorA: we have package epoch in OE?? Jun 02 14:47:36 so 1.0 would become 1:1.0 Jun 02 14:47:40 rsalveti_: yes Jun 02 14:47:43 rschuster: yes Jun 02 14:47:54 kgilmer: OE's versioning system is primarily debian based Jun 02 14:48:04 and this is supported by ipkg, dpkg and rpms? Jun 02 14:48:05 rschuster: was put in there mainly to recover from X.orgs wierd versions Jun 02 14:48:25 rschuster: ipkg/opkd/dpkg certainly Jun 02 14:48:32 * RP has no clue how it maps to rpms Jun 02 14:48:56 * rschuster thinkgs that Debian should consider moving to OE. that would make their m68k builds easier :) Jun 02 14:49:11 rschuster: almost certainly Jun 02 14:50:39 XorA: Ok good to know regarding PE. rschuster: I will post in the phoneme forums and get back. Jun 02 14:51:03 kgilmer: thanks Jun 02 14:53:25 m68k isn't a release arch in Debian any more. Jun 02 14:56:24 no more amiga debian :-( Jun 02 15:07:17 03koen 07org.oe.dev * rebf18a64... 10/ (3 files in 3 dirs): esound: clean configure of 0.36 a bit Jun 02 15:54:09 http://blog.haerwu.biz/2008/06/02/feel-the-power-of-usb-with-nokia-tablet/ Jun 02 16:03:42 03koen 07org.oe.dev * r0e0a01cf... 10/ (5 files in 3 dirs): hal: revert PIDFILE hunk Jun 02 16:03:47 03koen 07org.oe.dev * r62afea71... 10/ (4 files in 3 dirs): libfakekey: fix configure Jun 02 16:03:52 03koen 07org.oe.dev * r04ee23d7... 10/ (7 files in 3 dirs): ncurses: fix libtool issues with libtool 2.2.4, safe for general consumption Jun 02 16:22:05 hms Jun 02 16:22:16 koen changes somehow generate this Jun 02 16:22:31 "ncurses-native.do_rm_work_all" -> "glib-2.0-native.do_rm_work" Jun 02 16:22:40 dependcy Jun 02 16:29:35 03koen 07org.oe.dev * rd9f2d609... 10/ (3 files in 3 dirs): gtk 2.12.3: merge libtool patch with poky Jun 02 16:46:04 03koen 07org.oe.dev * r764a3b1e... 10/ (1 packages/images/beagleboard-demo-image.bb): beagleboard demo image: add image description of the demo used atLinuxTag 2008 Jun 02 17:03:24 re: BB_STAMP_POLICY, i still think itd be useful to figure out a generic mechanism for encoding package configuration into the output binary package. course you'd probably need a new binary pacakge manager to make use of such a thing, but.. :) Jun 02 17:20:22 finally my build worked, and glibc did not fail. Jun 02 17:20:39 i realized that i built outside my build dir so i have rebuilt everything Jun 02 17:21:02 can i just delete my build/tmp and move tmp to build/tmp ? Jun 02 17:21:50 since i ran bitbake bootstrap-image from openembedded/ and not openembedded/build Jun 02 17:22:45 markus: sure Jun 02 17:22:55 wait Jun 02 17:23:02 if i would've run bitbake bootstrap-image from my build-dir it would have failed i think. Jun 02 17:23:36 not sure if moving will work flawlessly Jun 02 17:23:52 to work around that problem you should make TMPDIR absolute Jun 02 17:25:08 hmm? Jun 02 17:25:30 if i continue to build fron openembedded/ and not openembedded/build, then it will work maybe Jun 02 17:25:50 but my local.conf is in openembedded/build/conf/ Jun 02 17:26:01 if you make TMPDIR fixed you can build from any dir in the future Jun 02 17:27:05 is that an environment variable? Jun 02 17:30:31 no, bitbake variable.. i set that in local.conf Jun 02 17:30:49 ok, i change tmpdir to /path/to/openembedded/tmp then Jun 02 17:34:42 you can try to move your bigger tmp/ dir to the new location. the worst thing that can happen is that you will get a strange compile error and have to start from scratch Jun 02 17:35:32 the thing is.. i want to use my new misplaced tmp Jun 02 17:35:43 since that one is fresh Jun 02 17:57:49 rp ping Jun 02 17:59:28 rp ping. Jun 02 17:59:41 ops i get these... ticks.. Jun 02 18:19:22 re cyril Jun 02 18:19:37 hey woglinde Jun 02 18:19:39 morning all Jun 02 18:52:41 03crofton 07org.oe.dev * r089195e6... 10/ (1 packages/linux/linux-davinci_2.6.x+git-davinci.bb): linux-davinci : Update to git rev e23e6d5c446074082d93285f21cf95ed0d537a6f.` Jun 02 19:23:25 Is anyone alive? Jun 02 19:23:34 yes Jun 02 19:25:18 A__A: no I'm dead for the moment Jun 02 19:25:53 Ah! Wonder of wonders. The last time I used IRC, there was no "always-online" broadband. I can't get used to hundreds of people being in a channel and maybe no one actually at a screen. I am about to flash my Zaurus and need some general advice. Jun 02 19:25:55 markus: you can also set TOPDIR to fix the build directory Jun 02 19:28:56 markus: I think it defaults to current directory in bitbake Jun 02 19:31:29 A__A: basic human behavior remains constant. You're far less likely to get a response from anyone in a crowd unless you ask a specific question! Jun 02 19:40:13 hm intressing Jun 02 19:40:43 gcc - 4.2.2 reports an not found gmp.h not as error Jun 02 19:40:48 gcc-4.3.0 does Jun 02 19:44:22 hi thesing Jun 02 19:44:31 hi everybody Jun 02 19:50:31 morning Jun 02 19:50:33 hi thesing Jun 02 19:50:49 thesing: I heard that you have nearly everything working on collie/2.6 Jun 02 19:51:30 hrw: thats the short marketing answer ;) Jun 02 19:52:04 there is no sound driver yet and mmc, usb is not really stable. Jun 02 19:52:14 but its works for me (tm) Jun 02 19:53:44 ;) Jun 02 19:53:53 hi hrw Jun 02 20:03:37 morning Jun 02 20:03:43 hi ant Jun 02 20:03:56 woglinde: hey Jun 02 20:04:01 thesing: ping Jun 02 20:04:13 ant_: pong Jun 02 20:04:17 hallo Jun 02 20:04:36 I put your work on the RP's stage yesterday... Jun 02 20:04:46 (about initramfs) Jun 02 20:04:58 he agrees we need a new discussion on the ML Jun 02 20:05:13 to coordinate your work with pfalcon one Jun 02 20:05:37 actually I'm trying the bootmenu-image and I'd need your advice Jun 02 20:05:59 I have an (old ?) copy of your patches Jun 02 20:06:18 and now I miss 'expr' Jun 02 20:06:30 ant_: Paul didn't use the function from kernel.bbclass so I already committed the initramfs stuff. Jun 02 20:06:31 kergoth: Its a nice idea from a theory standpoint ;-) Jun 02 20:06:46 kergoth: Until someone has time to implement it though... Jun 02 20:06:50 aye Jun 02 20:07:09 you can build kernels with build-in initramfs by setting INITRAMFS_IMAGE to your favorite image. Jun 02 20:07:12 my question is about 30-bootmenu.sh Jun 02 20:07:27 it contains a couple of '==' Jun 02 20:07:47 still after your patches Jun 02 20:08:05 they are probably bashisms Jun 02 20:08:17 * thesing is not shell guru. Jun 02 20:08:54 and as I said init cannot find 'expr' optarg=`expr "x$arg" : 'x[^=]*=\(.*\)'` Jun 02 20:09:12 I remeber you added 'wc' Jun 02 20:09:54 I think it would need 5 mins work (from you :-) Jun 02 20:10:05 looks very very promising Jun 02 20:10:06 is expr supposed to be build into sh ? Jun 02 20:10:27 busybox has it Jun 02 20:10:42 (it's there, paul put it there) Jun 02 20:11:21 == doesn't work in dash Jun 02 20:11:36 already corrected ;-) Jun 02 20:11:42 RP: we don't even have dash we have klibc ash ;) Jun 02 20:12:04 rp any idea why koen changes pull glib2 dep into ncurses-native Jun 02 20:12:08 second problem: no logs in console Jun 02 20:12:19 woglinde: not offhand Jun 02 20:12:23 helgeD: hi Jun 02 20:12:25 ant_: maybe you can replace expr with the [ ] stuff. Jun 02 20:12:30 depends and task-depends dont tell it Jun 02 20:12:39 its just their out of nowhere Jun 02 20:12:41 RP: You saw my mail? Jun 02 20:12:41 helgeD: Why can't we autoload the wifi crypto modules? Jun 02 20:12:47 helgeD: yes :) Jun 02 20:12:57 hms Jun 02 20:13:07 RP: No idea. It worked with other kernels before, but 2.6.24 has some problems. Jun 02 20:13:11 Maybe mis-compilation? Jun 02 20:13:17 gcc-4.3.0 makes trouble with gmp Jun 02 20:13:23 helgeD: What kind of problem? Jun 02 20:13:48 thesing: other problem: /initrd.d/30-bootmenu.sh 170: arith: syntax error: "cnt +1" Jun 02 20:13:50 helgeD: Is the problem the modules aren't loaded, won't load or ...? Jun 02 20:14:00 RP: modules were renamed Jun 02 20:14:01 RP: It's always "ieee80211_crypt_tkip: could not allocate crypto API arc4" Jun 02 20:14:19 helgeD: and if you modprobe arc4 first? Jun 02 20:14:33 RP: yes, should do, but doesn't. Jun 02 20:14:59 helgeD: Doesn't do what? The module doesn't load or it still can't find it? Jun 02 20:15:22 module loads. Everything looks ok. But arc4 crypto is not recognized. Jun 02 20:15:31 Even "cat /proc/crypto" does not show arc4 Jun 02 20:15:40 (or any other crypto module) Jun 02 20:16:00 ant_: as for logs I think some of "echo ... > CONSOLE " lines clear the screen. Jun 02 20:16:06 ah.. blockcypher-crypto-module Jun 02 20:16:12 so crpyto in 2.6.24 is totally bust. Surely there is a fix in mainline for that? Jun 02 20:17:46 RP: I didn't found something. I tried to debug it. No idea. Jun 02 20:18:00 thesing: I'm not sure this can work for klibc ash: cnt=$((cnt + 1)) Jun 02 20:18:10 RP: But, why is it working if built-in, but not as module? Jun 02 20:18:12 helgeD: Which device are you using? Jun 02 20:18:30 helgeD: I suspect something just isn't being loaded Jun 02 20:18:32 prism2-cf-based Jun 02 20:18:38 RP: it's not just me. Jun 02 20:18:51 ant: compile klibc for x86 linux and test on your box. Jun 02 20:18:55 helgeD: Which machine? Jun 02 20:19:03 RP: sorry, I'm using hostap Jun 02 20:19:08 RP: Poodle 5600 Jun 02 20:19:16 ant: maybe expr is needed for that. Jun 02 20:19:26 * mwester vaguely recalls similar problems with 2.6.24 and SlugOS; it was a poorly-documented module dependency problem. Jun 02 20:19:42 helgeD: ok, I was going to ask if 2.6.26-rc4 worked but that won't on poodle yet most likely Jun 02 20:20:11 mwester: Any recollections of the problem would be most welcome :) Jun 02 20:20:51 lets see if it works on spitz in 2.6.26-rc4... Jun 02 20:22:54 ieee80211_crypt_tkip loads fine with 2.6.26-rc4... Jun 02 20:23:11 Do you have ecb.ko modloaded? Jun 02 20:23:20 thesing: http://en.pastebin.ca/1037187 30-bootmenu.sh Jun 02 20:24:07 If I flash my Zaurus, and things go wrong, how do I restore it? That's to say, is it possible to "brick" my Zaurus by flashing? Jun 02 20:24:24 A_A: which zaurus? Jun 02 20:24:28 thesing: http://en.pastebin.ca/1037190 init Jun 02 20:24:42 SL-C3200. Or any. Jun 02 20:24:50 RP: "cat /proc/crypto" ? Does it lists "ecb(arc4)" ? Jun 02 20:25:37 helgeD: That depends if I load arc4 or not. the tkip module doesn't seem to use it Jun 02 20:25:56 helgeD: To get any further with this I suspect I'd need a 2.6.24 kernel Jun 02 20:25:58 A_A: http://www.trisoft.de/en_c3200howto.htm Jun 02 20:26:09 (not sponsored link) Jun 02 20:26:16 Yes. & WPA need it. Jun 02 20:26:27 RP: Do you have WPA to test? Jun 02 20:26:34 helgeD: no :/ Jun 02 20:26:39 hmm... Jun 02 20:26:42 :-( Jun 02 20:27:04 ant_: I've been there, but I want some reassurance from actual people who've done it. Jun 02 20:28:15 don't worry for your nand-flash, mine has survived thousend reflashes ! Jun 02 20:28:31 A_A: The 3200 is pretty much impossible to totally brick since it has recovery stuff in ROM Jun 02 20:29:09 rp what was the difference btw. inherit and require? Jun 02 20:29:45 woglinde: require includes a .inc file, inherit is for classes Jun 02 20:29:54 hm Jun 02 20:30:14 gcc-4.3.0 shows a fundamental error with gcc-cross-inital Jun 02 20:30:55 ant_: no idea you might try to find out how to do something like this line with dash Jun 02 20:31:34 RP: that's the kind of thing I want to hear. Is it a case of some "reset" after things go wrong and you get back to Sharp's original Japanese base system? Jun 02 20:33:15 waha gcc is a mess Jun 02 20:35:33 bye Jun 02 20:36:29 nite hrw Jun 02 20:36:31 A_A: yes, as long as you have a recovery image Jun 02 20:36:46 woglinde: you meant between include and require ? Jun 02 20:37:10 Kernel 2.6.24.4 -- ath0 is up and running with WPA. What's the question we need to answer on this? Jun 02 20:38:46 mwester: ath? Atheros? Do you by any chance have a hostap/prism2 as well to test? Which machine is this? Jun 02 20:39:09 RP: I don't want my existing Sharp OS. If the flash of Angstrom goes wrong, would there be anything stopping me (brick wise) from trying another distro? Or do I have to "Restore" in between flashes? Jun 02 20:39:20 helgeD: Which modules need to be loaded to get WPA running? Jun 02 20:39:38 A_A: Probably not Jun 02 20:40:01 RP: I reported that once in http://bugs.openembedded.org/show_bug.cgi?id=3686 as well. Jun 02 20:40:10 mwester: sorry, that comment was directed to you! Jun 02 20:41:23 This is a DSM-G600 (ixp4xx-based device). I have a zd1211 USB dongle about, but that's about it. Here's the lsmod output: http://www.pastebin.ca/1037226 Jun 02 20:42:02 rp do you have some time to look at the gcc .bb Jun 02 20:42:19 I think because of the lots of require Jun 02 20:42:34 DEPENDS for gcc-cross-inital gets lost somewhere Jun 02 20:42:48 mwester: Hmm, thats using wlan, not the ieee80211 stack by the look of it :( Jun 02 20:42:51 RP: On my fedora laptop (2.6.26-rc): af_packet, arc4, ecb, mac80211, maybe others. Jun 02 20:42:59 * RP doesn't know enough about wireless stuff :/ Jun 02 20:43:08 for gcc 4.2 configure only reports a warning for not having gmp.h Jun 02 20:43:14 Right. Ok, I can easily reflash a slug. Give me a few minutes... Jun 02 20:43:17 but gcc-4.3.0 reports an error on this Jun 02 20:43:18 I've just tried to get my router to talk WPA and I've broken my wifi access entirely Jun 02 20:43:26 RP: :-) Jun 02 20:43:34 ant_: try something like $((cnt=$cnt+1)) Jun 02 20:44:09 thesing: will try, thx Jun 02 20:44:20 * ant_ was lurking in /klibc-1.5/usr/dash/ Jun 02 20:44:21 ant_: I meant cnt=$(($cnt+1)) Jun 02 20:44:52 this sort of expressions work here. Jun 02 20:45:22 * helgeD is back in 20mins. Jun 02 20:45:42 thesing: now this hostile one optarg=`expr "x$arg" : 'x[^=]*=\(.*\)'` Jun 02 20:45:48 woglinde: looking at Gentoo ebuils, it seems that gcc4.2.x depends on gmp when built with fortran, whereas gcc4.3 strongly depends on gmp Jun 02 20:45:52 woglinde: help ^^^ Jun 02 20:46:05 ant? Jun 02 20:46:17 http://en.pastebin.ca/1037190 Jun 02 20:46:50 we don't have 'expr' in shell, how would you rewrite that Jun 02 20:46:55 ? Jun 02 20:46:58 cyrilromain when I look at the gcc recipe gcc-cross-inital shoud have DEPENDS on gmp-native and gmp-mpfr Jun 02 20:47:21 cyrilromain but there are so many require that it is lost somewhere Jun 02 20:48:39 ant cant you add expr support to the shell? Jun 02 20:48:47 would be the easiest Jun 02 20:49:05 eh, I'm reading /klibc-1.5/usr/dash/ Jun 02 20:49:08 It seems for this stuff you need expr. Jun 02 20:49:31 thesing hm you could rewrite it using shift Jun 02 20:50:55 right. Then this is the way to go. Jun 02 20:51:04 We need to save every bit we can. Jun 02 20:51:28 but the question is have you shift support *g* Jun 02 20:52:02 we have. Jun 02 20:52:22 but shift only works for $1, $2 etc, right? Jun 02 20:52:51 so we would rewrite the line as a new script. Jun 02 20:53:27 woglinde: indeed, gmp-native and mpfr-native are added to DEPENDS by gcc-cross4.inc, included by gcc-cross_4.3.0.bb Jun 02 20:53:51 cryil now try bitbake -g gcc-cross-inital Jun 02 20:53:56 and look at the deps Jun 02 20:54:04 initial Jun 02 20:54:15 always forgot the last i Jun 02 20:54:54 woglinde: thx a lot Jun 02 20:55:00 thesing yes Jun 02 20:55:51 cyrilRomain, pnig Jun 02 20:55:53 *ping Jun 02 20:56:25 cyrilRomain, have you submitted the virtual/mta patch yet? ;) Jun 02 20:56:45 Tartarus: pong Jun 02 20:56:58 Tartarus: nop :-/ Jun 02 20:57:01 rschuster1 ? Jun 02 20:57:53 morning all Jun 02 20:58:32 woglinde: now that I think about it how will shift help? we need a way to get from the string "Abcd=efg" the "efg" Jun 02 20:58:35 ah I got it Jun 02 20:59:09 thesing: woglinde: there are more in 30-bootmenu.sh Jun 02 20:59:28 cyrilRomain, aw, what's the hold-up? ;) Jun 02 20:59:29 more 'expr' Jun 02 20:59:40 cross-inital.inc Jun 02 20:59:46 is the root of failure Jun 02 21:00:43 woglinde: heh indeed Jun 02 21:00:43 ok we need expr Jun 02 21:01:04 woglinde: s/DEPEND =/DEPENDS +=/ Jun 02 21:01:35 cryil nope Jun 02 21:01:41 is not that easy Jun 02 21:01:53 hi florian Jun 02 21:02:12 * florian feels better with a bigger screen and keyboard Jun 02 21:03:32 florian hehe Jun 02 21:04:46 RP, helgeD_bbl: http://pastebin.com/d757b92ab Jun 02 21:06:11 mwester: thanks. helgeD: Can you make sure all that lot are loaded please? Jun 02 21:06:16 okay Jun 02 21:06:50 easiest way for add the gmp and mpfr -native deps to cross-init_4.3.0.bb Jun 02 21:06:53 hi khem Jun 02 21:07:22 woglinde: hello Jun 02 21:08:09 thesing: scripts/functions: Replace expr use with printf for skipping comments on /etc/modules. Works on both busybox ash and klibc dash. Prefix space is ignored by ... Jun 02 21:08:19 woglinde: yep. But isn't 'DEPENDS =' a problem when it's in both gcc-cross-initial.inc and gcc-4.3.0.inc ? Jun 02 21:08:44 cryil no Jun 02 21:09:09 cryil only esthetic Jun 02 21:09:13 for now Jun 02 21:09:30 some one should stand up and fix it saner Jun 02 21:09:34 the whole recipes Jun 02 21:12:40 cyril the problem is the require are there to dont duplicate code Jun 02 21:13:03 but when you look at the recipes there are a lot of duplicate code Jun 02 21:13:40 woglinde: I understand that. But isn't all 'DEPENDS =' and 'DEPENDS +=' before the one in gcc-cross-initial.inc ignored ? Jun 02 21:13:53 cryil right Jun 02 21:13:56 ant_: where is that from? Jun 02 21:14:17 http://changelogs.ubuntu.com/changelogs/pool/main/i/initramfs-tools/initramfs-tools_0.69ubuntu20/changelog Jun 02 21:14:37 cyril but gcc-cross-initial.inc depends binutils Jun 02 21:14:45 woglinde: hm ok, so it only 'looks' wrong Jun 02 21:14:50 and the put the most of the needed packages in for gcc-inital Jun 02 21:15:03 the=that Jun 02 21:15:17 besides mpfr and gmp Jun 02 21:15:23 -native *g* Jun 02 21:18:17 ant_: it doesn't help us Jun 02 21:18:36 google can help: +klibc +replace +expr Jun 02 21:18:40 ;-) Jun 02 21:18:46 *could Jun 02 21:22:37 RP: I'll try to build a kernel with your defconfig-poodle from -dev and try the modules as mwester mentioned. Jun 02 21:23:28 RP: It's late, so I'll try the next few days... Jun 02 21:23:44 mwester: thanks for the info. Jun 02 21:24:10 woglinde: hm adding 'DEPENDS += "gmp-native mpfr-native"' in a gcc-cross-initial4.inc would solve the problem for any gcc-cross-initial_4.x. But as you said someone has to fix it saner in whole recipes. Jun 02 21:24:31 no problem -- I remember fighting with this myself; at least one of those modules changed in 2.6.24, and the change was very poorly documented. Jun 02 21:24:47 woglinde: there is probably a better way to reorganise them.. Jun 02 21:25:04 is there an equivalent to Poky's meta-toolchain-sdk in OE? Jun 02 21:28:35 cyril I will check in the fix for now Jun 02 21:30:32 RP: btw, when doing a diff (-stable to -dev) of the files packages/linux/linux-rp.inc and packages/linux/linux-rp_2.6.24.bb you have reverted a lot of stuff.... Jun 02 21:30:55 helgeD: Depends what you define as "revert" Jun 02 21:31:15 e.g. "Check the kernel is below the 1264*1024 byte limit for the PXA Zaurii" Jun 02 21:31:29 helgeD: That was merged into kernel.bbclass Jun 02 21:31:38 * cbrake gives meta-toolchain-* in OE a try ... Jun 02 21:32:29 RP: oic. Ok, forget my comment. Jun 02 21:32:48 RP: I think I just pull over your new defconfig-poodle and try that with 2.6.24 from -stable. OK? Jun 02 21:33:37 helgeD: ok Jun 02 21:52:01 hi :) can anyone tell me if PR and PV can contain letters as well as numbers? Jun 02 21:54:28 ant_: klibc doesn't have regular expressions so we can't add expr. Jun 02 21:55:38 kgilmer -> http://oe.linuxtogo.org/wiki/StyleGuide Jun 02 21:55:40 the smaller I found atm is: http://www.krugle.org/examples/p-nrrcuNO3IxPObGNz/expr.c Jun 02 21:55:51 but #include "busybox.h" Jun 02 21:56:07 kgilmer and http://oe.linuxtogo.org/wiki/VersioningPolicy Jun 02 21:56:50 thesing/ant you could have a look at dietlibc maybee this helps your space problem Jun 02 21:57:00 ant_: I suggest writing a small tool in C for the things we need. (for example we need sth. like "tools abc=edf =" returning "edf". Jun 02 21:57:24 woglinde: I thing dietlibc is already to big. And its not in OE. Jun 02 21:58:27 * ant_ is down to 126254b for .cpio.gz (but lacks expr...) Jun 02 22:00:45 03woglinde2 07org.oe.dev * r79b32dd1... 10/ (1 packages/gcc/gcc-cross-initial_4.3.0.bb): Jun 02 22:00:45 gcc-cross-inital-4.3.0: fix DEPENDS Jun 02 22:00:45 * in gcc-4.2 a missing gmp.h is just a warning in gcc-4.3.0 it is an error Jun 02 22:00:45 * gcc-cross-inital.inc is overwriting DEPENDS so gmp-native and Jun 02 22:00:45 mpfr-native from gcc-cross4.inc is not in DEPENDS for gcc-cross-initial Jun 02 22:00:47 * for now put in the .bb self, but someone should step up and fix the Jun 02 22:00:49 whole gcc-recipes in a saner way Jun 02 22:00:53 03koen 07org.oe.dev * rd0dad3c0... 10/ (4 files in 3 dirs): libsndfile: fix configure.ac Jun 02 22:00:55 03koen 07org.oe.dev * r9c471eb1... 10/ (4 files in 2 dirs): fftw*: upgrade to 3.2alpha3 Jun 02 22:10:19 ant_: another solution might be to rewrite the whole init and bootmenu stuff in C. One could start from klibcs kinit. Jun 02 22:10:37 thesing hehe Jun 02 22:10:50 that would speed up a lot Jun 02 22:11:06 but lost some generalism Jun 02 22:11:06 RP: currently booting the new kernel... Jun 02 22:11:32 and when we are at it we could use some shiny gui. Jun 02 22:12:00 I don't know if there are small libs for framebuffer access. Jun 02 22:13:33 thesing: at this point would be faster to write fb for u-boot (zaurus) Jun 02 22:14:35 I found a guy with something ready (for pxa27X only) Jun 02 22:14:48 ant_: no this would be more low-level. Jun 02 22:14:56 joking Jun 02 22:15:12 too many zaurus models Jun 02 22:16:41 the other solution would be triple jump: boot->kexec->glibc bootmenu (in /boot) Jun 02 22:17:41 no it would be boot (zImage with bootmenu) -> boot real zImage Jun 02 22:18:01 this would be the optimum, I know Jun 02 22:18:12 nite Jun 02 22:18:26 woglinde: : good night Jun 02 22:18:29 or in more detail: bootloader->zImage(bootmenu)--kexec-->zImage Jun 02 22:18:33 nite woglinde Jun 02 22:18:33 woglinde: night Jun 02 22:19:46 I think this is quite doable. But somebody has to do it. ;) Jun 02 22:20:11 yes, the damned initramfs-bootmenu must fit in the small partition... Jun 02 22:20:28 RP: ayt? I think I found something... Jun 02 22:21:06 if you rewrite this stuf in C. this will be no problem. Jun 02 22:21:49 could be unified for linux-rp and linux-hh probably Jun 02 22:22:24 yes should work for all linux machines. Jun 02 22:23:02 you did most of the work for linux-rp, please revamp the discussion on the ML Jun 02 22:23:46 angstrom list, so we'll get koen's and paul's attention Jun 02 22:25:08 thanks again, Thomas Jun 02 22:25:12 good nite Jun 02 22:25:16 *all Jun 02 22:25:23 nite. Jun 02 22:28:29 bye Jun 02 22:31:02 * helgeD_ sent mail to RP Jun 02 22:36:58 03koen 07org.oe.dev * ra0a0beb1... 10/ (1 packages/icu/icu-3.6.inc): icu: fix do_configure Jun 02 23:12:50 gnite Jun 03 00:23:57 Er, is there a way to get inetutils xinetd files easily, that i'm missing? Jun 03 00:55:07 * * OE Bug 4351 has been RESOLVED (INVALID) by xjqian(AT)gmail.com Jun 03 00:55:09 * * navit cvs migrated to svn; edit navit_cvs.bb Jun 03 00:55:11 * * http://bugs.openembedded.net/show_bug.cgi?id=4351 Jun 03 02:41:08 03khem 07org.oe.dev * r94057a9d... 10/ (53 files in 15 dirs): Fix an obvious typo in the patch. The variable exported was 'CCP' should have been 'CPP'. I have tested build for gcc-cross-4.2.2 recipe. **** ENDING LOGGING AT Tue Jun 03 02:59:57 2008