**** BEGIN LOGGING AT Mon Jul 19 02:59:56 2010 Jul 19 06:13:57 Hi! I want to use not last version of gcc, what settings should I use? Tried DEFAULT_PREFERENCE_ = "-1" at 4.4.4 and "1" at 4.1.2, but OE still build 4.4.4 Jul 19 06:25:26 gim, in local.conf add PREFERRED_VERSION_gcc = "4.1.2" Jul 19 06:25:47 btw any particular reason why 4.4.4 is no good for you ? Jul 19 06:29:29 good morning Jul 19 06:29:45 gim, I'm using PREFERRED_GCC_VERSION_local, with minimal and ANGSTROM_GCC_VERSION when building angstrom Jul 19 06:38:17 eFfeM_work: Thanks, it works. Reason -- u-boot in OE fails to build for LPC3250 with patches from NXP, but builds fine in LTIB framework with gcc 4.1.2, so I decided to check whether errors are because of gcc version. Jul 19 06:38:50 gim ah ok Jul 19 06:39:02 might be the oe u-boot is missing a patch Jul 19 06:40:43 /j #git Jul 19 06:41:03 oopsie Jul 19 06:41:06 hi, I have added some packages to the task-beagleboard-demo.bb. It's getting compiled but the binaries are not there in rootfs. Jul 19 06:41:13 eFfeM_work: Maybe, some errors I've found in web and fixed them successfully, but then linking errors arise. Actually I don't think it's because of gcc version, checking it just to be on the safe side Jul 19 06:41:30 good evening all Jul 19 06:41:39 hi IgorK, gm Jul 19 06:42:10 gim, linking is binutils (but could also be lib related), Jul 19 06:42:33 How can i resolve this? Jul 19 06:46:39 it looks like glibc recipe ha(d) the wrong reference to the "ld.so.conf" Jul 19 06:47:01 removing it from the source files reference fixes the build Jul 19 06:47:26 did anybody took care of this? Jul 19 06:47:38 IgorK: which version ? Jul 19 06:47:52 glibc_2.9.bb Jul 19 06:48:51 IgorK: did you pull after sat 12.00 gmt ? Jul 19 06:49:31 it was missing from the tree from friday evening to saturday morning (my bad) fixed it sat morning Jul 19 06:49:58 IgorK: you do not want to remove it from the recipe Jul 19 06:51:04 ok, it's now fixed? what I need to do now? Jul 19 06:52:58 git pull Jul 19 06:56:00 do I need to kill work and rebuild everything? how will it work? Jul 19 06:57:24 IgorK: no, just go into your openembedded dir and do a git pull Jul 19 06:57:41 no need to kill work just reissue the bitbake command that had the failing glibc build Jul 19 06:57:49 and apologies for the trouble caused Jul 19 06:59:41 so it will rebuild glibc and all dependency it built? Jul 19 07:02:50 IgorK: it wil rebuild glibc and continue with the build process Jul 19 07:03:59 IgorK: ah, i get what you mean, you removed the line and have dependent recipes; well that should not be a problem as the file is not used at link time, it is just at the target where you run into problems when the file is missing Jul 19 07:08:41 yes, I removed the line from the recipe and continued the build Jul 19 07:09:53 so ALL I need is "git pull", right? and then "bitbake -k "? but will it rebuild everything? Jul 19 07:23:56 ehm, no; you need "git pull" for the latest recipe. but as this was not a functional change I didn't bump PR, so you need to explicitly remove glibc Jul 19 07:24:21 IgorK: command: "bitbake -cclean glibc glibc-initial" Jul 19 07:24:58 then you can continue your build process; this file does not affect the generation of other packages so no need to clean these Jul 19 07:25:18 continue your build process == "bitbake console-image" or whatever you wanted to build Jul 19 07:25:28 ok, thanx Jul 19 07:25:43 apologies accepted :) Jul 19 07:25:52 feel free to ping me if you run into problems Jul 19 07:25:53 ty :-) Jul 19 07:33:13 will do Jul 19 07:52:45 morning Jul 19 07:53:33 has the way conffiles overrides are handled changed in the last weeks? an override for such a file now creates a *-opkg file, while it did overwrite the file some time ago Jul 19 07:59:09 eFfem_work: by which time do u live? Jul 19 07:59:18 IgorK: GMT+2 Jul 19 07:59:27 (NL) Jul 19 08:00:34 good morning Jul 19 08:00:44 hi florian Jul 19 08:04:04 good morning, florian Jul 19 08:04:52 eFfem_work: I saw that file added in the "git pull" statistics. I cleaned glibc* and restarted the build Jul 19 08:05:08 IgorK: great! good luck! Jul 19 08:05:09 will report in the morning Jul 19 08:05:46 IgorK: ok, I'm online for the next 6.5 hr, but will stay in the channel so feel free to drop a msg Jul 19 08:06:02 or in my evening I might be online as eFfeM Jul 19 08:12:53 hi, I have some problem with building rootfs, in stable/2009 Jul 19 08:12:54 hi, I have some problem with building rootfs, in stable/2009? Jul 19 08:13:57 packages listed in task-*.bb, is not present in rootfs, can I have any idea? Jul 19 08:17:36 03Khem Raj  07org.openembedded.dev * r3d3d49a20d 10openembedded.git/recipes/uclibc/ (uclibc-git/mipsel/uClibc.machine uclibc_git.bb): Jul 19 08:17:36 uclibc-git: Refresh config for mipsel and bump to latest git master. Jul 19 08:17:36 Signed-off-by: Khem Raj Jul 19 08:51:41 03Koen Kooi  07org.openembedded.dev * rf73986f74c 10openembedded.git/recipes/devicekit/devicekit-power_014.bb: devicekit-power 014: fix build for offline machines Jul 19 08:52:57 hello Jul 19 08:53:03 hi Jul 19 09:07:17 hi xora Jul 19 09:07:33 mornig Jul 19 09:08:14 hi hrw Jul 19 09:11:05 morning Jul 19 09:28:01 mmh have some trouble to build openembedded with angstrom debian for an arm9 atmel qemu-native_010.3.bb failed every time :( Jul 19 09:29:26 where can i paste the compile log? to get some help Jul 19 09:29:50 ~pastebin Jul 19 09:29:51 [~pastebin] A "pastebin" is a web-based service where you should paste anything over 3 lines so you don't flood 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://bin.cakephp.org/ , http://asterisk.pastey.net/ , or install pastebinit with yum or aptitude. Jul 19 09:30:10 *sigh* as I said install the debian qemu Jul 19 09:30:29 and set ASSUME_PROVIDED += "qemu-native" in local.conf Jul 19 09:30:42 are you again hi woglinde Jul 19 09:33:49 03Xerxes RĂ„nby  07org.openembedded.dev * r6916986de6 10openembedded.git/recipes/openjdk/ (3 files in 2 dirs): Jul 19 09:33:49 openjdk-6-6b18: add pre1.8.1 releasebranch fixes up to 13 Jul 2010. Jul 19 09:33:49 fixes origin from http://icedtea.classpath.org/hg/release/icedtea6-1.8/shortlog Jul 19 09:33:59 k i change the local.conf and test it now again Jul 19 09:34:11 thx again for your great support (help) Jul 19 09:37:18 i ve some next question sorry im a noob... where can i find what display the kernel supported? at the moment i use an LB043WQ1 from philips lg its on the eval board Jul 19 09:39:28 hm Jul 19 09:39:39 xorg and framebuffer? Jul 19 09:39:47 has it edid? Jul 19 09:39:55 which graphic chipset? Jul 19 09:43:36 its an at91sam9g45 Jul 19 09:44:25 prinzenrolle: sam9g45 has a screen on board or vga/dvi output? Jul 19 09:48:57 nope no vga or dvi an LCD controller supporting STN and TFT displays up to 1280*860 Jul 19 10:05:07 03Koen Kooi  07org.openembedded.dev * r576f600ec1 10openembedded.git/recipes/usbutils/usbmodeswitch_1.1.3.bb: usbmodeswitch: fix virtual/libusb0 Jul 19 10:08:47 hm, anyone an idea on this: Jul 19 10:08:48 ERROR: Nothing RPROVIDES 'eglibc' (but /home/frans/workspace/gtd/openembedded.git/recipes/eglibc/eglibc_2.9.bb RDEPENDS on or otherwise requires it) Jul 19 10:10:53 i would expect eglibc_2.9.bb to deliver eglibc Jul 19 10:12:00 its an RDPENDS so it depends what it PACKAGES Jul 19 10:17:31 XorA, i don't think my understanding of the build system is good enough to understand that Jul 19 10:18:02 didn't change anything in this area, confused why I get this Jul 19 10:18:31 PACKAGES indeed mentions glibc, not eglibc Jul 19 10:20:14 ah, think i got it Jul 19 10:21:27 phew! as those files are obtuse :-D Jul 19 10:22:51 yes Jul 19 10:29:53 03Martin Jansa  07org.openembedded.dev * r57e0fe3728 10openembedded.git/conf/distro/include/angstrom-jalimo.conf: Jul 19 10:29:53 angstrom-jalimo: fix openjdk-6-jre preferred version Jul 19 10:29:53 Signed-off-by: Martin Jansa Jul 19 10:29:53 Acked-by: Koen Kooi Jul 19 10:42:44 03Koen Kooi  07org.openembedded.dev * rf87932855c 10openembedded.git/recipes/powervr-drivers/bc-cube_0.2.0.bb: bc-cube: add libgles-omap3-x11wsegl to x11 rdepends Jul 19 10:47:52 ~curse orange Jul 19 10:47:53 May you be reincarnated as a Windows XP administrator, orange ! Jul 19 10:49:14 they are blocking all ports.... Jul 19 10:49:25 I'll have to use a ssh proxy for git Jul 19 10:49:28 too Jul 19 10:49:33 s/git/oe Jul 19 10:51:07 GNUtoo|laptop: that is the reason I setup OpenVPN :-) Jul 19 10:51:14 blocked too Jul 19 10:51:19 it's on port 443 udp Jul 19 10:51:26 mine is on 443 tcp Jul 19 10:51:26 only ssh -D is left Jul 19 10:51:36 ssh -D is on port 443 tcp Jul 19 10:51:44 with openvpns ability to share with https it rocks Jul 19 10:52:21 I know Jul 19 10:52:25 before they added that support Jul 19 10:52:32 I used a perl script Jul 19 10:52:35 for doing that Jul 19 10:55:25 cool Jul 19 10:55:37 it would be extra cool to have all three :-) Jul 19 10:56:41 all 3 is impossible Jul 19 10:56:47 basically it work this way Jul 19 10:57:03 it detects https or ssh/vpn Jul 19 10:57:11 using the fact that one talk first Jul 19 10:57:18 and the other wait Jul 19 10:58:57 ah didnt know that Jul 19 11:18:44 morning all Jul 19 11:18:52 hi RP Jul 19 11:26:00 hi woglinde Jul 19 11:26:10 btw how do I put everything under a proxy? Jul 19 11:26:17 I need : Jul 19 11:26:21 git pull for oe Jul 19 11:26:25 autorev things Jul 19 11:26:27 etc... Jul 19 11:26:30 dont know Jul 19 11:26:32 sorry Jul 19 11:26:37 ok Jul 19 11:26:42 ~curse orange Jul 19 11:26:44 May you be reincarnated as a Windows XP administrator, orange ! Jul 19 11:26:49 (again) Jul 19 11:27:26 sockisfy? Jul 19 11:27:40 was called somelike that, you just do sockify bitbake spanner Jul 19 11:27:40 GNUtoo|laptop: setup your ~/.ssh/config file to use the proxy Jul 19 11:27:57 ah nice Jul 19 11:27:59 thanks a lot Jul 19 11:28:17 * XorA remembers having to do that many years ago at wolfson Jul 19 11:28:36 * RP does this behind the corporate firewall Jul 19 11:28:44 isn't there a variable Jul 19 11:28:48 like http_proxy? Jul 19 11:29:06 because my proxy is ssh Jul 19 11:29:16 I don't remember how to setup a vpn with ssh Jul 19 11:29:21 and that require root access Jul 19 11:29:33 and ubuntu-based uses sudo Jul 19 11:30:12 http://manpages.ubuntu.com/manpages/jaunty/man1/socksify.1.html Jul 19 11:31:13 thanks a lot!!! Jul 19 11:39:23 XorA, http://pastebin.com/AXKn8eAv Jul 19 11:39:27 and it didn't work Jul 19 11:39:33 it's dante.conf Jul 19 11:40:05 bugger it certainly used to, I used it many times a few years ago :-( Jul 19 11:40:07 03Martin Jansa  07org.openembedded.dev * rebc2e1d43a 10openembedded.git/recipes/linux/ (3 files in 3 dirs): Jul 19 11:40:07 linux-openmoko-2.6.32: add missing patch, enable XFS, brtfs, GPT Jul 19 11:40:07 Signed-off-by: Martin Jansa Jul 19 11:40:15 03Martin Jansa  07org.openembedded.dev * r6117802de8 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Jul 19 11:40:15 linux-openmoko-2.6.34: enable XFS, brtfs and GPT in defconfig Jul 19 11:40:15 Signed-off-by: Martin Jansa Jul 19 11:40:38 where should I look for help Jul 19 11:40:38 ? Jul 19 11:41:17 or I could try to do an ssh vpn but I don't remmber how to do it,specially the route part Jul 19 11:47:10 bbs Jul 19 11:49:17 is there some reason, a kernel module (option) doesn't get its entry in modules.dep ? Jul 19 11:49:35 modprobe fails, while insmod with the right path works :/ Jul 19 11:50:07 mlip: was depmod -a called from postinst of package with that kernel module? Jul 19 11:51:15 ehm, the package is generated by the standard kernel recipe? so it should i think Jul 19 11:51:41 JaMa: postins script included in the ipk seems to be right Jul 19 11:52:44 and if you do depmod -a, then modprobe works ok? Jul 19 11:54:58 testing Jul 19 11:58:34 JaMa: when I do it everything works Jul 19 11:59:06 http://pastebin.com/27FRVVmN Jul 19 11:59:10 standard postinst Jul 19 12:03:41 then it should work ok as long as 2.6.28.10 is right and the module is really in /lib/modules/2.6.28.10 Jul 19 12:05:30 it is ... Jul 19 12:06:05 usbserial module works; option, cdc-acm do not :/ Jul 19 14:38:43 good morning Jul 19 14:40:02 here is another trouble from me: http://paste.pocoo.org/show/239383 :) Jul 19 15:06:13 anybody knows what is going on here? Jul 19 15:12:10 This looks pretty much broken...which OE tree is this? Jul 19 15:12:49 oh and what tries to build firefox? Jul 19 15:14:30 "bitbake world" is never a good idea ;) Jul 19 15:14:50 good morning. It doesn't appear that the gumstix overo stuff is in the normal OE repo... is this the case, or am I missing something? Jul 19 15:14:57 seems some people have to much time and too much space Jul 19 15:15:04 I'd like to use the normal OE repo if possible, I really hate how every vendor seems to want their own "spin" of OE Jul 19 15:20:08 tzanger: the problem is that OE dev is a too fast moving target Jul 19 15:20:40 woglinde? Jul 19 15:20:55 tzanger: however you're right, their changes should be pushed into OE dev or stable sometimes. that would reduces headaches for everyone Jul 19 15:21:11 i ve set the qemu param but now the glibc_2.6.1.bb failed :( Jul 19 15:22:00 localdef returned an error Jul 19 15:22:01 thebohemian: yes, I have no problem with them going against point-in-time tags ro something, but this total separate repo blows Jul 19 15:22:33 tzanger: but the repo is a git one, is it? Jul 19 15:22:40 yes Jul 19 15:22:49 tzanger: so theoretically cherries could be picked and pushed into OE Jul 19 15:24:43 that sounds very problematic Jul 19 15:25:01 I like buglabs approach. they have a repo of repos and for every change to OE's metadata you have to copy stuff into a separate folder. this allows stuff to ripe there and you have a clear idea of what needs to be merged back into OE (as long as there is something in this overlay folder -> work is needed to port it back to OE) Jul 19 15:26:09 thebohemian: good to know that someone likes that layout Jul 19 15:26:30 I made it that way before my leave to make things easier for next devs Jul 19 15:26:34 hrw: its your baby, isnt it? Jul 19 15:26:51 yes Jul 19 15:26:56 hrw: yeah, I think it makes totally sense Jul 19 15:27:32 hrw: maybe we should document the approach in the wiki and so help 'customers' to find their way to use OE Jul 19 15:28:06 florian: its OE_HEAD, don't know who builds firefox and Im building angstrom-gnome-image Jul 19 15:28:09 there are many ways to do such setup Jul 19 15:32:31 IgorK: you don't wnat's this for your board ;) Jul 19 15:33:30 u mean I dont want gnome? is it too big for it? or something else? Jul 19 15:34:40 gpe from OE_HEAD is un-buildable... :-( Jul 19 16:02:22 I purged my sources/ svn directory and tarball cache files, but the svn checkout for a package keeps pulling a fixed number... SRCREV=${AUTOREV} in the bb file Jul 19 16:02:32 does anyone know where the svn revision is cached? Jul 19 16:04:36 nm... started working somehow Jul 19 16:48:09 mmh... gcc 4.4.4 miscompiles u-boot's vsprintf() :( Jul 19 16:51:54 03Robert Schuster  07org.openembedded.dev * r4f106a90c3 10openembedded.git/recipes/openjdk/openjdk-6-common.inc: openjdk-6: Reworked packaging (RDEPENDS) to match what users are used to from Ubuntu. Jul 19 16:52:05 03Robert Schuster  07org.openembedded.dev * r4a4c051546 10openembedded.git/recipes/openjdk/openjdk-6-common.inc: openjdk-6: Added manual RDEPENDS for vm-shark and vm-cacao. Jul 19 16:55:29 it seems to swap the two words of uint64_t in put_dec() Jul 19 16:57:21 does angstrom have a script htat is executed autmatically (like /etc/rc.local in debian); or do i need to create my own init script? Jul 19 17:25:39 ensc|w: interesting Jul 19 17:25:49 ensc|w: have you created a small testcase ? Jul 19 17:26:08 khem: I am about it... Jul 19 17:26:23 ensc|w: sometimes people dont use volatile for serialised access Jul 19 17:26:40 and rely on compiler which is incorrect assumption Jul 19 17:28:02 I do not think that this is a volatile think; when I do '++num' outside the loop in put_dec(), buffer will be filled with '0'. When I do it inside, buffer is '1' Jul 19 17:28:31 when doing num |= 0x23ffffffffull, there is given out '35' Jul 19 17:41:56 ensc|w: I did not get it but I will wait for testcase Jul 19 17:51:27 warflyr: it should have boot up scripts look into /etc/rc stuff if you want to customize it then you can Jul 19 18:54:04 khem: I am not able to reproduce the error under linux and have to stop now; I put the code + some debugging at http://pastebin.com/zZ2AjnJw Jul 19 18:54:26 im recieving the error: ERROR: QA Issue with autossh: No GNU_HASH in the elf binary: '/ninja/build/tmp/work/armv7a-angstrom-linux-gnueabi/autossh-1.4b-r0/packages-split/autossh/usr/bin/autossh'; i cant seem to find any solutions for it online but i did find one post where someone said they are ignoring QA errors; how can i ignore QA errors? Jul 19 18:54:34 put_dec() is called there with '3', '2', '1' (see r2 reg) Jul 19 18:55:01 'fp' should be filled with the 'r2' value but is always 0 Jul 19 19:09:17 hmm Jul 19 19:09:21 NOTE: The localpath does not exist 'sources/expat-2.0.1.tar.gz' Jul 19 19:09:24 how does one get around that? Jul 19 19:09:27 hmm... seems that strd is not working at this time... Jul 19 19:09:38 I tried bitbake -cclean expat and then bitbake expat, but it's the same thing Jul 19 19:09:39 perhaps a stack alignment problem... Jul 19 19:09:55 I had erased the stamp files for expat beforehand but then remembered tha tbitbake had a better way (-cclean) Jul 19 19:12:33 khem: yep; stack was not aligned. so no gcc problem... (at least not this time ;)) Jul 19 19:16:38 ensc|w: nice Jul 19 19:17:11 yeah casting is another nightmare where people assume a char* is same as int* Jul 19 19:17:48 warflyr: I would suggest not to ignore QA errors but fix them Jul 19 19:17:57 warflyr: this error has a simple fix Jul 19 19:18:57 warflyr: add TARGET_CC_ARCH += "${LDFLAGS}" in the autossh recipe Jul 19 19:23:54 RP: ping Jul 19 19:26:03 03Koen Kooi  07org.openembedded.dev * r9a3b603429 10openembedded.git/conf/distro/include/sane-srcrevs.inc: opkg: bump SRCREV to 543 Jul 19 20:25:36 khem thanks, that did the trick Jul 19 21:35:04 what would be the best way to install a program to run under qemu-386 inside of a oe linux image for my embedded system? Jul 19 21:37:10 you already have the image, and just want to add another package to it? Jul 19 21:37:42 well I want to install a package to run through qemu-386 emulation Jul 19 21:39:02 ? Jul 19 21:39:12 again, do you already have an oe built filesystem for qemu? Jul 19 21:39:48 yes I have the basic bootstrap image currently reading up on configuring packages for my own modified distro Jul 19 21:41:52 I'm not sure if it's as simple as running my image in qemu-system-arm and installing the packages etc... and then just copying that from my rootfs over to a sd card etc... or If I need to make a recipe that configures the package Jul 19 21:49:28 kergoth: I'm trying to accomplish the this... arm->qemu-386->x86 package Jul 19 21:52:42 Andy_: you can create an image with all stuff cross compiled Jul 19 21:52:47 that will be easy Jul 19 21:53:42 well the package I want to run isn't cross compilable at the end of that chain, or currently it isn't Jul 19 21:54:04 atleast not to my knowledge Jul 19 21:55:38 you can do lot of qemu dance but if it can be done make it compilable under cross env Jul 19 21:55:49 that will be easier to maintain also Jul 19 21:56:37 well for the time being aside from porting the code to cross compilable, is there a good approach for making a package that runs through x86 qemu? Jul 19 21:58:42 Andy_: look into impactlinux.com Jul 19 21:58:51 may be thats more suited for what you want Jul 19 22:00:52 I like what I've seen thus far of OE not sure why I would want to switch to a different cross compilation / linux image creation tool chain Jul 19 22:02:41 Andy_: you could boot qemu from nfs Jul 19 22:02:49 and then do the stuff you want Jul 19 22:02:59 in the end create an image Jul 19 22:03:11 which will have whatever you built Jul 19 22:03:25 this all will be a hack Jul 19 22:04:18 ok that is what I was thinking I was going to have to do, just wasn't sure if there was a cleaner approach Jul 19 22:06:42 you could look into how locales are generated in glibc Jul 19 22:08:18 how would changing the character set information with glibc help? Jul 19 22:08:44 this all looks like localization info to me Jul 19 22:11:26 glibc uses qemu-arm to generate locales for arm system on you rbuild box Jul 19 22:11:30 thats how it will help Jul 19 22:12:00 oh Jul 19 22:12:25 thanks khem Jul 19 22:29:54 khem: seems your PowerDeveloper proposal will be accepted Jul 19 22:39:08 03Khem Raj  07org.openembedded.dev * r1453178a02 10openembedded.git/conf/distro/include/sane-toolchain.inc: Jul 19 22:39:08 sane-toolchain.inc: Add armv4 to ARM EABI supporting architectures. Jul 19 22:39:08 Signed-off-by: Khem Raj Jul 19 22:39:20 03Khem Raj  07org.openembedded.dev * re9fe87a13c 10openembedded.git/conf/ (21 files in 3 dirs): (log message trimmed) Jul 19 22:39:20 distro: Move thumb into distro features Jul 19 22:39:20 * Currenlty we define the thumb specific overrides in machine tune files Jul 19 22:39:20 and as machine conf files are included before the distro conf files Jul 19 22:39:21 the overrides dont get right values and set incorrectly. Jul 19 22:40:08 hah! i forgot the old OZ buildroot is still in sourceforge cvs Jul 19 22:41:03 hmmmmm Jul 19 23:06:49 hey guys Jul 19 23:07:04 (dammit, now I have to watch The Goonies) Jul 19 23:07:06 NOTE: The localpath does not exist 'sources/expat-2.0.1.tar.gz' Jul 19 23:07:19 ^^ I'm trying to build expat, but it's throwing this up Jul 19 23:07:47 I had the connection drop so it failed to untar, so I deleted the file and also the stamp files in order to make OE grab it again Jul 19 23:08:14 when that didn't work I remembered that "bitbake -cclean expat" is a better way to do that, but after running it and then "bitbake expat" it's throwing this error now Jul 19 23:11:26 did you forget to remove the .md5? Jul 19 23:13:27 oh, perhaps Jul 19 23:16:45 the .md5 is the indicator that it doesn't need to fetch that file -- its complete Jul 19 23:17:03 ah, well, now console-image is building jack Jul 19 23:33:44 03Chase Maupin  07org.openembedded.dev * red482f3982 10openembedded.git/ (9 files in 9 dirs): Jul 19 23:33:45 remove dm3730-am3715-evm machine type references Jul 19 23:33:45 * Removed references to the dm3730-am3715-evm machine type Jul 19 23:33:45 and replaced with dm37x-evm and am37x-evm machine types. Jul 19 23:33:45 Signed-off-by: Chase Maupin Jul 19 23:54:36 kergoth_: hmm, okay that worked, unfortunately it seems that my connection wasn't interrupted; the archive's damaged Jul 20 00:17:49 hmm, anyone aware that expat-2.0.1 on sourceforge is a bad tarball? Jul 20 00:19:26 oh, interesting Jul 20 00:19:28 it's a gzip problem Jul 20 00:22:09 tzanger, expat seems fine to me Jul 20 00:22:37 ee8b492592568805593f81f8cdf2a04c expat-2.0.1.tar.gz Jul 20 00:22:45 (md5) Jul 20 00:23:20 grg: the md5 matches, but apparently ubuntu buggered up gzip somehow Jul 20 00:23:39 works on 9.04, doesn't on 10.04 Jul 20 00:23:56 i don't use ubuntu, so i can't comment on that Jul 20 00:23:58 https://bugs.launchpad.net/ubuntu/+source/gzip/+bug/524366 Jul 20 00:24:39 it affected redhat as well. :-) Jul 20 00:24:59 anyway, I gzip -d'd the file on 9.04, gzip'd it back up and copied the new one over and altered the .md5 in sources Jul 20 00:25:03 hopefully that fixes it Jul 20 00:27:56 tzanger, so now we need gzip-native? Jul 20 00:28:03 ugghhh Jul 20 00:29:45 tar-native too Jul 20 00:29:52 But even then, heh Jul 20 00:30:43 It's too bad the python only extraction stuff was noticeably slower, right kergoth? Jul 20 00:32:37 yep, it was enough to impact build times a fair bit Jul 20 00:32:52 hmm.. i actually think its better to tell people to use their distro fix than add more -native dependencies Jul 20 00:34:14 there isn't a distro fix for 10.04 yet Jul 20 00:34:19 it's one of those weird-ass corner cases Jul 20 00:34:28 who the hell makes the expat archive on OS/2 anyway? :-) Jul 20 00:34:59 you've got to be strange to write an xml parser Jul 20 00:36:14 Touche. Jul 20 00:38:04 it would be nice to reduce the number of -native dependent packages for any given image build Jul 20 00:38:26 surely there's some low hanging fruit... Jul 20 01:42:31 Some way to get a sane default environment to use and re-use would be nice Jul 20 01:42:59 There's very few -native packages we use that can't be made to work happily via env stuff to know where its junk is Jul 20 01:52:01 we need to distinguish between -natives that we have just to satisfy the dep instead of the build machine, and those we need because we require specifically patched versions Jul 20 01:54:12 You know, I don't think there's anything except for perl and I guess python that we need for doing the target, done up specially Jul 20 01:54:26 heck, we could use build host mtd tools if we wanted, for example :) Jul 20 01:54:39 (On recent and sane distros) Jul 20 01:55:37 libtool, coreutils both require patches, but i think we shuld move back to install-sh or a custom wrapper rather than patching coreutils just to get an install that doesn't strip Jul 20 01:55:40 good evening Jul 20 01:55:46 autoconf has patches we can kill, I expect, the whole gnu-configize thing Jul 20 01:56:32 building x11-gpe-image, Im getting: http://paste.pocoo.org/show/239569 Jul 20 01:58:03 building angstrom-gnome-image, Im getting http://paste.pocoo.org/show/239571 Jul 20 01:58:23 anybody have a suggestion? tia! Jul 20 02:09:41 ah that's right, autohell Jul 20 02:10:21 We only patch coreutils 7.2 to fix a bug, heh Jul 20 02:10:24 8.5 is unpatched Jul 20 02:11:01 * Tartarus does idly wonder if we really do need to patch libtool/etc over doing it to provide a known sane version Jul 20 02:39:48 Tartarus: there is no sane libtool. Jul 20 02:49:18 heh Jul 20 02:51:15 there are multiple aspects of its behavior that have to be changed in patches. it wants to add a -L of any -rpath, for example Jul 20 02:51:23 iirc, anyway Jul 20 02:51:48 so any idea on how 2 fix my problems? Jul 20 02:51:59 kergoth: Tartarus once I had tried to reduce the natives Jul 20 02:52:23 and I pretty much got it down to like 7 or 8 really needed one Jul 20 02:52:58 khem: nice. I think we need to decide on a way to distinguish these things. maybe a different prefix than -native if we alter it in a required way Jul 20 02:53:05 -oe-native seems like a good route Jul 20 02:53:10 "vendor" of a sort Jul 20 02:54:10 kergoth: yeah we should have something like native-oe-ubuntu9.04 Jul 20 02:54:14 IgorK try: bitbake iptables-native Jul 20 02:54:17 native-suse10.0 Jul 20 02:54:19 etc Jul 20 02:55:29 kergoth: I have made a list of package that one needs on say ubuntu 10.04 installed Jul 20 02:55:45 and then I keep adding to this list and remove from the -native Jul 20 02:55:59 khem: it would be nice to have a class/inc which determines your distro and sets ASSUME_PROVIDED accordingly. Jul 20 02:56:09 wouldn't be terribly difficult Jul 20 02:56:09 kergoth: good idea Jul 20 02:56:12 yep Jul 20 02:56:28 call the distros package manager to install the natives o.0 Jul 20 02:56:41 kergoth: cat /etc/lsb-release Jul 20 02:56:53 should give all info needed Jul 20 02:56:59 hopefully Jul 20 02:57:13 warflyr: heh not everybody is root Jul 20 02:57:17 on a build box Jul 20 02:57:31 good point o.0 Jul 20 02:58:57 kergoth: detect-buildos.bbclass :) Jul 20 02:59:53 we should think about separating INHERITed bbclasses from inherited ones **** ENDING LOGGING AT Tue Jul 20 02:59:56 2010