**** BEGIN LOGGING AT Sat Mar 11 10:59:56 2006 Mar 11 11:11:21 03koen 07org.oe.dev * rba5e9634... 10/packages/meta/ (meta-toolchain.bb task-sdk.bb): Mar 11 11:11:23 meta-toolchain: add a .bb to generate a tarball of the SDK toolchain Mar 11 11:11:25 task-sdk: add a task-sdk-bare only containing glibc bits Mar 11 12:01:16 03coredump 07org.oe.dev * r5e78a71a... 10/packages/altboot/ (altboot_0.0.5.bb altboot_1.0.5.bb): altboot: Sync with .oz branch Mar 11 12:08:32 hi, folks . Is there anybody use u-boot for pxa27x board ? Mar 11 12:08:46 hi, I setup an oe build environment to build openzaurus for my SL5600, it builds kernel 2.4.18 and 2.6.15 where can i disable building the 2.6.15 kernel? Mar 11 12:09:50 this is odd, virtual/kernel should point to _one_, not two kernels Mar 11 12:11:57 hmm, i did a monoton pull this morning Mar 11 12:12:12 where is virtual/kernel defined? Mar 11 12:14:04 you are using the oz354fam083 branch, are you ? Mar 11 12:14:12 virtual/kernel is defined in the $MACHINE.conf Mar 11 12:15:12 i did a pull for org.openembedded.dev Mar 11 12:16:31 if you're interested in building an openzaurus release, i recommend using the branch Mar 11 12:17:50 the poodle.conf states KERNEL_VERSION ?= "2.4", should I change that to KERNEL_VERSION = "2.4" Mar 11 12:18:06 morning Mar 11 12:19:14 ok, the build was fine, the problem arise when i tried building fbvncserver, which needs the kernel config, which is for 2.6 instead of 2.4 Mar 11 12:19:22 hey reenoo Mar 11 12:19:49 hi koen Mar 11 12:34:14 03koen 07org.oe.dev * r8cc816cb... 10/packages/linux/ep93xx-kernel/defconfig: ep93xx-kernel: update defconfig Mar 11 12:34:19 03koen 07org.oe.dev * ra5422e09... 10/packages/linux/ep93xx-kernel_2.6.15+2.6.16-rc5.bb: ep93xx-kernel: update to derovo15 Mar 11 12:36:27 03rw 07org.oe.oz354fam083 * rffdd8d95... 10/packages/dropbear/ (dropbear_0.45.bb dropbear_0.46.bb dropbear_0.47.bb): dropbear: never apply allow-nopw.patch for DISTRO=familiar Mar 11 13:35:47 hello all Mar 11 13:36:12 Did any one tried to build nylon distro lately with OE? Mar 11 13:39:24 hey pH5 Mar 11 13:39:30 hey koen Mar 11 13:39:35 i don't think so. nylon is a bit in flux since the last maintainer started a world trip. someone else has picked up this task, but i'm sure he would be glad to accept patches as well Mar 11 13:39:41 hrm, that gcc 4.1 .bb isn't working Mar 11 13:47:24 mickeI, may be you know tag that I can try to build against? Mar 11 13:49:10 blib: no, i have no idea. just try a build and lets see what problems you have Mar 11 13:50:42 it say that it cant' find linux-mtx-1 package, but it is there. So don't even know where to start with that Mar 11 13:55:48 hmm, can't help atm. since i'm busy doing oz builds, but how 'bout sending the exact chain of commands plus the error message to oe@handhelds.org? the new nylon maintainer is supposed to read that list as well Mar 11 13:57:23 ok Mar 11 14:09:02 All: any news on the OeNI meeting? Mar 11 14:12:26 hi Mar 11 14:12:38 hi Bernardo Mar 11 14:12:46 03rw 07org.oe.dev * rfb9d7b21... 10/conf/distro/familiar-unstable.conf: familiar-unstable: reinstate PREFERRED_VERSION_gcc* one more time. Mar 11 14:19:45 hi all Mar 11 14:19:54 03rw 07org.oe.dev * rc3f657e2... 10/conf/distro/familiar.conf: familiar: set MAINTAINER and ENABLE_BINARY_LOCALE_GENERATION. Mar 11 14:31:41 03rw 07org.oe.dev * rd4c53927... 10/packages/neon/neon_0.24.7.bb: neon: don't link against openssl for licensing reasons. inherit binconfig rather than duplocating code. Mar 11 14:32:01 duplocating Mar 11 14:32:06 * reenoo rotfl Mar 11 14:32:11 ~lart typos Mar 11 14:32:12 * ibot beats typos severely about the head and shoulders with a rubber chicken Mar 11 14:33:23 guys, trying to build a module, but "inherit module" seems to be switching into the kernel dir before doing the make, and I want the make to be done in the package dir Mar 11 14:33:52 reenoo: so https support in neon is off-limits? Mar 11 14:35:44 reenoo: also, I didn't inherit binconfig in my patch for oz354 for the reasons I put on the bug report Mar 11 14:38:51 Bernardo: sounds like you want to fix subversion's configury Mar 11 14:39:11 reenoo: I wanted to, but not enough time or expertise... :) Mar 11 14:40:18 I just proposed bug #732 after getting subversion to work with https for a friend that was leaving to the usa and wanted to use his collie to build some stuff Mar 11 14:40:33 I know it's a quick'n'dirty hack Mar 11 14:44:41 morning Mar 11 14:44:49 hi chouimat Mar 11 14:45:07 ROFL http://www.askmen.com/jokes/2006_mar/mar11.html Mar 11 15:39:58 morning all Mar 11 15:41:53 hi RP Mar 11 15:44:28 hey RP Mar 11 15:47:30 hi RP Mar 11 15:47:51 can any of you guys tell me what is the recommended way to add stuff to modules.conf? Mar 11 15:48:11 the same way hostap-modules adds to pcmcia-cs.conf, in install? Mar 11 15:48:13 let update-modules handle it Mar 11 15:48:17 I found the colour problem under the qemu images Mar 11 15:48:23 RP: yay! Mar 11 15:48:25 RP: and a fix? Mar 11 15:49:06 koen: I have a fix. It was a matchbox issue - just awaiting approval to commit it. Mar 11 15:49:16 heh Mar 11 15:49:40 I wonder if that fixes florian's MB problem on his 15bit sgi as well Mar 11 15:51:33 koen: I certainly have a good idea of where any 15bpp problems would be Mar 11 15:52:10 * koen needs a website/html/css guru soon Mar 11 15:52:26 koen: This "fix" just gets the reversed R and B case working for now until we can fix the code properly... Mar 11 15:52:38 for the Angstrom site? Mar 11 15:52:56 angstrom and others Mar 11 15:53:26 I can do php, html and sql. My design skills aren't good though ;-) Mar 11 15:53:46 koen: ok, thanks Mar 11 16:08:01 I needed this patch to build xserver-kdrive_20050207 against X11R7.0 xlibs: http://en.pastebin.ca/45241 Mar 11 16:08:22 Can I drop this and the X11R7 xlibs into org.oe.dev tomorrow? Mar 11 16:08:49 hmm, I got a question: At first, I build a bootstrap-image. The gcc-cross will use "gcc-cross-3.4.4-initial". After that, I bitbake LAB-kernel, The gcc-cross will be "gcc-cross-3.4.4_kernel-3.4.4". Mar 11 16:09:03 pH5: send a mail to oe@ saying "this is going in tomorrow" Mar 11 16:09:25 pH5: Is this backwards compatible? Mar 11 16:09:26 leoncamel: yes, it build a seperate gcc to build the kernel Mar 11 16:09:43 pH5: (The xserver patch I mean) Mar 11 16:09:52 Then, I bitbake mplayer, what the gcc-cross will be ? gcc-cross-initial or gcc-cross-3.4.4_kernel-3.4.4 ? Mar 11 16:10:04 neither Mar 11 16:10:17 gcc-cross-initial is only used to build glibc Mar 11 16:10:44 koen, which gcc-cross will use then ? Mar 11 16:10:59 pH5: I'm certainly fine with the X11R7 libs. Tieing the old xserver to the new libs does worry me slightly though Mar 11 16:11:03 gcc-cross-3.4.4 (depending on your distro) Mar 11 16:11:39 RP: good point, I'll have to try this. Mar 11 16:12:00 RP: the _cvs version works flawlessly though. perhaps we should just make a new snapshot? Mar 11 16:12:01 koen, So, hmm, ok. The gcc-3.4.4_kernel-3.4.4 is only for building kernel ? Mar 11 16:12:17 leoncamel: yes, some kernels are picky about the compiler Mar 11 16:12:51 pH5: That would be better. Also feel free to add the patch for that version as I for one might fine it useful but just have it commented out in the .bb for now. Mar 11 16:13:27 koen, which gcc-cross will put into tmp/cross/bin then ? gcc-cross-3.4.4 ? Mar 11 16:13:43 pH5: Which device did you test the xserver on btw? Mar 11 16:14:01 hx4700, that's the only one I have... Mar 11 16:15:32 pH5: ok. I should be able to test on the Zaurus but not soon :-/ Mar 11 16:19:50 RP: great. I'll write a warning to oe@ Mar 11 16:26:14 03tmbinc 07org.oe.dreambox * r42ec8d00... 10/packages/meta/dreambox-image.bb: dreambox-image: add enigma2-plugin-systemplugins-configurationbackup to image Mar 11 16:26:18 03tmbinc 07org.oe.dreambox * ra951b187... 10/packages/tuxbox/ (3 files in 2 dirs): tuxbox-tuxtxt-32bpp: fix unsigned char, fix input device detection, bump PR Mar 11 17:46:04 koen, while we are on the gcc-cross/glibc topic, how is the version picked? I see PREFERRED_VERSION_glibc in the various confs. I set this in my generic.conf, but an older version gets picked. Mar 11 17:46:50 if you override it in your local.conf it will use that Mar 11 17:47:09 or machine or distro as well i belive Mar 11 17:47:54 lemme check - i don't think it appears in any other file that gets read... Mar 11 17:47:57 emte: generic.conf is in conf/distro :) Mar 11 17:47:59 * emte tossed in my 1/2 cent incase koen went for coffee Mar 11 17:48:45 koen, yeah but can you use it with the "non-version" bitbake code now? Mar 11 17:48:53 er version-only Mar 11 17:49:43 nope, PREFERRED_VERSION_glibc only appears in the generic.conf. wierd. Mar 11 17:50:06 emte: the version only code is only familiar, angstrom and OZ Mar 11 17:50:14 ah Mar 11 17:50:27 thought it was OE-wide Mar 11 17:50:33 not yet Mar 11 17:50:34 is some other package specifically naming a dependency on the older version? Mar 11 17:50:44 RP's sanity.conf will do that Mar 11 17:51:19 HopsNBarley, does the glibc bitbake your trying to call have a bb file? Mar 11 17:52:33 emte, i'm trying to build a ppc toolchain with bitbake virtual/powerpc-linux-gcc. Mar 11 17:52:41 yeah, it's this one: 2.3.2+cvs20040726 Mar 11 17:52:48 http://linuxdevices.com/news/NS2786595347.html Mar 11 17:53:00 if only it had a better screen Mar 11 17:53:14 HopsNBarley: there's a bug in recent bitbake about selectors Mar 11 17:53:25 03rw 07org.oe.dev * rc110c77d... 10/packages/dropbear/ (4 files): dropbear: never apply allow-nopw.patch for DISTRO=familiar Mar 11 17:53:33 HopsNBarley: as a workaround remove all other glibc .bbs Mar 11 17:53:52 koen, great idea. Mar 11 17:54:41 koen, while i think about it, you know where to look for more docs on trying to "inject" python test code into bb metafiles? Mar 11 17:55:11 emte: nope, sorry Mar 11 17:55:34 apart from the usual 'over 3k examples' Mar 11 17:56:25 koen, yeah, i just havent git it to quite work Mar 11 17:56:29 got* Mar 11 17:58:05 http://oe.pastebin.com/596542 can you see my problem? Mar 11 18:01:32 i think my problem is with the bb.data.setVar('EXTRA_OECONF', ' --with-ipc=tcp --program-prefix= ', d) line Mar 11 18:02:13 03rw 07org.oe.dev * rdfcb7791... 10/packages/dropbear/dropbear.inc: dropbear: fix typo in last commit Mar 11 18:03:42 grr stupid power ... worst thing about the power doing out is that my system rebooted with a new kernel and now my sounds card doesnt behave again :( Mar 11 18:03:47 going* Mar 11 18:07:48 aha i think i found my problem :) Mar 11 18:10:07 hey Mar 11 18:10:21 my NSLU2 arrived, tomorrow I will have the time to brick it Mar 11 18:10:56 congrats Mar 11 18:12:26 * emte actually has 2 problems, with the first one being that i've no clue what i am doing Mar 11 18:13:37 okay so i just checked and the gloabal variable IS getting passed but EXTRA_OECONF is not getting modified ... therefore my test is failing ... Mar 11 18:16:04 emte: is it overwritten later? Mar 11 18:17:06 trying to find out now Mar 11 18:17:21 just doing some prints to check what i expect Mar 11 18:17:53 well its behaving Mar 11 18:17:56 NOTE: Handling BitBake files: \ (0464/3113) [14 %]1 Mar 11 18:17:56 --with-ipc=tcp --program-prefix= Mar 11 18:18:05 so it must be getting overridden Mar 11 18:19:11 zecke, this is what i have currently Mar 11 18:19:13 http://oe.pastebin.com/596570 Mar 11 18:20:16 including both my print tests Mar 11 18:21:00 hmm i wonder ... Mar 11 18:22:44 the "inherit autotools" is the only line in the non-native bb file, will that override the native settings? Mar 11 18:23:05 have you traced all the open calls? I found this pretty helpful in understanding/diagnosing what's going on in bitbake Mar 11 18:23:46 emte: I just grepped the classes and none should overwrite the EXTRA_OECONF variable Mar 11 18:24:00 emte: try __anonfunc instead of python() (should be the same though) Mar 11 18:24:07 hmm Mar 11 18:24:12 emte: what does the shell say if you peek into the variable? Mar 11 18:24:26 will do, i am just trying the function in both files now Mar 11 18:24:38 emte NOTE: Handling BitBake files: \ (0464/3113) [14 %]1 Mar 11 18:24:38 emte --with-ipc=tcp --program-prefix= Mar 11 18:24:43 as i would expect Mar 11 18:25:04 the "1" is the setting in my local.conf Mar 11 18:25:22 so that part behaves and so does the function Mar 11 18:26:17 and the HOST_NONSYSV i have set to default "0" in the bitbake.conf Mar 11 18:26:24 for compatability Mar 11 18:28:08 aha! Mar 11 18:28:11 it loves me Mar 11 18:28:37 btw, what does "--program-prefix= " do? Mar 11 18:28:49 emte: you specified it? Mar 11 18:28:57 ' ' gets prepended to fakeroot Mar 11 18:29:18 zecke, yes i did Mar 11 18:29:20 so it is ' fakeroot' e.g. you can use it to prefix a gcc with arm-linux- ;) Mar 11 18:29:29 ah Mar 11 18:30:27 it seems i have to have the function in the called bb as well Mar 11 18:30:33 or that is how it just worked Mar 11 18:31:34 so, do i submit all three files to bugtracker or a patch for all three? Mar 11 18:33:18 patch Mar 11 18:33:22 * emte tests non-setting Mar 11 18:33:44 k Mar 11 18:34:07 * Bernardo wonders why when using oe_runmake or plain inheriting do_compile from module.bbclass, arm-linux-gcc gets called instead of arm-linux-gcc-2.95 Mar 11 18:34:52 fakeroot shouldnt need a PR bump from this because i never actually changed fakeroot , correct? Mar 11 18:35:27 * Bernardo gives up for now... better to test why rt2570 builds and loads but doesn't work Mar 11 18:37:56 Bernardo: the kernel gcc is not 2.95? (at least not set as this) Mar 11 18:39:33 hrm ... bad function bad! Mar 11 18:41:47 hi, folks. How can I create a nfsroot for target ? well, the tmp/rootfs load for building rootfs images. And HOW can I want setup a nfs root for target? Mar 11 18:42:41 hi zecke Mar 11 18:43:02 zecke: I finally found the colour problem under qemu. Matchbox bug :) Mar 11 18:44:12 03hrw 07org.oe.dev * r5de21ce1... 10/packages/dropbear/dropbear_0.48.bb: dropbear: add 0.48 - this time for sure Mar 11 18:46:59 03koen 07org.oe.dev * r97dd2ea8... 10/packages/binutils/ (10 files in 2 dirs): add binutils 2.6.91.0.6 Mar 11 18:54:02 i think i got it :) Mar 11 18:54:10 needed a pair of single quotes Mar 11 19:00:25 woohoo test 3 works :) Mar 11 19:00:45 time for some -nurding :) Mar 11 19:28:13 RP: hehe Mar 11 19:28:25 RP: did you make mallum release a 1.0.1 yet? Mar 11 19:28:41 enough writing for now, time to brick the NSLU2 Mar 11 19:30:48 zecke: yo can even brick it over ethernet! Mar 11 19:30:54 it was never so easy ;) Mar 11 19:31:28 koen: I hope it can read and write HFS+ media Mar 11 19:31:44 if 2.6.15 can, the nslu2 can too Mar 11 19:35:25 zecke: http://www.macosx.nl/?p=showarticle&art_id=2247 Mar 11 19:40:06 * france is back (gone 09:59:11) Mar 11 19:43:38 koen: I have to try that PDF thing Mar 11 19:43:47 koen: I'm annoyed I can't zoom in safari Mar 11 19:43:58 that SafariSpeed thing is also great Mar 11 20:08:16 zecke: not yet :) Mar 11 20:09:04 Why is it me+OE+ext3 = damaged partitions? :-/ Mar 11 20:12:28 because ext3 sucks Mar 11 20:12:35 xfs hasn't let me down yet Mar 11 20:13:51 koen|tv: I have seen all Linux FS's fail Mar 11 20:13:54 yup xfs loves me Mar 11 20:14:05 even after the power fail last night Mar 11 20:14:11 zecke: yeah, I'd love to use zfs Mar 11 20:14:30 koen|tv: have you had redboot access from OSX? Mar 11 20:15:54 once I get the pl2303 osx driver working, yes Mar 11 20:16:11 ~lart bazaar Mar 11 20:16:11 * ibot runs at bazaar with an origami Swiss Army knife, and inflicts a nasty paper cut Mar 11 20:18:43 koen|tv: oh you haven't tried over ethernet? Mar 11 20:19:09 zecke: no, I have a linux box ready for stuff like that Mar 11 20:22:34 koen|tv: I tried xfs and I find that easier to break :-/ Mar 11 20:24:07 lets hope for zfs support in linux :) Mar 11 20:25:24 it will confuse the hell out of zaurus users ;) Mar 11 20:28:58 zecke: mm, I don't think I've ever seen ext3 fail in a serious way. XFS does seem to be a bit of a liability, though. Mar 11 20:29:11 I've had severe problems with XFS and powerfailures Mar 11 20:29:32 NAiL: I have a ups for that :) Mar 11 20:29:36 all open files were overwritten with ^@ when the power went out. Mar 11 20:29:49 (All files written to) Mar 11 20:30:10 not sure what other file systems are worth trying. Has anybody used JFS? Mar 11 20:32:34 pb_: I have seen ext3 fail on syscall fuzzinh Mar 11 20:32:56 zecke: I'm not sure I know what "syscall fuzzinh" is, but it sounds eleet. Mar 11 20:33:00 pb_: which is nothing one does normally though :) Mar 11 20:33:18 pb_: make syscalls with weird arguments Mar 11 20:33:25 oh right Mar 11 20:33:30 pb_: e.g. let Opie run Mar 11 20:33:48 nothing you would normally do :) Mar 11 20:34:00 indeed Mar 11 20:34:15 I guess I feel that you get what you deserve in that case :-} Mar 11 20:34:53 pb_: JFS loses the speed contest for most things you'd use a FS for Mar 11 20:35:08 it is reportedly very stable though Mar 11 20:35:24 ah, right Mar 11 20:35:58 hey guys, how would i make wpa-supplicant in the rootfs for openzaurus-unstable image updated to the new version; it now links against gnutls Mar 11 20:36:02 and works Mar 11 20:36:04 i've used it successfully Mar 11 20:36:20 the other reasons I gave up on xfs were that grub hates it, and the online resizing thing can only grow partitions, not shrink them Mar 11 20:36:23 currently there is some werid wpa-supplicant-nossl package, i would like to depreciate that entirely Mar 11 20:36:40 jnc: why? Mar 11 20:36:41 send patches to bugzilla, I guess Mar 11 20:37:01 jnc: I guess ssl is implemented with OpenSSL? Mar 11 20:37:13 I'm not sure why you have that antipathy to wpa-supplicant-nossl though. There are some folks who might not want to install any tls library. Mar 11 20:37:33 zecke: it was, gnutls is in the rootfs for many of those images, so i think supporting openssl in wpa-supplicant is pointless when gnutls works Mar 11 20:37:45 pb_: Thread Locale Storage? ;) Mar 11 20:37:47 pb_: gnutls is already in many images Mar 11 20:37:50 as it happens, for gpe we already ship gnutls in the images, so ssl support in the supplicant would come for free, but the Opie crew might balk at installing a TLS library. especially one whose name starts with g. Mar 11 20:37:50 no no Mar 11 20:38:01 TLS is a term for SSL stuff, not thread local storage Mar 11 20:38:10 I know :) Mar 11 20:38:12 jnc: many images, but not all of them Mar 11 20:38:22 pb_: jnc : I only mind OpenSSL Mar 11 20:38:23 okay sure Mar 11 20:38:30 I don't really understand what would be gained by removing the option to not have tls. Mar 11 20:38:43 but the wpa-supplicant-nossl that is in there does not work at all with the 2.6 kernel Mar 11 20:38:49 and we are targetting 2.6 aren't we? Mar 11 20:39:10 doesnt the wpa-supplicant-nossl end up with a crippled package taht cannot auth ? Mar 11 20:39:19 exactly Mar 11 20:39:20 jnc: I'm not sure who "we" is in that context. Mar 11 20:39:37 pb_: RP mentioned something about completely depreciating all 2.4.x kernels Mar 11 20:39:54 jnc: I suppose he must have been drinking. Mar 11 20:39:59 jnc, i doubt that will happen for a very long time Mar 11 20:40:00 he may have been thinking of OZ, but i thought it was in the context of all distros in OE Mar 11 20:40:39 anyhow wpa-supplicant-nossl is broken either way AFAIK Mar 11 20:40:45 http://www.handhelds.org/moin/moin.cgi/HandheldsDevices_2fh3600 Mar 11 20:40:57 is the status page for the h36 2.6 kernel Mar 11 20:41:01 RP is welcome to deprec(i)ate the 2.4 kernels as much as he likes in a personal capacity, of course, but there are some platforms that are still using 2.4 and will probably continue to do so for a while yet. Mar 11 20:41:23 That is a slight misquote :) Mar 11 20:41:28 ah okay Mar 11 20:41:35 RP: tell me what you meant again? Mar 11 20:41:35 lol Mar 11 20:41:49 * emte is off Mar 11 20:42:01 i'll submit my non-sysv stuff later today Mar 11 20:42:19 emte: what are you using instead? init-ng? Mar 11 20:42:50 I said that a number of zaurus 2.4 kernels are end of life and only 2.6 will be developed in future for the models c7x0, akita, spitz and borzoi, plus probably tosa Mar 11 20:43:39 oh okay, my mistake Mar 11 20:43:54 i'm really asking what you guys think about this wpa-supplicant thing Mar 11 20:44:04 jnc: did you see the gcc 4.1.0 .bb on the mailinglist? Mar 11 20:44:08 Also, it was mentioned that angstrom might be 2.6 only at first Mar 11 20:44:23 koen|tv: i haven't yet subscribed to the ML, i.e. i didn't see where to do that Mar 11 20:44:36 RP: till 2.4 eabi patches surface it will be 2.6 only Mar 11 20:45:25 jnc: As long as you make sure the licencing issue is ok, I can't see a problem with some new wpa-supplicant packages and changing the image depends to use them where appropriate Mar 11 20:45:30 jnc: http://handhelds.org/mailman/listinfo/oe Mar 11 20:45:31 koen|tv: exactly Mar 11 20:45:54 jnc: http://bugs.openembedded.org/show_bug.cgi?id=733 Mar 11 20:46:08 RP: and eabi is also armv5 only at the moment Mar 11 20:46:21 thanks koen Mar 11 20:47:44 koen|tv: FYI the ssl cert for "localhost.tqghQq" expired on 12/06/2005 Mar 11 20:52:56 anyone know where I can get the upslug from? Mar 11 20:53:00 sf.net is broken ATM Mar 11 20:53:39 pb_: Does the collections handling in bitbake cover class/*.bbclass files? Mar 11 20:53:54 zecke: http://pdo.debian.net/testing/otherosfs/upslug2 Mar 11 20:55:43 RP: yes Mar 11 20:55:54 RP: it also covers conf Mar 11 20:56:02 RP: site, etc... Mar 11 20:57:01 zecke: thanks, that makes sense, I just wanted to confirm Mar 11 20:57:44 RP: I had a hard time creating the right regular expressions at first Mar 11 20:57:55 RP: do you suffer from the same problem? Mar 11 20:59:56 zecke: I'll find out if I have it right shortly :) Mar 11 21:00:27 zecke: I'm going by what the bitbake manual says and it seems quite clear so hopefully its just going to work Mar 11 21:21:28 for some reason when I try to build -native package during the build bitbake uses include files from target platform instead of host platform. How can I debug it? Mar 11 21:31:03 blib: change into the workdir Mar 11 21:31:18 blib: look at the log files, config.log etc Mar 11 21:42:28 hi Mar 11 21:42:35 hey hrw Mar 11 21:43:29 RP: where did you hide linux-openzaurus_git.bb? Mar 11 21:43:55 http://www.rpsys.net/openzaurus/temp/linux-openzaurus_git.bb Mar 11 21:44:28 thx Mar 11 21:53:58 NOTE: package collie26-image-1.0: started Mar 11 22:00:58 what is openmn Mar 11 22:01:03 ?$ Mar 11 22:02:28 distro Mar 11 22:02:58 oh hmm Mar 11 22:03:19 #@DESCRIPTION: Destribution for Devices produced by M&N Solutions GmbH, Germany, Rosbach Mar 11 22:04:09 fetching zaurus.git tree... Mar 11 22:04:22 jnc: Bero also got glibc 2.4 working Mar 11 22:04:28 interesting Mar 11 22:04:43 so we'll have all the necessary bits for EABI soon(TM) Mar 11 22:04:51 Real Soon Now (TM) Mar 11 22:05:00 where 'working' read 'compiling' Mar 11 22:05:29 i'm still very strongly for seperating out the SDK from the tree Mar 11 22:05:47 jnc: use different BBFILES? Mar 11 22:05:53 more than that Mar 11 22:05:58 make the SDK a distro Mar 11 22:06:02 whatever :) Mar 11 22:06:29 you build the SDK, then you change the configs to point at the SDK as a pre-built verison of the tools needed to compile Mar 11 22:07:04 combining development with quality assurance is pretty much impossible in the current configuration Mar 11 22:07:21 put staging under package management control! Mar 11 22:07:22 ? Mar 11 22:07:47 jnc: ah hand having two different trees will magically increase quality? Mar 11 22:08:04 it will make it easy to keep the SDK from moving around too much Mar 11 22:08:07 jnc: in that case I could write a bucket sort like tool to split up the tree Mar 11 22:08:18 ~lart git fetcher for lack of progress info Mar 11 22:08:19 * ibot cats /dev/urandom into git fetcher's ear for lack of progress info Mar 11 22:08:20 plus it would speed up some of this Mar 11 22:08:31 jnc: PREFERRED_VERSION, PROVIDER and BBFILES are enough for that Mar 11 22:08:31 not by much, but the complaint i hear is that OE is too slow Mar 11 22:08:35 jnc: seriously Mar 11 22:08:59 jnc: I recommend you try to understand how OE works and what the relevant design decisions where Mar 11 22:09:05 hm Mar 11 22:09:09 jnc: bitbake recipes parser is slow Mar 11 22:09:11 jnc: what has speed to do with SDK? Mar 11 22:09:13 were even Mar 11 22:09:27 jnc: I thought we talk about quality? Mar 11 22:09:35 :) Mar 11 22:09:48 zecke: how can i send you my SDK to use for building your OE tree? Mar 11 22:09:50 i am on amd64 Mar 11 22:09:51 reenoo: how can I thank you for your review? Mar 11 22:09:59 are you on amd64? Mar 11 22:10:02 zecke: which review? Mar 11 22:10:14 reenoo: you do on the oe commit mailinglist Mar 11 22:10:37 zecke: ah heh Mar 11 22:10:58 zecke: how about unborking bitbake's current dependecy issues? ;) Mar 11 22:11:00 jnc: dunno, maybe by getting me an AMD64? Mar 11 22:11:06 zecke: no. Mar 11 22:11:17 s/dependecy/dependecy handling/ Mar 11 22:11:18 how would i use the current setup, and build you an SDK for your platform Mar 11 22:11:29 reenoo: that is NP complete I guess, I'm in the process of building a gadget to prove that Mar 11 22:11:49 jnc: 1. select which licenses you want to distribute Mar 11 22:11:51 and by platform i meant build platform Mar 11 22:12:18 jnc: 2. build a OE meta-sdk in a lsb environment Mar 11 22:12:26 jnc: 3. wait Mar 11 22:12:37 jnc: 4. have a SDK that works on any lsb compilant system Mar 11 22:12:44 ... Mar 11 22:12:46 for any arch? Mar 11 22:12:51 cu Mar 11 22:12:59 jnc: for any lsb compilant system Mar 11 22:13:03 heh Mar 11 22:13:13 has the lsb stopped being a farce? Mar 11 22:13:24 no, it still is Mar 11 22:13:34 koen: well it has at least some ABIs defined :} Mar 11 22:13:34 their testsuite that is Mar 11 22:13:37 Aha. The symlinks in my paths are breaking everything! The ibot logs are useful :) Mar 11 22:13:48 RP: libtool? Mar 11 22:13:48 what does lsb compliant mean... that you build one binary and it magically works on any arch, now? Mar 11 22:13:57 i thought LSB was a layout for the fs Mar 11 22:14:07 jnc: well lsb is IA-32 only Mar 11 22:14:11 yeah Mar 11 22:14:16 jnc: so it works on any lsb compilant distribution Mar 11 22:14:28 jnc: well nobody else standardized the ABI on different CPUs Mar 11 22:14:33 okay so you're not even talking about the same thing Mar 11 22:14:34 koen: correct Mar 11 22:14:55 jnc: well I'm saying what is possible on Linux/ELF Mar 11 22:15:00 i'm talking about building an SDK for a given build platform from a single build machine Mar 11 22:15:07 jnc: on OSX/GnuStep you could create FAT binaries Mar 11 22:15:23 so we could set a build machine to build the SDK for many types of build host archs Mar 11 22:15:26 jnc: but that is not even in the focus on OE Mar 11 22:15:29 zecke: FAT is old, Universal is all the rage now Mar 11 22:15:34 cross compiling is not the focus? Mar 11 22:15:50 jnc: creating one binary for all host systems Mar 11 22:16:02 why would you create one binary for all host systems Mar 11 22:16:07 jnc: and I think I personally do not want to distribute prebuild OE toolchains Mar 11 22:16:17 jnc: how else should it run on all systems? Mar 11 22:16:24 jnc: cross compiling cross toolchains is going to break horribly given that nobody ever does that Mar 11 22:16:58 reenoo: debian does that a lot, so i don't know what you get that nobody does it Mar 11 22:17:08 jnc: ?? Mar 11 22:17:22 debian is a pretty huge project, and the build system allows for exactly that sort of thing Mar 11 22:17:24 jnc: you do realize that you're talking nonsense don't you? Mar 11 22:17:31 perhaps Mar 11 22:17:34 it makes sense to me Mar 11 22:17:46 debian does native builds on all supported archs Mar 11 22:18:03 which is why the armeb port is taking months Mar 11 22:18:30 that's a policy thing though, much of say, 68k development, is cross-compiled from other faster machines Mar 11 22:19:04 ack... i just remembered, 68k is not exactly supported in etch :/ Mar 11 22:19:25 wookeys emdebian talk raises some problems with debian's policy Mar 11 22:20:11 jnc: that's a different story, you proposed cross compiling $any->$any toolchains on the same build box Mar 11 22:20:36 it confuses me why you wouldn't build a single toolchain package for a given arch, for all non-toolchain developers to use Mar 11 22:20:50 to make sure everyone is reporting against the same thing exactly Mar 11 22:22:24 jnc: putting staging/ and cross/ under package management would be a nice first step to achieve that goal Mar 11 22:23:22 hm .. Mar 11 22:24:43 reenoo: hmm RP and me are working on it... we need more input first though Mar 11 22:28:32 jnc: This is something we'd like to give the option of in future. The solution is to put staging under package control. This is planned but needs a lot of careful thought and work Mar 11 22:28:37 any objections to overwriting the default wpa-supplicant.conf file with something vastly more simple? Mar 11 22:29:10 jnc: What breaks as a result of that change? Mar 11 22:29:23 nothing really, actually it should work more often Mar 11 22:29:28 that's why i want to change it Mar 11 22:29:43 currently the default wpa-supplicant.conf does not work Mar 11 22:30:04 i.e. if you try to associate with an access point, and the SSID is 'example', it won't work. Mar 11 22:30:10 not likely Mar 11 22:30:36 It certainly needs some attention Mar 11 22:30:40 i propose to change it so that by default, unless you change the config file, wpa-supplicant.conf will be set up such that it will not do anything in any case, allowing WEP to work Mar 11 22:31:04 also changing it so that update_config=1 by default Mar 11 22:31:34 I'd mention the changes on oe@ maybe linking to a patch. If nobody complains, commit it Mar 11 22:31:41 hmm. Mar 11 22:51:17 after this push. Mar 11 22:52:41 03jnc 07org.oe.dev * r53e14a99... 10/ (7 files in 4 dirs): Mar 11 22:52:43 WPA Supplicant updates, and inclusion of version 0.5.1 for distros with gnutls Mar 11 22:52:45 already being built. Resolves licensing issues, and non-working supplicant Mar 11 22:52:47 issues for those with 2.6/2.4 kernels. Report breakage to Eric (jnc). Mar 11 22:53:54 * emte hates alsa Mar 11 22:58:30 jnc: just to understand it, you just discarded RP's comment and pushed? Mar 11 23:01:54 see ya'll next week. gotta do actual real work in san fran this week. Mar 11 23:06:23 jnc: the latest wpa-supplicant... Does it include the zd1211-patches I added earlier? Mar 11 23:07:29 * NAiL pulls Mar 11 23:15:30 ah, it doesn't. Good Mar 11 23:35:22 03nail 07org.oe.dev * r8df12327... 10/packages/zd1211/ (zd1211-r59/makefile-unslung.patch zd1211_r59.bb): zd1211: Fixup for Unslung 2.4-kernel Mar 11 23:55:23 'night all Mar 12 00:26:05 anyone awake to take a quick look at this before i submit it? Mar 12 00:26:07 http://oe.pastebin.com/597219 Mar 12 00:41:10 guess no one is arround ... i'll just submit it then Mar 12 00:43:51 hope my documentation was good enough Mar 12 00:46:03 <[g2]> so what's the deal with dropbear.inc ? bb -D says ERROR but doesn't identfy much Mar 12 00:47:13 can someone look and test this if they have time? http://bugs.treke.net/show_bug.cgi?id=735 Mar 12 00:47:23 its the NONSYSV compatability patch Mar 12 00:47:24 * jacques was just wondering the same thing about dropbear.inc Mar 12 00:47:41 it passed all my tests Mar 12 00:49:58 Afternoon/evening, all Mar 12 00:58:54 <[g2]> jacques there's a 1 line require that dies on the parsing error Mar 12 00:59:18 <[g2]> dropbear_... is just a 1 liner that does a require dropbear.inc Mar 12 01:00:07 <[g2]> my workaround (HACK) was to toast the _48.bb and symlink dropbear.inc to it Mar 12 01:00:19 <[g2]> fyi Mar 12 01:10:12 okay i fixed and updated my patch Mar 12 01:10:16 :) Mar 12 01:11:44 Anyone here a Jlime ninja? Mar 12 01:13:08 anyone in here giving me a strange look for asking for a ninja? Mar 12 01:13:30 anyone in here wonder what Jlime is? Mar 12 01:13:35 * emte sticks up hand Mar 12 01:14:12 haha my mistake Mar 12 01:14:21 jlime is embedded linux for the HP Jornada series Mar 12 01:14:34 ah Mar 12 01:14:39 I gogoled around a bunch and saw #oe (presumably on freenode?) come up frequently Mar 12 01:14:42 i was guessing it was some java app Mar 12 01:16:30 nice Mar 12 01:16:59 maybe i need to toss tmp ... Mar 12 01:29:43 This is all starting to look greek to me. I'm totally green when it comes to cross compiling anything. My host pc is x86 and my handtop is sh3. I've googled for days and I'm not quite getting it. Can anyone recommend a good tutorial for crosscompilation for sh3? Mar 12 01:34:13 maou2025: assuming it's supported in OE (and I think it is) just follow the normal OE stuff Mar 12 01:34:19 maou2025: openembedded.org Mar 12 01:37:55 excellent Mar 12 01:40:31 hmm... that site seems to not be responding, while otehrs are fine. Is oe down at the moment? Mar 12 01:41:32 I thought it had been brought back up.... Mar 12 01:41:35 try oe.handhelds.org Mar 12 01:44:24 does oe have any ipkg feeds for sh3? Mar 12 01:55:08 maou2025: that's a distro question, OE is the metadata Mar 12 01:55:16 I have no idea what distro that would be Mar 12 02:15:01 * emte wonders what an sh3 is Mar 12 02:16:05 JustinP, i dont suppose you would like to test my patch would ya? Mar 12 02:16:12 emte: Processor Mar 12 02:16:25 not familiar with it Mar 12 02:16:39 One of the old ARM cores, IIRC.. Mar 12 02:16:49 pre strongarm? Mar 12 02:19:32 Er, looks like I was wrong.. Mar 12 02:19:35 http://72.14.207.104/search?q=cache:y7hJmItFSW8J:encyclopedia.ekist.de/s/su/superh.html+superh+timeline&hl=en&gl=au&ct=clnk&cd=4&client=firefox-a Mar 12 02:19:50 blech Mar 12 02:19:55 oe is screwed Mar 12 02:20:08 wth is it building qemu ? Mar 12 02:20:21 its supposed to be building fakeroot Mar 12 02:21:04 dont tell me someone stuck qemu in the bootstrap stuff ... Mar 12 02:21:09 or the toolchain Mar 12 02:22:08 it should be an option, not part of the tool chain Mar 12 02:22:19 i want to run the stupid code on my host Mar 12 02:22:22 dont* Mar 12 02:25:14 okay lets toss tmp and try bootstrap to confirm this sillyness Mar 12 02:26:45 * emte reads log to figure out why this happened Mar 12 02:28:37 RP, did you make qemu a toolchain default ? Mar 12 02:31:07 no it doesnt look like it was Rp Mar 12 02:32:26 this is dumb Mar 12 02:32:42 | ERROR: "gcc" looks like gcc 4.x Mar 12 02:32:42 | QEMU is known to have problems when compiled with gcc 4.x Mar 12 02:32:42 | It is recommended that you use gcc 3.x to build QEMU Mar 12 02:32:42 | To use this compiler anyway, configure with --disable-gcc-check Mar 12 02:32:42 | FATAL: oe_runconf failed Mar 12 02:34:12 PREFERRED_VERSION_gcc ?= "3.4.4" Mar 12 02:34:12 PREFERRED_VERSION_gcc-cross ?= "3.4.4" Mar 12 02:34:12 PREFERRED_VERSION_gcc-cross-initial ?= "3.4.4" Mar 12 02:42:54 emte: which? Mar 12 02:47:20 both Mar 12 02:47:39 i dont want qemu and it doesnt obey the distro gcc setting Mar 12 02:51:35 there goes my stuff for redboot i guess, no point i changing it's toolchains Mar 12 02:51:46 in* Mar 12 02:52:06 rm -Rf packages/redboot Mar 12 02:57:04 * emte needs to teach xchat to ignore commands i type in the wrong window Mar 12 02:59:53 have it try running everything you say as a command, if it works, then you must have meant it for another window and it shouldn't display it Mar 12 03:01:15 that is a valid partial solution Mar 12 03:01:28 commands the user doesnt have access to will still print Mar 12 03:01:52 its one of those silly wishfull thinking features Mar 12 03:02:20 this must be my day to patch bitbake.conf or something Mar 12 03:04:21 in this case, it would also have attempted to trash some random dir, depending on what working directory your client runs in :-) Mar 12 03:09:08 lol Mar 12 03:09:10 true Mar 12 03:09:21 but you could chrrot and test for safety Mar 12 03:09:26 chroot* Mar 12 03:12:01 ohay lets see if my new magic works Mar 12 03:13:35 i am actually curious why this didnt exist before actually ... Mar 12 03:14:57 if this behaves you might be interested in it actually jnc Mar 12 03:15:33 its a local.conf version override for CC Mar 12 03:16:16 so you dont need to change your HOST's gcc you can specify bitbake to use a specific one Mar 12 03:21:27 woohoo :) Mar 12 03:21:30 ccache gcc-3.3 Mar 12 03:21:34 looks good :) Mar 12 03:21:56 lets see if qemu balks again Mar 12 03:32:46 jnc, http://bugs.treke.net/show_bug.cgi?id=736 if your interested Mar 12 03:33:20 hmm Mar 12 03:34:32 seems qemu ignores this for testing ... Mar 12 03:37:03 nice ... qemu doesnt obey our globals ... Mar 12 03:37:39 # default parameters Mar 12 03:37:39 prefix="" Mar 12 03:37:39 interp_prefix="/usr/gnemul/qemu-%M" Mar 12 03:37:39 static="no" Mar 12 03:37:39 cross_prefix="" Mar 12 03:37:42 cc="gcc" Mar 12 03:37:44 host_cc="gcc" Mar 12 05:28:02 emte: wrong way to do it. noble, but wrong. Mar 12 05:28:54 jnc, what is the correct way? Mar 12 05:29:00 emte: the solution is to fix qemu Mar 12 05:29:02 i am about to remove my post Mar 12 05:29:14 http://www.gibix.net/projects/qemu/ Mar 12 05:29:20 personally i dont want it to begin with Mar 12 05:29:48 the solution is always to fix upstream when possible Mar 12 05:31:09 yes i agree Mar 12 05:31:17 http://www.google.com/search?hl=en&q=qemu+gcc4&btnG=Google+Search Mar 12 05:31:24 i even patched most of qemu Mar 12 05:31:41 but now i've decided to build a custom patch to remove qemu Mar 12 05:31:46 for myslef Mar 12 05:31:46 the trouble with qemu is the author does not agree with the licensing of qemu Mar 12 05:32:02 ? Mar 12 05:32:16 the author of qemu doesnt agree with his own licensing>? Mar 12 05:32:16 amongst other reservations of "i need to profit off my creation"isms Mar 12 05:32:45 guy does all this work, and does not get paid. he seems to have some kind of moral issue with that Mar 12 05:32:59 so it's been kind of a pain to submit patches Mar 12 05:33:11 if you submit a patch, there is a fear your work will be sold Mar 12 05:33:19 however rediculous that may be in reality Mar 12 05:33:42 so why is this being forced into OE then? Mar 12 05:34:08 i'm not sure. i don't have a problem with it. Mar 12 05:34:17 upstream qemu is just a tricky subject to deal with Mar 12 05:35:10 hmm is see where the issue with qemu is Mar 12 05:35:31 i take it we are using qemu without the accelerator module? Mar 12 05:35:54 the accelerator module is not really of value IMO Mar 12 05:35:55 since it is under propritary license Mar 12 05:36:09 it only seems to apply to x86 emulation on x86 host Mar 12 05:39:11 NAiL: i wasn't sure how to include that. it is open to be re-added Mar 12 05:39:28 NAiL: currently the wpa-supplicant thing is in flux, pending a simplification Mar 12 05:41:29 NAiL: if you have a patch, please try to work with upstream though Mar 12 05:43:37 it would appear that the patch broke stuff, and having removed it was A Good Thing (tm) Mar 12 05:43:48 i still dont understand why we are using qemu Mar 12 05:44:01 or why we are using an emulator at all Mar 12 05:44:32 NOTE: Task failed: Fetch failed: http://ftp.handhelds.org/pub/projects/gpe/source/xst-0.15.tar.gz Mar 12 05:44:35 how the hell. Mar 12 05:44:48 ? Mar 12 05:45:02 http://FTP.HANDHELDS.ORG/PUB/... Mar 12 05:45:02 i wish i could compile anything anymore Mar 12 05:45:13 * jnc looks confused Mar 12 05:45:25 no idea Mar 12 05:45:52 maybe someone else has a poor idea of subdomain names liek my school does Mar 12 05:46:04 its ftp://www.blah Mar 12 05:46:25 ftp is how it was done before http, IIRC Mar 12 05:46:33 gopher too Mar 12 05:46:46 www is for html Mar 12 05:47:08 by default and rfc Mar 12 05:47:32 is where my comment came from Mar 12 05:48:36 whats the command to roll back monotone? Mar 12 05:48:59 just pull an earlier revision, i guess Mar 12 05:49:10 i pushed a fix for the URL btw Mar 12 05:50:13 03jnc 07org.oe.dev * r130a0e55... 10/conf/bitbake.conf: s/ftp.handhelds.org/handhelds.org/ for teh mirror URL. fixes b0rkage. Mar 12 05:50:25 yeah i am trying to figure out how Mar 12 05:50:32 well, i don't know Mar 12 05:50:39 man page only lists date stuff for commiting Mar 12 05:51:34 i wonder if njs is awake or not Mar 12 05:51:45 hmm? Mar 12 05:51:53 hey! :) Mar 12 05:51:54 jnc: what do you want to do? Mar 12 05:51:59 err, misread Mar 12 05:52:02 emte: what do you want to do? :-) Mar 12 05:52:09 how do i specify a date to rollback monotone to? Mar 12 05:52:27 taht part seems to be missing from the man page Mar 12 05:52:31 that* Mar 12 05:52:46 what i would do is read the monotone log, look for a revision you want, and pull it Mar 12 05:53:05 hmm Mar 12 05:53:19 * emte tries to figure out where my monotome log lives Mar 12 05:54:04 it's not an actual file, reallly Mar 12 05:54:12 'monotone log' command will give you the log Mar 12 05:54:16 emte: what do you mean by 'rollback to'? change your working copy so it contains something older? Mar 12 05:54:17 monotone log | less Mar 12 05:54:29 yes Mar 12 05:54:38 oh! Mar 12 05:54:42 hrm Mar 12 05:54:47 emte: then yeah, find the rev you want with 'monotone log', then 'monotone update -r ' Mar 12 05:55:30 njs: i am so pleased you offer help Mar 12 05:55:34 you could also try doing clever things with selectors, I guess Mar 12 05:55:36 * jnc dances joyfully Mar 12 05:55:44 ? Mar 12 05:56:01 but there isn't really any selector for "some random rev on this date" Mar 12 05:56:23 you could say 'monotone up -r d:2006-01-01', but it would probably say "there are several revs matching that, which do you want?: " Mar 12 05:56:42 hey, that would be shorter Mar 12 05:57:01 i am currently waiting for my buffer to overflow .... Mar 12 05:57:23 sounds like a cooking process for food Mar 12 05:58:12 (in fact, just -r 2006-01-01 would probably work, because monotone will try to guess what sort of selector you want if you give it something ambiguous :-) but the names like "d:" are short enough I like to use them, personally.) Mar 12 05:58:19 http://venge.net/monotone/docs/Selectors.html Mar 12 05:58:51 the next time i hit "end" i shoudl slap myself Mar 12 05:58:58 jnc: glad I could help Mar 12 05:59:24 thanks njs i'll try the date thing here in a min Mar 12 05:59:32 njs: now that i'm using monotone, i really like it Mar 12 05:59:42 i wonder how it would compare to something like perforce Mar 12 06:00:09 * emte likes that metal one :) Mar 12 06:00:17 jnc: cool Mar 12 06:00:47 bah Mar 12 06:00:50 jnc: I don't really know; perforce isn't in the universe of projects I can really use :-) (both because it's expensive, and just to avoid any whiff of legal issues.) Mar 12 06:00:54 i am looking at the wrong end :( Mar 12 06:01:08 ah good idea there Mar 12 06:01:33 jnc: p4 is much more like cvs than like monotone, though -- it still uses RCS as a backend, still uses a central server you check out from/commit to, etc. Mar 12 06:02:05 i forget, are we able to give arbitrary names to revisions? Mar 12 06:02:10 jnc: 'tag'? Mar 12 06:02:15 hmm okay Mar 12 06:02:30 'monotone tag ' Mar 12 06:02:40 * jnc swears at bitbake.conf Mar 12 06:02:45 then you can say 't:' anywhere you would otherwise give a revision id Mar 12 06:03:04 it's so simple to use Mar 12 06:03:54 ... Mar 12 06:04:09 i think that was what they used to say about BK Mar 12 06:05:03 tat was yr-mth-day ? Mar 12 06:05:14 %Y-%m-%d Mar 12 06:09:44 aha thankyou Mar 12 06:09:56 hopefully i went far enough back to build again Mar 12 06:14:27 you can do any prefix of --T:: Mar 12 06:15:22 yeah just wasnt sure it it was y-m-d or y-d-m Mar 12 06:15:22 plus some other things like "2 weeks ago", see that page from the manual I linked to... Mar 12 06:15:42 i missed that url ... Mar 12 06:15:47 * emte scrolls up Mar 12 06:16:02 nod. I'm pretty sure there's never any such thing as y-d-m Mar 12 06:16:44 not standard, bit i see all types of permutations with colors etc Mar 12 06:16:48 but* Mar 12 06:16:49 it's the ISO order, which was partly chosen to be unambiguous wrt day versus month, since no-one except ISO puts the year first... Mar 12 06:17:05 true Mar 12 06:17:15 i like the iso order since its easily sortable Mar 12 06:17:15 I guess I can believe people do it anyway, though, just to screw things up for the rest of us ;-) Mar 12 06:17:21 kergoth: yeah, that too... Mar 12 06:18:07 hey kergoth Mar 12 06:18:11 hey Mar 12 06:18:30 kergoth, do you happen to know the reason for making qemu a part of the toolchain? Mar 12 06:18:57 qemu is now used for part of the glibc locale generation. read teh changelogs Mar 12 06:19:05 thats all i know, havent touched oe in like a year Mar 12 06:19:43 * emte wonders how using an emu for glibc is at all useful Mar 12 06:20:18 uh? Mar 12 06:20:21 do you happen to know which server the changelogs are? Mar 12 06:20:27 my guess would be it's to solve one of those obnoxious cross-compiling issues, where some build system first builds a program, and then runs it to build some other part of the system Mar 12 06:20:29 they end up having to run an arm binary to generate the binary locale files Mar 12 06:20:54 so if you're cross-compiling, you either have to figure out how to _not_ cross-compile that one part, or else use something like qemu to run the cross-compiled binary locally... Mar 12 06:21:07 we're now using the old style granular binary locale files, apparently stops from running into the OOM condition on the first boot locale generation Mar 12 06:21:09 exactly Mar 12 06:21:14 * kergoth shrugs Mar 12 06:21:21 qemu is used to binary compile the locales Mar 12 06:21:24 takes *forever* Mar 12 06:21:48 especially on my poor 700Mhz box Mar 12 06:22:08 i take it this is more of an issue for MAD machines that dont like intel code? Mar 12 06:22:11 AMD* Mar 12 06:22:18 huh? Mar 12 06:22:22 that has nothing to do with this Mar 12 06:22:34 no? hmm Mar 12 06:22:48 emte: it is an issue for machines that don't like arm code, i.e., all the machines that aren't arms :-) Mar 12 06:22:50 it's used to run an ARM binary so that the locales can be compiled for the machine so that the post-conf doesn't have ot do it Mar 12 06:23:06 i thought all AMR chips adheared to intels compatability Mar 12 06:23:11 arm* Mar 12 06:23:22 oh, is that what you meant? Mar 12 06:23:31 yeah Mar 12 06:23:33 well...you can't run an arm binary on a normal PC....AFAIK Mar 12 06:23:45 correct, arm is a different architecture with a completely different instruction set Mar 12 06:23:52 anyway, qemu could also possibly be used for compiling, say, emacs ;-) Mar 12 06:23:56 exactly Mar 12 06:24:03 so what that has to do with Intel vs. AMD.... Mar 12 06:24:36 ARM isnt intel anyway. ARM is ARM. intel has some processors with arm cores, but thats about it Mar 12 06:24:44 my misinterpretation about intel and its instruction compatability Mar 12 06:24:45 i think emte should do some reading on the web Mar 12 06:24:49 that's what I thought Mar 12 06:25:13 start with googling on what exactly locale data is Mar 12 06:25:15 and go from there Mar 12 06:25:31 * jnc reluctantly nukes his oetmp and starts over Mar 12 06:25:35 i read, but about far too many different things at once Mar 12 06:25:40 *mutter* stupid URL breakage *mutter* Mar 12 06:25:59 they kinda merge after a while Mar 12 06:34:21 nice, codesourcery's arm work is being put into gcc4.1 Mar 12 06:34:30 now i can del that bookmark Mar 12 06:44:22 that news is soooo last week Mar 12 06:47:08 why didnt you tell me? Mar 12 07:40:04 * france is away: Away Mar 12 08:02:26 03rpurdie 07org.oe.dev * r1ba45cf5... 10/packages/linux/linux-openzaurus_2.6.15.bb: linux-oz-2.6: Update LED patches to match those in -mm Mar 12 08:29:41 mhhh ... why for building bootstrap-image qemu is need ??? (bitbake is downloading right now) ... Mar 12 08:30:39 gremlin[it]: OE uses it for locale generation but this was disabled again in later versions of OE Mar 12 08:32:46 thank RP : I did a oe chechout this morning and is the first time i notice qemu since months ... Mar 12 09:47:21 hi Mar 12 09:47:29 someone use git fetcher? Mar 12 10:03:55 hrw, I do . Mar 12 10:05:04 leoncamel: why it does not use tarballs after once fetched (when tag is used)? Mar 12 10:08:34 ARHG! Mar 12 10:08:42 hrw@bitbake:~/devel/build/test$ bitbake -b ../../oe/org.openembedded/packages/linux/linux-openzaurus-collie_git.bb -f -ccompile Mar 12 10:08:45 NOTE: package linux-openzaurus-collie-git: started Mar 12 10:08:48 NOTE: package linux-openzaurus-collie-git-r0: task do_fetch: started Mar 12 10:08:52 aRGH! Mar 12 10:10:03 good morning all Mar 12 10:10:09 http://pastebin.com/597592 Mar 12 10:10:10 hrw: did you specify a tag? Mar 12 10:10:13 ~curse git fetcher Mar 12 10:10:16 May the fleas of a thousand camels infest your most sensitive regions, git fetcher ! Mar 12 10:10:27 SRC_URI = "git://kernel.org/pub/scm/linux/kernel/git/pavel/zaurus.git;tag=2de82410a61d8c76b5c26f1d59f7e1657dc4ea40 \ Mar 12 10:11:05 this behaviuor makes using git fetcher too painful Mar 12 10:11:30 I will probably switch to fetch->tarball and use tarball in SRC_URI Mar 12 10:12:02 -rw-r--r-- 1 hrw hrw 1183412038 Mar 12 11:10 /home/hrw/devel/sources/dl_dir/git_kernel.org.pub.scm.linux.kernel.git.pavel.zaurus.git.tar.gz Mar 12 10:12:06 -rw-r--r-- 1 hrw hrw 51641471 Mar 12 11:13 /home/hrw/devel/sources/dl_dir/git_kernel.org.pub.scm.linux.kernel.git.pavel.zaurus.git_2de82410a61d8c76b5c26f1d59f7e1657dc4ea40.tar.gz Mar 12 10:12:10 but which one it will be ;( Mar 12 10:12:19 hi koen Mar 12 10:13:12 hey hrw Mar 12 10:13:31 hrw, hmm, if you specific a tag, I think it will be the second one. Mar 12 10:13:41 #define git totally-evil-stuff-which-I-refuse-to-use-in-src-uri-because-of-those-over-1G-archives Mar 12 10:14:34 leoncamel: I was thinking like that before Mar 12 10:14:41 sent 124 bytes received 187298 bytes 53549.14 bytes/sec Mar 12 10:14:41 total size is 1209948353 speedup is 6455.74 Mar 12 10:14:41 Already up-to-date. Mar 12 10:15:04 but that total size looks not like 50M.. Mar 12 10:15:17 brb Mar 12 10:17:55 hrw, maybe it dose not contains the whole git repository ? Mar 12 10:20:31 leoncamel: ? Mar 12 10:20:39 * hrw is green when it comes to git Mar 12 10:21:41 50M archive contain linux source. 1.2G archive contain git indexes/packs Mar 12 10:22:18 monotone is much better then git - atleast I do not need to buy new harddrive just for metadata Mar 12 10:22:58 monotone storing the whole kernel history probably wouldn't be any smaller. Mar 12 10:23:45 we plan to add ways to do partial pulls, grab the latest snapshot without history, that sort of thing, but other things have been higher priority. Mar 12 10:25:10 SRC_URI = "file://home/hrw/devel/sources/dl_dir/git_kernel.org.pub.scm.linux.kernel.git.pavel.zaurus.git_2de82410a61d8c76b5c26f1d59f7e1657dc4ea40.tar.gz works much beter Mar 12 10:25:56 hrw, hmm, I don't think so . the git is good than monotone. the performance of monotone is very poor.. Mar 12 10:26:37 monotone is much faster for me than git Mar 12 10:26:48 hrw, maybe there is a bug about git fetcher .. the git SCM tools is very good, and very stable now. Mar 12 10:26:50 with monotone I don't have to google for documentation for every action ;) Mar 12 10:26:59 the things that monotone takes a long time to do are things that git does not do at all :-) Mar 12 10:27:10 (e.g., detect corruption in your data) Mar 12 10:28:02 maybe this is a good thing, maybe not, but it seems important to point out if comparing them side-by-side... Mar 12 10:28:23 koen, hmm, the git is more effective. for example, the first time I clone the repository, it will eat up all my cpu .... what a crazy thing .. Mar 12 10:28:36 I like how cvs allows you to store the merge conflict inside the file itself Mar 12 10:28:42 leoncamel: but I just want to fetch one git tag - there should be a way to fetch snapshot Mar 12 10:28:51 koen, and git is widely use in kernel development .. :) Mar 12 10:29:08 heh Mar 12 10:29:11 emacs is as well Mar 12 10:29:23 and I'm not going to use emacs in my life anymore Mar 12 10:29:46 leoncamel: I have 1.5Mbps at home - but I also use 128kbps lines sometimes and does not feel that I have to fetch 1.2G of data for 50M archive Mar 12 10:29:56 hrw, hmm, I havn't read the git fectcher now. and I think the git fetcher should use cogito as well. Mar 12 10:30:38 hrw, the git fetcher should clone whole repository into sources directory, and cg-export a package into working directory . Mar 12 10:31:34 anyway I share many people feeling that there should be a way to fetch only snapshot of metadata Mar 12 10:32:09 hrw, hmm, you needn't clone whole respository every time. It is something like monotone. You should update your git respsitory before your building .. Mar 12 10:32:28 hrw: no, that's all wrong, Linus fetches the complete tree, so that is the *only* *correct* way to do that Mar 12 10:32:37 linus is *always* right Mar 12 10:32:37 in monotone, the most common web frontend has a way to grab a tarball... maybe gitweb is similar? Mar 12 10:32:43 leoncamel: I know that git is distributed and how it works Mar 12 10:33:16 leoncamel: I'm just refusing to use it when I have to fetch something from git tree Mar 12 10:33:36 hrw, hmm, AFAIK, the git is something like monotone. They are all distributed SCM tools. Mar 12 10:33:53 git isn't actually and scm Mar 12 10:34:07 leoncamel: read my broken lips: I know git is distributed and how it works Mar 12 10:34:09 since it doesn't have a sane way to push data Mar 12 10:34:18 and please stop that discussion... Mar 12 10:34:55 hrw, first you shuould "clone" whole respository at the first time. after that, you should do "pulling" from remote( or parent) respsitory. the git respsitory will place into a ".git" directory". Mar 12 10:35:02 when I realise how long it would take to fetch one git tree just to test collie-2.6 kernel on my home connection... Mar 12 10:35:07 ~hail koen for ewi account Mar 12 10:35:09 * ibot bows down to koen for ewi account and chants, "I'M NOT WORTHY!!" Mar 12 10:35:18 :) Mar 12 10:35:49 atleast it has big enough pipe to not waiting few days before git will fetch 50M of sources Mar 12 10:35:49 I'm still trying to get the connection upgrade to Gb/s, but I need a better excuse :) Mar 12 10:36:09 hrw, :) Mar 12 10:36:23 koen: oz 3.5.4 will suck your conenction... we had 150G/month on sf.net (if not more) Mar 12 10:36:51 hrw: heh, I used to have that per day :) Mar 12 10:37:14 ;) Mar 12 10:37:42 the server is inside a lab, so it's for 'educational' purposes :) Mar 12 10:37:56 which means unlimited bandwidth Mar 12 10:37:58 koen, you should build more mirror site .. :) Mar 12 10:38:18 270G/month on last release Mar 12 10:38:43 I have 10G/month limit on hosting... Mar 12 10:38:56 bittorrent! Mar 12 10:39:06 hrw: angstrom also has unlimited traffic Mar 12 10:39:11 (I'm am shocked that we still have "mirrors", I mean, whose idea is this...) Mar 12 10:39:14 koen: nice Mar 12 10:39:39 njs: I prefer to use wget then bittorrent to fetch data Mar 12 10:39:47 njs: if it weren't for firewalls I'd use bittorrent more Mar 12 10:40:04 hrw: hmm... sounds like a bug in wget :-) Mar 12 10:40:23 ok. enough for now Mar 12 10:40:40 njs: how about putting bittorent style fetching into monotone? Mar 12 10:40:41 koen: mickeyl said something about 3.5.4 building progress? Mar 12 10:40:56 hrw: not that I know off Mar 12 10:41:01 ok Mar 12 10:41:27 koen: hmm. arguably useful for initial pulls. have to see how much of a bandwidth drain they are. Mar 12 10:42:41 cu Mar 12 10:42:45 hrw: cu Mar 12 10:43:03 njs: and change 'monotone serve' to 'monotone distribute' ;) Mar 12 10:43:57 koen: it really should be possible for monotone to advertise via mDNS, tunnel over XMPP, etc., ... just need someone to do it :-) **** ENDING LOGGING AT Sun Mar 12 10:59:56 2006