**** BEGIN LOGGING AT Wed Jun 08 02:59:56 2011 Jun 08 06:20:18 JaMa|Wrk, do you have anything to push today? Jun 08 06:22:19 oe.dev? I fear not Jun 08 06:22:46 okay, I did some work last night on the server...and if you did I was going to watch things. Jun 08 06:23:22 I've pushed few commits to meta-oe-contrib repo and I can again if it's usefull for you Jun 08 06:23:38 yes, it is. Jun 08 06:23:40 thanks. Jun 08 06:25:02 done Jun 08 06:25:13 went fast... Jun 08 06:25:32 good...about 2X as fast and a lot less overhead. Jun 08 06:25:49 nice, good job Jun 08 06:26:00 was it some hungry hook or something? Jun 08 06:26:24 no it was running as an sv task, I switched it to xinetd. Jun 08 06:27:23 and it's so much faster when started from xinetd? interesting Jun 08 06:27:45 ya, I was surprised too...testing earlier proved that out as well. Jun 08 06:28:20 would not have been my first choice but its working out. Jun 08 06:28:32 in about 30minutes I'll really see. Jun 08 06:28:38 as CET gets going. Jun 08 06:36:28 hello--????? Jun 08 06:40:34 i am compiling a file connMan.cxx for mini6410 ,through openemmbedded ,the file gets compiled properly but when I run the complied binary on mini6410 target ,it gives seg fault on the xterm at the same time on the gtk term it gives the following error Jun 08 06:40:56 Unable to handle kernel paging request at virtual address 4034b000 Jun 08 06:40:56 pgd = c719c000 Jun 08 06:40:56 [4034b000] *pgd=57182031, *pte=00000000, *ppte=00000000 Jun 08 06:40:56 Internal error: Oops: 817 [#1] Jun 08 06:40:56 Modules linked in: Jun 08 06:40:57 CPU: 0 Not tainted (2.6.28.6-FriendlyARM #3) Jun 08 06:40:59 PC is at v6_coherent_kern_range+0x4/0x24 Jun 08 06:41:03 LR is at arm_syscall+0x1e0/0x328 Jun 08 06:41:05 pc : [] lr : [] psr: 80000013 Jun 08 06:41:07 sp : c7191e80 ip : c7186e70 fp : c7191fa4 Jun 08 06:41:09 r10: 00000000 r9 : c7190000 r8 : 00000000 Jun 08 06:41:11 r7 : 000f0002 r6 : 00000000 r5 : 40374000 r4 : 4033a000 Jun 08 06:41:13 r3 : c0620be4 r2 : c7186e88 r1 : 40374000 r0 : 4034b000 Jun 08 06:41:15 Flags: Nzcv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user Jun 08 06:41:17 Control: 00c5387d Table: 5719c008 DAC: 00000015 Jun 08 06:41:19 Process connMan.bin (pid: 1091, stack limit = 0xc7190260) Jun 08 06:41:21 Stack: (0xc7191e80 to 0xc7192000) Jun 08 06:41:23 1e80: c7191eb4 c7191e90 c00ffe74 c00ffce4 c0622800 c7061380 c70613b0 c7061380 Jun 08 06:41:25 1ea0: 00000000 4033e66c c7191ed4 c7191eb8 c0622800 c7186dc0 c7191efc c7191ec8 Jun 08 06:41:27 1ec0: c0141bb4 40374000 4033a000 c7182ce8 40374000 c7186e70 40374000 40374000 Jun 08 06:41:29 1ee0: c7191efc c7191ef0 c042f254 60000013 c719a034 00000000 40374000 00000005 Jun 08 06:41:33 1f00: 60000013 c719a034 c7191f2c c7191f18 c042f23c c0227418 00000000 c719a034 Jun 08 06:41:35 1f20: c7191f5c c7191f30 c0222468 c042f234 c7061380 00000000 40374000 00000000 Jun 08 06:41:37 1f40: 40374000 00000005 c7190000 00000000 c7191f6c c7191f60 c011ede8 c0222330 Jun 08 06:41:39 1f60: c7191fa4 c7191f70 c0142e0c c011ede4 08100075 4033e66c c7191f94 c7186e70 Jun 08 06:41:41 1f80: c010c1d0 4001f348 bef88938 4001f348 bef88938 40348084 00000000 c7191fa8 Jun 08 06:41:43 1fa0: c00ecd80 c00f1250 4001f348 bef88938 4033a000 40374000 00000000 00000015 Jun 08 06:41:45 1fc0: 4001f348 bef88938 40348084 000f0002 403486ac 4033b4dc 4033a000 bef889f4 Jun 08 06:41:47 1fe0: 00019c20 bef88928 4000c5e8 4000c604 60000010 4033a000 00000000 00000000 Jun 08 06:41:49 Backtrace: Jun 08 06:41:51 [] (arm_syscall+0x0/0x328) from [] (ret_fast_syscall+0x0/0x2c) Jun 08 06:41:53 r6:40348084 r5:bef88938 r4:4001f348 Jun 08 06:41:55 Code: ee070f15 e1a0f00e e1a0f00e e3c0001f (ee070f3a) Jun 08 06:41:57 ---[ end trace a4bfee1378497cd9 ]--- Jun 08 06:42:03 has this problem occured to any one Jun 08 06:42:05 ? Jun 08 06:42:35 my arguments when compiling throuh open emmbedded is as folloes Jun 08 06:43:13 ultimate_warrior: use pastebin Jun 08 06:43:50 pastebin.ca is nice and quick Jun 08 06:44:01 ok Jun 08 06:44:41 seems to be bug in arm kernel... Jun 08 06:45:31 uploading it on pastebin Jun 08 06:51:41 http://pastebin.ca/2076588 Jun 08 06:52:35 CruX: is there any method to solve that bug Jun 08 06:53:29 ultimate_warrior: can you try more recent kernel ? Jun 08 06:54:47 I am using the kernel version 2.6.28 Jun 08 06:58:48 Crux: Is there any problem with the command or arguements that i am passing ,can u suggest any higher kernel version for running on mini6410 ... Jun 08 07:04:12 CruX|: bug in arm kernel? It might be bug in the vendor (friendlyarm) patchset (I would suspect network driver) or whatever the connman is trying to access Jun 08 07:04:38 yes, the kernel shouldn't crash, but who knows what's corrupted where by some buggy driver Jun 08 07:05:06 advice, move to the latest stable kernel, apply no patches if possible and try to reproduce the problem Jun 08 07:05:28 if it persist, report it, otherwise contact your vendor and whine there :) Jun 08 07:05:30 easy Jun 08 07:08:42 ynezz: i will try it with a stable kernel ,but is there a VFP arguements passing problem in ARM 11 Jun 08 07:22:24 good morning Jun 08 07:30:05 anyone who has compiled kernel for ARM .target board mini6410 ,plz tell me the version number Jun 08 07:52:17 been a while since I asked, so: does OE "work" in Mac OS X? Jun 08 08:10:43 Hi everybody Jun 08 08:11:50 is anybody explain me how logging files are numbered with task number? Jun 08 08:12:10 for instance log.do_configure.2240 Jun 08 08:12:18 that's pid, not task number Jun 08 08:12:24 hm Jun 08 08:12:45 can I get this number in bb files? Jun 08 08:14:27 is there a variable name which bitbake use? Jun 08 08:17:43 morning all Jun 08 08:22:04 'morning Jun 08 08:24:30 gm! Jun 08 08:24:54 good morning Jun 08 08:54:58 JaMa|Wrk: about xserver-xorg and randr, does xf86-video-fbdev support it now? Jun 08 09:38:12 when I run a command < Xorg :1 -once -query 192.xxx.xxx.xxx > on mips dev3210 ... I m getting connected to X server.... but the problem is that the color resolution changes... Jun 08 09:38:19 why is it so.... Jun 08 10:50:54 03Paul Eggleton  07master * r32e190e5cc 10bitbake.git/lib/bb/runqueue.py: (log message trimmed) Jun 08 10:50:54 bitbake/runqueue: fix clash when setscene & real tasks done in same build Jun 08 10:50:54 If a build causes a real task to be run when the setscene task has already Jun 08 10:50:54 run then it was possible for dependent packages to be rebuilding at the same Jun 08 10:50:54 time as a rebuild of the packages they depended on, resulting in failures Jun 08 10:50:55 when files were missing. This change looks in the setscene covered list and Jun 08 10:50:56 removes anything where a dependency of the real task is going to be run (e.g. Jun 08 10:51:01 03Richard Purdie  07master * r088d7607e6 10bitbake.git/lib/bb/runqueue.py: Jun 08 10:51:01 bitbake/runqueue: When we successfully fork off a task and there are more processes we can start do so immediately Jun 08 10:51:01 (From Poky rev: 974ea1a190167dcfd831ba1fc5f733e0dc9a6fda) Jun 08 10:51:01 Signed-off-by: Richard Purdie Jun 08 10:51:03 03Richard Purdie  07master * r6108039172 10bitbake.git/lib/bb/ (cooker.py server/process.py): Jun 08 10:51:03 cooker/process.py: Move profiling code to a place it can be reused by different server mechanisms Jun 08 10:51:03 The cooker profiling code isn't server specific so move it to a place Jun 08 10:51:03 where different server backends can use it. Jun 08 10:51:04 Signed-off-by: Richard Purdie Jun 08 10:51:04 03Richard Purdie  07master * r7905257b3d 10bitbake.git/lib/bb/runqueue.py: Jun 08 10:51:05 bitbake/runqueue: When we have successfully collected a pid value return True to ensure we don't sleep in the main loop Jun 08 10:51:05 (From Poky rev: 5a178b547bbf6d99c42cba6821b3f559e585a079) Jun 08 10:51:06 Signed-off-by: Richard Purdie Jun 08 10:51:27 03Paul Eggleton  07master * r1dfeb84f7d 10bitbake.git/lib/bb/runqueue.py: (log message trimmed) Jun 08 10:51:27 bitbake/runqueue: show correct task name for setscene task failure Jun 08 10:51:27 If a setscene task failed previously it was showing an incorrect task Jun 08 10:51:27 name in the error line. This patch ensures we show the correct name, also Jun 08 10:51:28 including the "_setscene" suffix. Jun 08 10:51:28 (From Poky rev: 9d9c68e3536387150e3ce7261f4770576e724dd9) Jun 08 10:51:29 Signed-off-by: Paul Eggleton Jun 08 10:53:01 03Richard Purdie  07master * raa84a900e1 10bitbake.git/lib/bb/cooker.py: Jun 08 10:53:01 bitbake/cooker: Ensure delays are accurately transfered to the idle loops from runqueue Jun 08 10:53:01 (From Poky rev: 579ecccc2244da43b9bc03fb09bb71d8a0219a8b) Jun 08 10:53:01 Signed-off-by: Richard Purdie Jun 08 11:27:07 I need to re-build qt4-embedded-4.6.3 with -qt-gfx-transformed flag Jun 08 11:27:33 what is the way to modify the conf files to make it build successfully with this flag Jun 08 11:28:27 through bitbake Jun 08 11:28:32 through bitbake ? Jun 08 12:19:59 03Richard Purdie  07master * rb420e865f6 10bitbake.git/lib/bb/cooker.py: Jun 08 12:19:59 bitbake/cooker.py: Fix debug message arguments Jun 08 12:19:59 (From Poky rev: 06bb59d) Jun 08 12:19:59 Signed-off-by: Richard Purdie Jun 08 12:20:00 03Richard Purdie  07master * rc30c1349db 10bitbake.git/lib/bb/server/none.py: Jun 08 12:20:00 bitbake/server: Add the none server back for now Jun 08 12:20:00 Signed-off-by: Richard Purdie Jun 08 12:20:00 03Richard Purdie  07master * rd21d5954d0 10bitbake.git/ (bin/bitbake lib/bb/server/process.py lib/bb/server/xmlrpc.py): Jun 08 12:20:01 bitbake/server: Move server specific code into the server backend and create a server API Jun 08 12:20:01 Move the server implementation specific code into the server backend where Jun 08 12:20:02 it belongs and replace this with a set of object method calls which Jun 08 12:20:02 establish the server, detach it and then connect to it using appropriate Jun 08 12:20:03 IPC. Jun 08 12:20:03 Signed-off-by: Richard Purdie Jun 08 12:20:04 03Richard Purdie  07master * r91611c8b88 10bitbake.git/lib/bb/ (event.py ui/uievent.py): Jun 08 12:20:04 xmlrpc/event: Add ability to send pickled events to UI if requested Jun 08 12:37:27 03Joshua Lock  07master * ra33b34b27a 10bitbake.git/lib/bb/ui/crumbs/hobeventhandler.py: Jun 08 12:37:27 bitbake/hob: only show one progress dialog when changing machine Jun 08 12:37:27 Remove a spurious signal emission which resulted in two progress dialogs Jun 08 12:37:27 being shown after changing the machine. Jun 08 12:37:27 (From Poky rev: 7738e86e8b4881056bd44d0f767d17b36159a7bb) Jun 08 12:37:27 Signed-off-by: Joshua Lock Jun 08 12:37:28 Signed-off-by: Richard Purdie Jun 08 12:37:35 03Joshua Lock  07master * r558bbb4ef9 10bitbake.git/lib/bb/ui/depexp.py: Jun 08 12:37:35 bitbake/depexp: closing progress dialog kills gui Jun 08 12:37:35 It seems safe to assume a user hitting the close button Jun 08 12:37:35 on the dialog wants to kill the whole UI. Jun 08 12:37:35 (From Poky rev: e21150f2b18ff4ac85e64ed8814f0268007830c0) Jun 08 12:37:35 Signed-off-by: Joshua Lock Jun 08 12:37:35 Signed-off-by: Richard Purdie Jun 08 12:37:37 03Joshua Lock  07master * r8a56794cd1 10bitbake.git/lib/bb/ui/goggle.py: (log message trimmed) Jun 08 12:37:37 bitbake/goggle: closing the progress dialog kills the UI Jun 08 12:37:37 It's unlikely that someone wants to close the progress dialog Jun 08 12:37:38 yet leave the UI (and BitBake process) running, so hook up Jun 08 12:37:38 the progress dialogs delete-event to exit gtk. Jun 08 12:37:39 (From Poky rev: c9bb30b232396bbdd3c97c1059e972d6a4abf637) Jun 08 12:37:39 Signed-off-by: Joshua Lock Jun 08 12:37:40 03Richard Purdie  07master * r517366b449 10bitbake.git/lib/bb/ui/knotty.py: Jun 08 12:38:27 bitbake/hob: fix cancel button Jun 08 12:38:27 An accidental logic inversion (aka thinko) had the cancel button only Jun 08 12:38:27 cancel a build when the user didn't confirm the cancellation (i.e. clicked Jun 08 12:38:27 no)... Jun 08 12:38:28 (From Poky rev: ff3e326269d27662c830b04e05b571d257b89346) Jun 08 12:38:28 Signed-off-by: Joshua Lock Jun 08 12:38:29 03Richard Purdie  07master * r15af83452c 10bitbake.git/lib/bb/ui/crumbs/hobeventhandler.py: Jun 08 12:38:29 hob/crumbs/hobeventhandler: Remove server specific implementation details missed in the precious cleanup Jun 08 12:38:30 Signed-off-by: Richard Purdie Jun 08 13:17:32 Hi All, playing around with building some images with angstrom-layer on openembedded-core and getting a lot of issues with do_popluate_sysroot failing first time (works second) Jun 08 13:17:44 e.g. NOTE: package autoconf-native-2.68-r0: task do_populate_sysroot: Started ERROR: Package already staged (/home/jic23/src/beagle/setup-scripts/build/tmp-angstrom_2010_x-eglibc/sstate-control/manifest-x86_64-autoconf-native.populate-sysroot)?! NOTE: package autoconf-native-2.68-r0: task sstate_task_postfunc: Failed ERROR: Function 'sstate_task_postfunc' failed Jun 08 13:18:01 Anyone able to guess what is happening or suggest how to figure it out? Jun 08 13:34:19 I'll rephrase based a few more starts. Pretty much every populate_staging is failing. Jun 08 13:34:32 (first time) Jun 08 13:47:15 No one have any idea on the populate_staging issues I'm seeing? Jun 08 14:01:53 jic23: is there a pointer to a log file? I can't recall if we write out logs for sstate Jun 08 14:03:38 nope Jun 08 14:03:58 bluelightning: nope Jun 08 14:04:45 hm... maybe try with -DDD added to the bitbake command line, does it output anything relevant? Jun 08 14:05:27 bluelightning: nothing that I can see. Jun 08 14:05:34 jic23: is SSTATE_DIR set in your local.conf? Jun 08 14:06:09 bluelightning: no. Should it be? Jun 08 14:06:59 I would think so yes, it should be set to point to somewhere (supposed to be outside of TMPDIR, in poky the default is ${TOPDIR}/sstate-cache) Jun 08 14:07:56 where did you get your local.conf btw? was it created by the angstrom setup scripts? Jun 08 14:08:26 bluelightning: yup to angstrom scripts. Was being lazy as haven't build oe of any variety for a few months. Jun 08 14:09:06 jic23: there should be no problem with that approach, I'm just guessing that value is missing from the default file Jun 08 14:09:20 surprised nobody picked up on that... let me give it a try here Jun 08 14:11:09 bluelightning: thanks. Will try again here with that set and report back. Jun 08 14:11:37 jic23: hmm, looks like it indeed does not set it Jun 08 14:14:14 bluelightning: hmm. adding has triggered a whole new set of fun. Jun 08 14:14:37 ERROR: Package already staged (/home/jic23/src/beagle/setup-scripts/build/tmp-angstrom_2010_x-eglibc/sstate-control/manifest-x86_64-m4-native.populate-sysroot)?! NOTE: package m4-native-1.4.16-r0: task sstate_task_postfunc: Failed ERROR: Function 'sstate_task_postfunc' failed ERROR: Task 696 (/home/jic23/src/beagle/setup-scripts/sources/openembedded-core/meta/recipes-devtools/m4/m4-native_1.4.16.bb, do_populate_sysroot) failed with exit code '1' ERROR: '/ Jun 08 14:15:13 should sstate-control be somewhere outside tmp as well? Jun 08 14:15:36 jic23: no it's correct that sstate-control is under tmp Jun 08 14:17:14 bluelightning: maybe I have some dead caches. I'll clear out tmp. Jun 08 14:17:55 bluelightning: Actually whilst doing that I've found a sstate-cache under build (same dir as where angstrom seems to be putting tmp) Jun 08 14:18:09 so I guess it might have sensible defaults. Jun 08 14:18:15 jic23: right ok so it does have a default Jun 08 14:18:45 jic23: which version of bitbake are you using? Jun 08 14:19:51 bluelightning: 1.12 was using HEAD but dropped back to see if something weird was going on with that. Jun 08 14:20:51 well this is definitely odd behaviour... Jun 08 14:21:54 well it is a gentoo box so could be a weird dependency issue on the host that no one else has hit. Jun 08 14:24:02 clearing tmp didn't help. Fails on first package (quilt-native - I'm doing bitbake meta-toolchain to test) Jun 08 14:24:57 quilt is fine on rerunning. Jun 08 14:27:23 perl-native then fails. at least this is quick to test ;) Jun 08 14:27:44 am having a quick look at the code now Jun 08 14:28:07 bluelightning: thanks. Jun 08 15:03:03 jic23: probably doesn't help you much, but I just downloaded and ran the angstrom setup scripts and ran "bitbake meta-toolchain" and it's not giving me those errors Jun 08 15:03:13 jic23: did you change any configuration? Jun 08 15:06:29 g'day kergoth Jun 08 15:06:48 hm, is there any special trick to getting two git:// URIs to work in SRC_URI? Jun 08 15:07:54 hi pb Jun 08 15:07:57 hi kergoth Jun 08 15:08:21 * kergoth knows bb.fetch2 is an improvement, but is still greatly tempted by rewriting the damn thing Jun 08 15:08:26 hi woglinde Jun 08 15:09:01 every time i look into it, though, i get distracted by wanting to replace the git fetcher with one that uses dulwich (pure python git implementation) or pygit2 (libgit2 bindings) Jun 08 15:09:04 heh Jun 08 15:09:13 heh Jun 08 15:18:10 jeeze, now i remember why i left debian proper. its installer still doesn't support wpa, and someone asked about it on the mailing list and got flamed, practically, acting like its some crazy newfangled thing Jun 08 15:18:21 uh, no, sorry, even a regular end user could buy an AP that uses it Jun 08 15:18:36 ubuntu releases break shit, and debian is a pain in the ass Jun 08 15:18:37 Hrm? Jun 08 15:18:52 * kergoth thinks about an oe-based distro on a desktop Jun 08 15:18:59 I'm pretty sure I've doen installs over WPA wireless. Jun 08 15:19:08 the 6.0.1a installer doesn't support it Jun 08 15:19:16 at least the netinst doesn't, haven't looked at the full dvd Jun 08 15:19:16 Is it not a firmware issue? Jun 08 15:19:18 kergoth fc-15 losing emails? Jun 08 15:19:27 no, it found the wifi just fine Jun 08 15:19:31 and then prompted for a WEP password Jun 08 15:19:34 and that was the only option Jun 08 15:19:44 pathetic, really Jun 08 15:19:46 problem is how often you need to install a debian? Jun 08 15:19:46 URL for the flameage? Jun 08 15:19:54 kergoth: kergothix? ;) Jun 08 15:19:56 that is a bit sad Jun 08 15:20:13 "> I guess you need to get *your* act together then, and get a clean Jun 08 15:20:13 > solution for it, propose that to be merged and maybe get it in for now Jun 08 15:20:13 > or for wheezy. Whining on a mailinglist isn't overly helpful. Jun 08 15:20:14 " Jun 08 15:20:19 http://lists.debian.org/debian-devel/2010/11/msg00129.html Jun 08 15:20:24 course, thats about the beta release, not the final Jun 08 15:20:30 just some quick googling Jun 08 15:21:02 kergoth: That's not a user, that's a bunch of DDs arguing with each other. Jun 08 15:21:14 broonie *g* Jun 08 15:21:29 i didn't say it was a user, i said someone asked about it Jun 08 15:21:39 i said an end user may well want wpa Jun 08 15:21:53 I was more worried about the flaming than the lack of support. Jun 08 15:23:07 ah Jun 08 15:23:54 still, I dread to think what oe on the desktop would be like. Jun 08 15:24:17 agreed. we don't really have enough devs to maintain it Jun 08 15:24:18 presumably if it discovered a WPA access point then you'd get a dozen lines of python stack trace and a message telling you that your hard disk was the wrong size. Jun 08 15:24:28 haha Jun 08 15:24:33 *g* Jun 08 15:24:36 haha Jun 08 15:24:38 he he Jun 08 15:24:43 pb_: No, you'd get a prompt to enter your WPA token with the touchscreen. Jun 08 15:24:53 and the touchscreen would be non-functional Jun 08 15:25:25 On a desktop it'd not be physically present normally which falls within that category. Jun 08 15:25:45 yeah, I've just dealt with broken ts a lot lately, not bitter or anything ;) Jun 08 15:26:06 I wonder what all we'd have to do to enhance it to better support a desktop. I guess the main issue would be the widely varied hardware, would need the full fledged kitchen sink kernel builds Jun 08 15:26:26 and naturally the whole installer situation Jun 08 15:26:29 but yeah, maintaining installers in general is a particularly thankless activity. given that oe struggles with quality even now, I don't think that trying to find devs to work on an installer system is a good plan. Jun 08 15:26:39 * kergoth nods Jun 08 15:26:54 other than that, I can't think offhand of any particular reason that one couldn't run angstrom or something on a desktop if one wanted to. Jun 08 15:27:09 of course, if we got a live cd functioning well enough, you could almost reduce the install to: fdisk, dd disk image Jun 08 15:27:11 * bluelightning smells a challenge Jun 08 15:27:13 or something Jun 08 15:27:25 heh Jun 08 15:27:45 true. plus the regulation 7-hour fight with grub, but you get that with any distribution. Jun 08 15:27:53 heh, indeed Jun 08 15:28:05 oh, I guess we don't really do lvm+raid setup kind of stuff, need that for any credible desktop distro nowadays Jun 08 15:28:08 I'm convinced grub has actually got worse over the years... Jun 08 15:28:09 would probably want the whole update-grub thing to avoid manually screwing with grub.conf Jun 08 15:28:19 is grub2 any better? haven't tried it yet Jun 08 15:28:30 not from a usability standpoint as far as I noticed Jun 08 15:28:36 :\ Jun 08 15:28:43 yeah, grub2 rocks compared to grub1, usability wise Jun 08 15:28:58 if it breaks it's still just as obtuse with its error messages Jun 08 15:29:04 but it does have some extra ways in which you can get it wrong when installing, and debian at least has made full use of that ability in the past Jun 08 15:29:16 heh, i'm not surprised Jun 08 15:29:31 pb why you fight 7 hours with grub? Jun 08 15:29:44 for years i refused to install grub on the mbr, didn't trust something that was split into two pieces to partly reside there Jun 08 15:30:25 heh, I spent about six months netbooting my work desktop off another machine because I couldn't bear to debug the hosed grub install Jun 08 15:30:30 hah Jun 08 15:30:40 i know how that is Jun 08 15:30:43 I've never had any issue with it. Jun 08 15:30:48 just want the damn thing to work so you can move on with your day Jun 08 15:30:53 so sick of debugging crap like that Jun 08 15:31:00 yeah Jun 08 15:31:16 even now, on my current machine, grub seems to have gotten itself into some bad state where it doesn't boot correctly off one of the drives Jun 08 15:31:20 nowadays when my desktop acts up, i just fscking reinstall / and rerun my basic setup steps. keep /home as its own partition and its safe enough Jun 08 15:31:27 need to go into the bios boot menu and select the second stata as the boot disk Jun 08 15:31:28 there is an OE+LVM based distro though...and it works well. Jun 08 15:32:06 I was tired of seeing: Jun 08 15:32:07 LI Jun 08 15:32:11 and that was it... Jun 08 15:32:13 ha Jun 08 15:32:13 haha Jun 08 15:32:15 yes indeed Jun 08 15:32:17 that brings back memories Jun 08 15:32:38 heh Jun 08 15:33:26 yeah, the nice thing about grub, and grub2 in particular, is that it does have a decent enough shell built in that you can generally recover from most of the trivial errors that would have tended to leave you hosed under lilo Jun 08 15:34:04 unfortunately Jun 08 15:34:13 (having fought grub too) Jun 08 15:34:17 and grub2 does know how to boot off raid+lvm, no need for separate /boot anymore Jun 08 15:34:27 so, some evidence of progress, but it's still not exactly smooth sailing Jun 08 15:34:33 bluelightning: no changes to config Jun 08 15:35:05 my ubuntu desktop spontaniously started booting me back to the gdm login screen when trying to login. /me mutters.. Jun 08 15:35:20 jic23: I'm guessing it's some form of host problem though pinning it down would probably involve instrumenting sstate.bbclass further Jun 08 15:35:37 kergoth install windows7 Jun 08 15:36:05 heh, you make kergoth sound like a package manager Jun 08 15:36:19 woglinde: i dual boot that. most recently i've been booting the linux install under vmware player and opening a lot of putty sessions :) Jun 08 15:37:19 kergoth: so any progress on the build-oe-on-windows thing then? ;) Jun 08 15:38:03 You'll need to use a VM at some point. Jun 08 15:38:16 ugh, i can startx but can't login via gdm Jun 08 15:38:19 You can't even store the Linux source on a case insensitve filesystem. Jun 08 15:38:21 * kergoth grumbles Jun 08 15:39:14 broonie: doesn't cygwin provide some workarounds for that? Jun 08 15:39:28 Not to my knowledge. Jun 08 15:40:21 hmm, oh well... not something I'm particularly bothered about Jun 08 15:40:25 * woglinde runs now win2008R2 64bit on 32bit virtualbox Jun 08 15:40:43 NTFS does support case sensitivity, and you used to be able to turn it on in the windows kernel too at one point. Jun 08 15:40:52 I don't know if that still works though, I haven't tried it for ages. Jun 08 15:41:18 atleast ntfs supports symlink somehow Jun 08 15:41:29 <3 junction link extension Jun 08 15:41:37 (the explorer extension, that is) Jun 08 15:45:59 i need to play with the NixOS distro more, it's awfully nifty, but I'm not sure if it's something I'd want to really *use*, or just look at and say thats nifty Jun 08 15:48:03 kergoth: interesting idea for sure Jun 08 15:48:14 (nixos) Jun 08 15:48:21 * kergoth nods Jun 08 15:48:35 after my experiences with gentoo I'm loath to try another source-based distro though Jun 08 15:48:38 I really like how the packages are isolated, and how it can reconstruct the entire system from that + its config db Jun 08 15:48:46 yeah, I can see that Jun 08 15:49:15 I like how you can basically undo any system operation. drop back to a previous configuration via a boot option, or manually Jun 08 15:50:23 yeah Jun 08 15:51:24 I'm not entirely sure I buy into their claim that this is all due to the "purely functional" build language though. Or maybe they're using the term "functional" in a way that I don't quite grasp. Jun 08 15:51:56 they stole it from windows? Jun 08 15:51:56 It'd seem to make things easier. Jun 08 15:52:09 It seems like they're trying to say that the build are largely free of side effects. input, output, thats it Jun 08 15:52:19 which seems reasonable Jun 08 15:53:43 ah, yeah, that makes sense Jun 08 15:54:58 (god, i want that for oe..) Jun 08 15:55:10 kergoth: an oe-based distro on a desktop ia a great idea Jun 08 15:55:39 bluelightning: that's weird. The whole issue seems to have gone away.... Far as I can tell I didn't do anything.... Jun 08 15:55:49 kergoth: I proposed it at OEDEM 2009 without success Jun 08 15:56:11 jic23: ?! that's odd Jun 08 15:56:15 mckoan: heh, well, we're short hand as is, its hard to say hey, lets do even more, and this is a bigger can of worms than most of our current platforms.. Jun 08 15:56:18 mckoan: :) Jun 08 15:56:23 s/hand/handed/ Jun 08 15:57:07 mckoan: are you volunteering to maintain it? ;) Jun 08 15:57:14 hehe Jun 08 15:57:35 bluelightning: I think I did an update using the angstrom scripts. It grabbed something but I didn't notice what. Jun 08 15:57:42 not enough time to maintain it, but contribute yes Jun 08 15:58:42 bluelightning: if I had time to mantain it I'd already done a few years ago ;-) Jun 08 15:58:50 I think it's not a bad idea... I'd love to see more distros using bitbake/oe as a base rather than rolling their own tools Jun 08 15:59:40 basically a sort of openwrt replacement Jun 08 16:00:21 plus fluxbox wm Jun 08 16:00:30 just to begin Jun 08 16:01:01 we probably already have all the pieces, it's just the integration, testing and fixing that would be needed Jun 08 16:01:11 i.e. the "boring" stuff Jun 08 16:01:13 "just" Jun 08 16:01:14 heh Jun 08 16:01:17 heh Jun 08 16:01:17 yes Jun 08 16:01:32 understatement of the year perhaps Jun 08 16:01:33 without graphics it already works Jun 08 16:01:47 right, we already have a bunch of zeros and ones. it's just a case of putting them in the correct order. Jun 08 16:01:49 problema are wm, mouse and keyboard Jun 08 16:02:01 pb_: haha Jun 08 16:05:31 as long as we don't have to build things like LibreOffice :( Jun 08 16:08:18 pb_, are you handling the oe lists these days? Jun 08 16:10:46 not in any official capacity, but I do have administrator access to mailman and I am happy to look at it if you are having a problem Jun 08 16:11:35 I want to make sure we add the new ev members the members list Jun 08 16:11:55 ah, right Jun 08 16:12:04 I can do that if you send me a list of their email addresses Jun 08 16:12:14 the original ballot only had their names Jun 08 16:12:30 hmm, there's a goggle bug. doesn't always close the parsing progress bar when it hits 100% Jun 08 16:14:00 kergoth: we're not still losing some events somewhere are we? Jun 08 16:19:24 didn't think so, but perhaps Jun 08 16:19:25 hmm Jun 08 16:19:43 and of course, now if i close out the 100% progress bar, it quits the build Jun 08 16:19:45 hehe Jun 08 16:19:49 * kergoth lets it float Jun 08 16:21:38 kergoth: seems that was a recent change Jun 08 16:21:53 bah, and it blocks the ability to click on the main window :) Jun 08 16:21:58 * kergoth will have to investigate more closely Jun 08 16:27:15 pb_, I will work on that Jun 08 16:27:21 thanks Jun 08 16:52:38 bluelightning: progress bar doesn't hang open in hob, just goggle, hopefully just a goggle bug rather than an event issue Jun 08 16:52:56 * kergoth crosses fingers and pokes at the code Jun 08 16:53:58 hmm, some performance issues in hob, need a busy cursor after you check something for inclusion while it processes the packages Jun 08 16:54:01 looks nifty though Jun 08 16:55:07 kergoth: incandescant was soliciting feedback on bitbake-devel recently, I'm sure he'd welcome your comments Jun 08 16:55:15 * kergoth will jot down some notes Jun 08 16:55:25 yow, we really need a search box for the package list :) Jun 08 16:55:31 that's one looong list Jun 08 16:56:24 yeah I think that's on his plan Jun 08 16:56:35 or isn't there a box already? Jun 08 16:56:56 don't see one, yet, but yeah, presumably its a pretty obvious thing Jun 08 16:56:59 looks pretty slick though Jun 08 16:57:06 i imagine the end users will love it Jun 08 16:58:00 screenshot here has one at the bottom of the list: https://wiki.pokylinux.org/wiki/BitBake/GUI/PostOneOh Jun 08 16:58:21 it could be this stuff is in a branch Jun 08 16:58:47 hm hm windows 2008R2 server is very nice Jun 08 16:58:49 could also be i was blind, hard to say without reopening it Jun 08 16:58:52 hehe Jun 08 17:01:09 it definitely looks a bit different than that shot in current master. no 'base image' selection for example, just pick and choose the images whose content you want to include Jun 08 17:03:00 kergoth: ah right, his work seems to be in the poky-contrib josh/hob branch Jun 08 17:03:37 ah Jun 08 17:03:56 it's advanced quite a bit since the version in bitbake master Jun 08 17:04:03 nice, will take a look Jun 08 17:04:05 thanks for the pointer Jun 08 17:05:08 I wonder if he plans on extracting configuration variables from the metadata and presenting a UI for them based on the variable types, and if so, if he plans to put that in a common place. I'd imagine that would be useful in multiple UIs Jun 08 17:05:10 * kergoth ponders Jun 08 17:06:38 kergoth: you'd have to ask him :) I think the approach has been to put in channels to list possible values for specific variables (where the UI currently provides a drop-down), but I haven't looked at the code lately Jun 08 17:07:12 I know he has been careful not to break the client/server split Jun 08 17:09:14 ugh, damnit Jun 08 17:09:21 we still have a lot of sstate relocation bugs lurking Jun 08 17:11:25 kergoth: you've just hit a specific one? Jun 08 17:14:39 this particular case was reuse of autoconf-native to build automake-native, it said autoconf wasn't functional. cleaning and killing the autoconf-native sstate fixed it. i suspect some of these may be cases of missing 'vardeps' flags resulting in reuse of sstates that shouldn't be reused, as opposed to relocation issues. both are concerns. unfortunately missing vardeps are particularly hard to nail down, since it doesn't always result in a Jun 08 17:14:39 visible failure Jun 08 17:15:16 * khem dreamnt of writing bitbake in golang Jun 08 17:15:22 khem: i thought about that :) Jun 08 17:15:30 kergoth: its damn good Jun 08 17:15:38 I am experimenting with it at work Jun 08 17:15:50 I need to play with it some more, have made simple apps, but no real projects with it yet Jun 08 17:16:02 and takes advantage of threading at no cost where python really is constipated Jun 08 17:16:10 using multiple CPUs Jun 08 17:16:26 I saw like 5 times speedup in some programs Jun 08 17:16:50 thats on 32core network processors but you get the idea Jun 08 17:17:01 yeah I really like the channels Jun 08 17:17:10 explicit message passing is way way better than memory sharing Jun 08 17:17:17 yep Jun 08 17:17:32 (hence the use of queues in bitbake's ui<->server and parallel parsing) Jun 08 17:18:25 i think there's even lua stuff for go somewhere, could use that to provide scripting Jun 08 17:18:28 heh Jun 08 17:19:05 yes that will be a perfect combo Jun 08 17:19:07 or ocaml Jun 08 17:19:21 we need to use ocaml for something Jun 08 17:19:25 hehe Jun 08 17:19:52 but rewriting bitbake in go hmmm Jun 08 17:20:02 I have to think about it Jun 08 17:20:45 first step for that would probably be to get the recipe parser written in it. need a lexer with states Jun 08 17:21:26 bah, spoke too soon wrt hung up progress bar, just happened with hob in josh/hob branch Jun 08 17:21:27 * kergoth mutters Jun 08 17:22:14 yeah clearly make a frontend Jun 08 17:22:47 kergoth: and may be that could be an excuse to develop .bb2 language as well Jun 08 17:22:56 the problem with any rewrite is they tend to take forever to reach feature parity with the original Jun 08 17:22:58 and get rid of pains from .bb1 Jun 08 17:23:01 * kergoth nods Jun 08 17:23:22 yeah like hollywood sequels Jun 08 17:23:28 hehe Jun 08 17:23:31 the bar is too high Jun 08 17:23:42 shit gcc ICEd Jun 08 17:23:44 hmm Jun 08 17:23:55 I'd be interested in trying it in go even just as an exercise. something to hack on and improve go skill, even if it never replaced the original Jun 08 17:23:59 would be fun Jun 08 17:24:21 hmmm thumb Jun 08 17:24:23 damn it Jun 08 17:24:39 unable to find a register to spill in class 'LO_REGS' Jun 08 17:24:54 hi khem Jun 08 17:24:57 ofcourse if you hide away half the registers Jun 08 17:25:03 woglinde: yo Jun 08 17:25:13 how is weather in dland Jun 08 17:25:44 my friend toured south germany last week and is totally happy Jun 08 17:25:55 thunderstorms Jun 08 17:26:03 oh Jun 08 17:26:06 yeah last week Jun 08 17:26:27 but the last days we had nice "Knalls" Jun 08 17:27:22 I'd really love to see the coding in the OE metadata get shifted out. I think it'd be nice to provide explicit plugins / scripting hooks into bitbake's operations instead. let the recipes/classes stick to declarative. let the plugin set up the tasks, and the classes/recipes just *define* them Jun 08 17:27:27 not sure if others would agree though Jun 08 17:28:50 kergoth: yeah that would be more semantically robust too Jun 08 17:29:10 woglinde: you had reported some issue with scanf Jun 08 17:29:12 the only difficulty there is ensuring that the checksumming can still capture the changes to behavior by the scripts/plugins Jun 08 17:29:15 but i think its doable Jun 08 17:29:15 what was it ? Jun 08 17:29:25 especially if we name/version the addons, and capture that in it Jun 08 17:31:57 khem, how do we make this go awau when we run ldconfig Jun 08 17:32:03 ldconfig: /usr/lib/libstdc++.so.6.0.14-gdb.py is not an ELF file - it has the wrong magic bytes at the start. Jun 08 17:32:09 I am getting beat up over it Jun 08 17:32:18 and me saying ignore it is not working Jun 08 17:32:29 Crofton: hmmm its harmless Jun 08 17:32:42 I know, but it produces user questions Jun 08 17:32:50 * mwester would argue that nothing that creates user concern is harmless. Jun 08 17:33:08 * mwester spent some (unhappy) months in a customer-service job.... Jun 08 17:33:27 exactly Jun 08 17:33:41 * kergoth worked in tech support for 3 years ages ago Jun 08 17:33:42 Crofton: what do customer complain Jun 08 17:33:54 they ask about the message Jun 08 17:34:01 since it is not what they expect to see Jun 08 17:34:04 Jefro, ping Jun 08 17:35:22 Crofton: ok is it packaged in -dev ? Jun 08 17:35:33 if not then it belongs to -dev package Jun 08 17:35:36 not sure Jun 08 17:35:39 so thats first thing we need to do Jun 08 17:35:51 git://git.openembedded.org down? I can get to http://cgit.openembedded.org/cgit.cgi/openembedded/ but can't clone Jun 08 17:36:01 I cant look at it now, in MV and have a telecon and need to add debug to verilog ... Jun 08 17:36:20 I just got "yelled" at over it again since we move it around in the bug tracker Jun 08 17:36:41 khem yes scanf Jun 08 17:36:52 uclibc dont honors m flag Jun 08 17:37:26 for alloc space for char* and store it Jun 08 17:37:36 tharvey: getting that here too Jun 08 17:38:43 woglinde: hmm Jun 08 17:39:02 it dont honors a flag too Jun 08 17:39:12 but a flag shouldnt be used because of some clashes Jun 08 17:39:56 for systemd I used ugly char[] stuff to workaround it Jun 08 17:45:50 khem btw. -> https://bugs.busybox.net/show_bug.cgi?id=3829 Jun 08 18:08:51 hehe.. Jun 08 18:09:06 I'm reading kergoth idea about oe-based desktop Jun 08 18:09:25 it's not a new idea, people have thrown it around on occasion for a long time Jun 08 18:09:28 I remember that there was idea to create self-containing recursive OE distro Jun 08 18:09:35 i.e. distro able to build OE Jun 08 18:09:36 yeah, that'd be nice Jun 08 18:09:40 or image at least Jun 08 18:12:31 that is very nice idea. Jun 08 18:14:15 but this idea is more applicable to oe-core now :) Jun 08 18:14:21 nslu2-linux.org tried to make a master Makefile that would setup everything necessary to build itself...but not a full distro. Jun 08 18:14:46 it should may be qemu or lxc rootfs Jun 08 18:15:09 I'm not sure people will use this distro on real HW.. only for autobuilders may be Jun 08 18:15:43 or as a VM of some sort. Jun 08 18:17:21 in years past it would have been a CD image...but now USB Flash image. Jun 08 18:32:14 I am trying to write a recipe for pyudev - it claims no dependencies and just needs python-ctypes. The recipe I have is very simple and gets most of the way through but blows-up on the native system trying to import ctypes, specifically, trying to import _ctypes. Jun 08 18:32:32 http://pastebin.com/R3gTznxu Jun 08 18:32:45 does anyone have any experience with this? I feel like I am missing something simple. Jun 08 18:34:04 ka6sox-home: git-daemon-run got uninstalled on melo Jun 08 18:34:18 ka6sox-home: xinetd got updated Jun 08 18:34:29 thats like today Jun 08 18:36:10 khem, yup, I did it. Jun 08 18:36:27 solved quite a few problems. Jun 08 18:37:16 hmm KDE 4.7 will have openGL-ES support Jun 08 18:37:39 so probably we can think of putting KDE on mobile divices Jun 08 18:37:46 * khem thinks of meta-kde Jun 08 18:40:06 khem yes you mean kwin Jun 08 18:40:57 03Tom Rini  07master * rca688dd2de 10openembedded.git/recipes/udev/ (udev.inc udev/cache udev/init): Jun 08 18:40:57 udev: Make the dev cache work with /dev/shm Jun 08 18:40:57 At this point in the boot process we may not have a writable /tmp Jun 08 18:40:57 directory. We will have a writable /dev/shm and the data we need Jun 08 18:40:57 is small enough that we use this directory. Jun 08 18:40:57 Signed-off-by: Tom Rini Jun 08 18:42:03 woglinde: yes Jun 08 18:42:35 woglinde: btw. those two patches Jun 08 18:42:59 I think cpusets patch is fine but execvpe needs to be controlled by a knob Jun 08 18:43:16 khem sure I didnt write the test Jun 08 18:43:17 and also run a bloatometer how much size is increased Jun 08 18:44:46 tharvey, can you clone now? Jun 08 19:25:41 ka6sox-home, git clone does work now thx Jun 08 19:37:33 03Richard Purdie  07master * r70232a8abc 10bitbake.git/bin/bitbake-layers: Jun 08 19:37:33 bitbake-layers: Add comment header Jun 08 19:37:33 Signed-off-by: Richard Purdie Jun 08 19:37:47 03Richard Purdie  07master * r4d15aaabf3 10bitbake.git/lib/bb/runqueue.py: Jun 08 19:37:47 runqueue.py: Export further hash information into the task environment Jun 08 19:37:47 Signed-off-by: Richard Purdie Jun 08 19:37:48 03Richard Purdie  07master * re92c694a53 10bitbake.git/lib/bb/data.py: Jun 08 19:37:48 data.py: Ensure vardeps flags get expanded Jun 08 19:37:48 Signed-off-by: Richard Purdie Jun 08 19:37:48 03Richard Purdie  07master * r5b345d482f 10bitbake.git/lib/bb/ (build.py runqueue.py): Jun 08 19:37:49 build/runqueue: Pass quieterrors flag around to supress errors at task execution time Jun 08 19:37:49 Signed-off-by: Richard Purdie Jun 08 19:37:50 03Richard Purdie  07master * rea0ca90dba 10bitbake.git/lib/bb/cooker.py: Jun 08 19:37:50 cooker.py: Drop duplicate parseCommandLine call Jun 08 19:37:51 Signed-off-by: Richard Purdie Jun 08 20:06:45 khem, we now have 0.9.32, fyi. And thanks alot for your dedication! Jun 08 20:08:11 khem, woglinde, i take it that we will have a point release during the next 14 days due to stuff you and carmelo already scheduled. Jun 08 20:09:04 hiho blindvt Jun 08 20:09:23 ehlo Jun 08 20:09:27 means finally .32 Jun 08 20:09:37 without systemd able to run Jun 08 20:09:40 *g* Jun 08 20:11:11 woglinde, yea. but without your systemd stuff (attachment to bugzilla means that i cannot read it 98% of the time since i cannot afford appropriate SSL certs for the bugzilla to read on my phone; it's far better to git send-email it to the list) Jun 08 20:13:36 woglinde, i'll get that fixed once i have a chance to read the patches (which will likely be this friday or next thursday, depending on the phase of the moon and real work workload) Jun 08 20:14:16 hm as I said I didnt write the excevpe testcase Jun 08 20:14:17 sorry Jun 08 20:14:21 wasnt in the mood Jun 08 20:15:57 that delay sucks, i know, but i have no choice otherwise ATM. They pay me to get other stuff done so no chance to do something really useful currently Jun 08 20:16:24 * blindvt shrugs Jun 08 20:17:39 woglinde, if you're in the mood to contribute a regtest, the better. Would actively prevent grief in the future (so would be awesome in my POV, certainly) Jun 08 20:18:47 anyway. i'm digressing -- & Jun 08 20:24:37 blindvt: great !!! Jun 08 20:25:50 blindvt: so we can merge stuff from prelink branch into master now Jun 08 20:26:09 and also the patches for mtune stuff removal Jun 08 20:26:18 I will do the latter soon Jun 08 20:33:17 khem, i'm about to merge future in now, then you can do prelink then ldso-future. Jun 08 20:36:17 khem, i'll start to draft a future mail beginning tomorrow with a vision about a (radically) changed selection of POSIX-2008/POSIX-2004/POSIX-2001 selection via config reflecting dependant stuff and API aiming towards a 1.0 release and correspoding ABI. Jun 08 20:36:50 blindvt: is future good enough to merge ? Jun 08 20:37:29 khem, i will suggest to completely drop SUSv2 legacy nonsense there too (yay!!) Jun 08 20:37:43 khem, i think it is, yes Jun 08 20:38:30 khem, if you could arrange auto-testers to build oe flavors against future that would of course be of great help for sure! Jun 08 20:39:17 khem, but i fear we -- still -- don't have any such buildbots, do we? :( Jun 08 20:39:44 no Jun 08 20:39:51 it's a shame Jun 08 20:39:57 and my concern is that we should not break anything Jun 08 20:40:06 its hard to debug failures Jun 08 20:40:20 so any change on massive scale has to be tested a bit Jun 08 20:40:34 atleast on active architectures Jun 08 20:41:07 blindvt: startssl.com does valid certs that are free @ 1 year expr Jun 08 20:41:19 perhaps i should just go gentoo and benefit on them paying the bill for the cycles we would need for a sane check-bed. Suckage Jun 08 20:42:29 hi Jun 08 20:43:20 blindvt: oe-core/meta-oe can now build serious images using uclibc Jun 08 20:43:32 on arm/mips/ppc/x86/x86_64 Jun 08 20:43:44 some gui images as well Jun 08 20:43:49 and they all boot and run on qemu Jun 08 20:44:04 so it can definitely be used as a playground Jun 08 20:44:33 BlindMan: it can build efl-nodm-image Jun 08 20:44:48 which is fairly extensive Jun 08 20:44:56 hi. I'm, using yocto. Im trayng to compile fuse package.this is in oepenembedded, not in yocto.I've putted the oe git in meta-openembedded.and added to my bblayers.conf. but, wen I try : bitbake fuse. http://pastebin.com/cBKdzfFT Jun 08 20:45:31 egonzalez_ergio you mean you use oe-core layer and meta-oe? Jun 08 20:45:58 egonzalez_ergio: copy the recipes/fuse from oe.dev into meta-openembedded/meta-oe/recipe-extended or somethin Jun 08 20:46:37 * khem will try oneiric tonight to experience KDE 4.7 Jun 08 20:48:35 Tartarus, i (and erik) am willing to pay DNS for a hobby and /me is willing to spend significant parts of my sparse sparetime for everybody else's profit. But i just have no time for the other stuff like taking care of certs or buildbots. I've been taking care of that infrastructure stuff for a decade now, for several projects, but i cannot afford to continue doing that, unfortunately. I have to get my rent payed and if nobody is in a w Jun 08 20:48:36 ay sponsoring that administrative stuff and my efforts then i have to bail our of certain parts. I'm pretty sure you know the dilemma. Jun 08 20:48:53 txs...trying Jun 08 20:58:57 * darknighte chuckles over the re-write of bitbake in go from khem Jun 08 20:59:19 rphillips started looking at the same thing last year. not sure how far he got. Jun 08 21:00:10 blindvt: what kind of builds you need? Jun 08 21:00:18 I can do some Jun 08 21:00:47 my home workstation is idling lot of time Jun 08 21:22:22 Jay7, full set of each several configs for alpha Jun 08 21:22:22 arm Jun 08 21:22:23 avr32 Jun 08 21:22:23 bfin Jun 08 21:22:23 c6x Jun 08 21:22:23 cris Jun 08 21:22:24 e1 Jun 08 21:22:26 frv Jun 08 21:22:28 h8300 Jun 08 21:22:30 hppa Jun 08 21:22:32 i386 Jun 08 21:22:34 i960 Jun 08 21:22:38 ia64 Jun 08 21:22:40 m68k Jun 08 21:22:42 microblaze Jun 08 21:22:44 mips Jun 08 21:22:46 ehm Jun 08 21:22:46 nios Jun 08 21:22:48 nios2 Jun 08 21:22:50 powerpc Jun 08 21:22:52 sh Jun 08 21:22:54 sh64 Jun 08 21:22:56 sparc Jun 08 21:22:58 v850 Jun 08 21:23:00 vax Jun 08 21:23:01 don't make me kick you :P Jun 08 21:23:02 x86_64 Jun 08 21:23:04 xtensa Jun 08 21:23:08 ahem, too many newlines, sorry Jun 08 21:23:19 blindvt: oh :) Jun 08 21:24:37 well.. it's task for small cluster, not for one powered workstation :) Jun 08 21:24:39 * mwester hands blindvt a sack full of commas. Jun 08 21:24:41 Jay7, it's a bit involved, i know. Been there, done that and keeping it up is not exactly fun, at least not my notion of fun Jun 08 21:26:20 blindvt: so are looking for cluster or you have cluster but have no admin staff? Jun 08 21:27:03 Jay7, yea, i used a some nodes to do the matrix but i lost access to clusters recently, so no >1000 nodes where one can sneak in backfillers to do useful stuff in some 16 nodes Jun 08 21:27:57 Jay7, i don't have sensible hardware nor the time to take care of setting up that stuff afresh Jun 08 21:28:32 yeah, that's not exactly fun as you said Jun 08 21:28:56 that's why I'm talking about distributed builds Jun 08 21:29:24 autobuilders is nice thing but not every company/people can have it Jun 08 21:29:26 it's fun if you can sneak it in, but it gets cumbersome if you have to collect slaves and setup the thing for the Nth time Jun 08 21:30:06 but anyway. Testing is for whips either way ;P Jun 08 21:30:10 have to run Jun 08 21:30:12 * blindvt & Jun 08 21:31:28 that is application of oe distro.. Jun 08 21:31:47 build it, place it on node in cluster and control it Jun 08 21:34:07 Tartarus: wrt udev, great, now I can reenable device cache without hacking base-files Jun 08 21:34:36 heh Jun 08 21:36:09 though, standard fstab needs some care Jun 08 21:44:29 Here at least, ramdisks work without complaining about doing stuff before root is rw Jun 08 21:47:54 Tartarus: neat trick with udev, btw! Jun 08 21:48:47 denix: yeah, but from what koen has said it's not nearly as neat as just fixing the rules Jun 08 21:49:36 * tzanger missed this neat trick Jun 08 21:50:44 rebuilding and testing it right now Jun 08 21:52:57 tzanger: In short, OE is currently adding a bunch of rules that seem to do nothing more than slow the system down Jun 08 21:53:21 oe-core has this fix Jun 08 21:54:52 Tartarus: ok, but what's the ramdisk do to fix it, are you using initramfs and rm /etc/udev.d/rules.d/* early on? Jun 08 21:55:05 tzanger: oh, what I just pushed? Jun 08 21:55:17 I just made the cache stuff use /dev/shm which is rw that early not /tmp, which is not Jun 08 21:55:28 not doing anything clever to save the cache Jun 08 21:55:42 Tartarus: I retry now that pb_ knocked out uclibc... why is klibc machine specific? Because is using kernel headers? And then? Jun 08 21:56:46 oh I see Jun 08 21:57:37 Tartarus: when I reordered the initscripts I was wondering nobody else did complain Jun 08 21:58:52 udev scripts were starting too early Jun 08 22:31:05 hi, there is no sdl-config-native Jun 08 22:31:27 that breaks crimsonfields-native_0.4.8.bb Jun 08 22:31:30 in oe.dev Jun 08 22:31:55 ouch it has a do_stage **** ENDING LOGGING AT Thu Jun 09 02:59:56 2011