**** BEGIN LOGGING AT Mon Oct 27 02:59:57 2008 Oct 27 03:31:37 03Mike Westerhof  07org.openembedded.dev * rc91edd60bd 10openembedded.git/conf/distro/include/slugos.inc: SlugOS: slugos.inc - Set preferred version for cairo as well Oct 27 03:31:42 reverting commits due to comment format non-compliance. Now *that's* funny :-) Oct 27 03:32:18 * rwhitby thinks the OE commit format should just be changed to include standard practice elsewhere. Oct 27 07:00:13 after i've finished an OE build, is there a way i can pick just the cross toolchain from it and use it for other things? Oct 27 07:57:23 hi , whats the conf option again for cleaning up tmp during a build ? Oct 27 08:00:25 Good morning Oct 27 08:51:10 morning Oct 27 08:51:28 rob_w: INHERIT += "rm_work" Oct 27 08:58:16 hello Oct 27 08:59:19 bbs Oct 27 09:24:34 hrw|afk, thx Oct 27 10:08:32 florian_: good morning Oct 27 10:15:31 pb__: really? what git status says? Oct 27 10:17:17 otavio: "fatal: cannot do a partial commit during a merge." Oct 27 10:17:43 let me try running git status on the whole tree Oct 27 10:17:54 hi lrg_ Oct 27 10:18:28 pb__: mind to paste the while git status output? Oct 27 10:18:41 morning pb__ Oct 27 10:18:43 otavio: yah, I'm just waiting for git status to complete now Oct 27 10:18:55 this is an nfs folder so it takes a while to run Oct 27 10:19:07 pb__: wow! ok Oct 27 10:19:52 I guess I should move my git tree onto a local disk. Oct 27 10:19:53 ah, here we go Oct 27 10:19:57 hi Oct 27 10:20:18 otavio: there are many hundreds of lines, are you sure you want me to paste it all? Oct 27 10:20:18 pb__: it is much faster working locally Oct 27 10:20:27 pb__: yes, please do Oct 27 10:20:35 okay, let me capture it to a file Oct 27 10:20:37 pb__: but use a pasting service, of course :P Oct 27 10:20:41 heh, sure Oct 27 10:24:01 otavio: http://www.pastebin.ca/1237678 Oct 27 10:25:17 morning all Oct 27 10:27:22 hi thesing Oct 27 10:28:51 pb__: looking at it Oct 27 10:29:03 hi woglinde Oct 27 10:30:03 pb__: http://www.pastebin.ca/1237687 Oct 27 10:30:18 pb__: those are the pending merges, check them and git add them Oct 27 10:31:53 http://blog.haerwu.biz/tmp/anuluj.jpg Oct 27 10:32:41 otavio: ah, right, I see Oct 27 10:32:47 hi hrw Oct 27 10:33:22 otavio: on this occasion I have already switched to a fresh checkout, but I will remember that for next time this happens. Oct 27 10:33:36 pb__: great. Oct 27 10:34:17 pb__: you want git-mergetool probably Oct 27 10:35:26 XorA: probably :-} Oct 27 10:36:09 pb__: fires up your favourite graphical merger Oct 27 10:37:10 hi, all! Oct 27 10:37:39 hi slapin Oct 27 10:37:53 if I need to set additional CFLAGS for glibc, which file should I change? Oct 27 10:38:20 and, also - for commit access using git, what should I do? Oct 27 10:40:38 slapin: the answer to the first question depends on what flags you want to set, and why you need them. Oct 27 10:40:58 for the second, you need to find an existing developer to propose you for commit access. Oct 27 10:41:22 pb__, I have commit access using monotone. Oct 27 10:41:37 oh, right. in that case, just follow the procedure that was mentioned on the mailing list. Oct 27 10:42:23 : Scratchbox has unsatisfying version: automake 1.4-p6 Oct 27 10:42:35 i.e. email your ssh public key to one or more of the überhackers and ask them to add it to the gitosis access list. Oct 27 10:43:32 pb__, thanks, as for glibc, I have some requirements to build glibc using -fno-omit-frame-pointer -mapcs -mno-sched-prolog Oct 27 10:44:29 slapin: in that case, you would probably be best served by setting SELECTED_OPTIMIZATION_glibc = "..." in either your MACHINE.conf, DISTRO.conf or local.conf file Oct 27 10:44:31 pb__, it is not something which is really important, but I just need to comply any way possible, but just setting CFLAGS= before configure somehow doesn't work. Oct 27 10:45:11 pb__, thanks! will look into this Oct 27 10:52:39 03Thomas Kunze  07org.openembedded.dev * r505790b6e8 10openembedded.git/packages/kexecboot/linux-kexecboot-2.6.26/ (akita/defconfig c7x0/defconfig poodle/defconfig): Oct 27 10:52:39 linux-kexecboot: add ext3 to some defconfigs Oct 27 10:52:39 * added for akita, c7x0 and poodle Oct 27 10:52:39 * closes 4688 Oct 27 10:53:01 thesing this means you are done with your diploma-work? Oct 27 10:55:54 No. But I found out that it doesn't help to stop programming. I would find other stuff to distract myself. Now I use my freetime from diploma work for better stuff than TV ;) Oct 27 10:58:25 thesing haha I know this too Oct 27 10:58:34 03Koen Kooi  07org.openembedded.dev * rad9d0faf22 10openembedded.git/packages/xorg-driver/xf86-video-omapfb_git.bb: xf86-video-omapfb: bump SRCREV for stride fix Oct 27 10:58:34 you mean you ran out of TV :-D Oct 27 10:59:23 do someone know why | * ERROR: Cannot satisfy the following dependencies for task-base: Oct 27 10:59:26 | * update-rc.d (= 0.7-r1) * Oct 27 10:59:48 * otavio is fighting with it by one hour already without progress Oct 27 11:01:29 XorA: no. I just don't want to waste so much time by watching all the crime series. Oct 27 11:02:24 thesing: probably dangerous for a hacker anyway, we know what happens when they watch too much crime series :-) Oct 27 11:03:23 do someone knows why it is failing? Oct 27 11:03:59 03Koen Kooi  07org.openembedded.dev * r8f9b3ebf73 10openembedded.git/ (2 files in 2 dirs): xf86-input-tslib: add tslib driver for Xorg xserver Oct 27 11:05:19 otavio: do you, in fact, have that version of update-rc.d built? Oct 27 11:06:16 it does seem a bit odd that there would be a strictly versioned dependency on update-rc.d. any version of that ought to be ok. Oct 27 11:06:32 pb__: I think so; I agree that it looks strange Oct 27 11:06:33 if you check the tmp/deploy/ipk/Packages file, you should be able to find out which package has this dependency Oct 27 11:07:17 Package: update-rc.d Oct 27 11:07:17 Version: 0.7-r1.1 Oct 27 11:07:26 XorA: Not really. Any particular movie I don't know? Oct 27 11:07:30 Looks like it depends on the wrong version Oct 27 11:07:44 thesing: I was thinking of reiser Oct 27 11:08:01 otavio: yeah, it seems. I guess update-rc.d has had its PR bumped (maybe due to DISTRO_PR) and whatever depends on it has not noticed. Oct 27 11:08:21 pb__: however I did a from scratch built Oct 27 11:08:23 otavio: I think the real bug is that the package doing the depending should not specify an exact version. I can't think of any good reason why it would want to do that. Oct 27 11:08:28 otavio: oh, that is even stranger then Oct 27 11:08:33 pb__: so it needs to work Oct 27 11:08:37 otavio: can you tell which package contains this spurious dependency? Oct 27 11:08:53 * otavio is checking it Oct 27 11:09:03 XorA: nearly forgot about him.. Oct 27 11:09:13 * pb__ heads to the hardware lab for a few minutes Oct 27 11:09:14 bbiab Oct 27 11:10:01 otavio@neumann:~/hacking/ossystems/oe$ git grep update-rc.d | grep DEPE | grep 0.7 Oct 27 11:10:02 otavio@neumann:~/hacking/ossystems/oe$ Oct 27 11:10:08 No one does it. Oct 27 11:10:16 * otavio is crazy :P Oct 27 11:11:44 otavio: did you inspect the Packages file? Oct 27 11:12:07 it's very unlikely to be a visible dependency in the source metadata Oct 27 11:12:23 you need to look at the output packages Oct 27 11:12:54 Package: update-rc.d-dev Oct 27 11:12:55 Version: 0.7-r1.1 Oct 27 11:12:55 Depends: update-rc.d (= 0.7-r1), update-rc.d-dev-locale* Oct 27 11:13:02 This is a bug Oct 27 11:13:15 that is indeed a bug for all kinds of reasons. Oct 27 11:13:21 does anything depend on update-rc.d-dev? Oct 27 11:13:29 no Oct 27 11:13:38 * otavio checking if others has this bugged depends Oct 27 11:14:00 firstly, there shouldn't be a -dev package for update-rc.d at all Oct 27 11:14:15 secondly, if there was, it should obviously depend on the correct version of update-rc.d Oct 27 11:14:25 I would guess that the latter is due to an incomplete implementation of the DISTRO_PR thing. Oct 27 11:14:43 might be worth checking with rp for that, I think he wrote the code for the way that is done now Oct 27 11:17:05 all looks fine Oct 27 11:21:55 03Koen Kooi  07org.openembedded.dev * r82240aa226 10openembedded.git/packages/xorg-driver/ (2 files in 2 dirs): xf86-input-tslib: ship hal fdi file to avoid xorg.conf incantations Oct 27 11:25:06 03Philip Blundell  07org.openembedded.dev * rf6a927d799 10openembedded.git/packages/glib-2.0/ (files/atomic-thumb.patch glib-2.0_2.16.4.bb): glib-2.0: add workaround for compilation with -mthumb Oct 27 11:25:08 03Philip Blundell  07org.openembedded.dev * r00de032e49 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 27 11:30:29 so, you mean there are no other packages with that spurious dependency? Oct 27 11:30:45 if that's the case then it would suggest a bug in opkg. Oct 27 11:33:42 pb__: no, it does not Oct 27 11:33:51 pb__: it looks only update-rc.d-dev has it Oct 27 11:34:12 and nothing ought to be causing update-rc.d-dev to be installed, right? Oct 27 11:34:17 in that case, opkg should not have a problem. Oct 27 11:34:18 yes Oct 27 11:34:39 it would be interesting to know whether it works if you manually delete the entry for update-rc.d-dev in Packages Oct 27 11:34:50 I can try Oct 27 11:34:54 if that allows you to build an image successfully, it is surely an opkg bug Oct 27 11:35:41 Well it is a bug Oct 27 11:35:59 since it puts a versioned depends and does not respect the DISTRO_PR Oct 27 11:36:18 yes, that's a separate bug. that bug is in oe somewhere, probably package_ipk.bbclass Oct 27 11:36:25 as I mentioned before, best to talk to rp about that one. Oct 27 11:36:48 there is also a third bug, perhaps in package.bbclass, which is causing oe to output an empty and useless package for update-rc.d-dev Oct 27 11:37:25 update-rc.d isn't a library and doesn't have any development files, so there shouldn't be a -dev package at all for it. Oct 27 11:41:07 yes Oct 27 11:41:12 and the package is empty Oct 27 11:42:46 ~curse buildbot's git fetching Oct 27 11:42:47 May the fleas of a thousand camels infest your most sensitive regions, buildbot's git fetching ! Oct 27 11:43:34 03Koen Kooi  07org.openembedded.dev * r7acee89ac0 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 27 11:43:43 03Koen Kooi  07org.openembedded.dev * r3b353b83ec 10openembedded.git/packages/xorg-xserver/ (6 files): xorg xserver: enable xcalibrate to get touchscreens working a bit better Oct 27 11:43:55 oh! I guess I got it Oct 27 11:44:41 no i did not Oct 27 11:45:40 if I do 'git fetch' then to merge fetched changes into my working copy I should do 'git merge' (no options) or other? Oct 27 11:46:00 hrw: you should do git merge Oct 27 11:47:37 humm Oct 27 11:47:50 mornin' channel Oct 27 11:49:54 http://paste.debian.net/20099/ Oct 27 11:49:57 pb__: http://paste.debian.net/20099/ Oct 27 11:50:36 -dbg and -dev packages are empty but are built anyway Oct 27 11:53:59 thesing: ping Oct 27 11:54:17 ant_work: pong Oct 27 11:55:46 hi, thx for the ext3 commit (I am to lazy to follow commit's guidelines) Oct 27 11:56:47 I have some more commits for kexecboot in my git local branch...mostly referring to dusty status of Zaurus Oct 27 11:57:12 otavio: buildbot use "git fetch git://git.openembedded.net/openembedded org.openembedded.dev" to fetch Oct 27 11:57:58 hrw: what you want to do? Oct 27 11:58:02 thesing: would you think to fix oebug 4545 e.g. Oct 27 11:58:18 otavio: have latest source without hacking buildbot to use pull instead of fetch Oct 27 11:58:41 ant_work: is that the ext3 one? I just commited it. Oct 27 11:58:50 otavio: latest source checked out I mean Oct 27 11:58:54 I've seen, thx Oct 27 11:59:22 (but it is also needed for standard kernel, and we need to decide about kernel in /boot oneday...) Oct 27 12:00:41 hrw: use: git merge origin/org.openembedded.dev Oct 27 12:01:35 otavio: origin/org.openembedded.dev - not something we can merge Oct 27 12:01:45 thats result of 'git merge' Oct 27 12:01:52 hrw: how are the testing with 2.6.27 on c7x0? Would you BUMP 2.6.26 as default for Zaurus clamshells? Oct 27 12:02:04 hrw: git branch -r Oct 27 12:02:10 hrw: what does it output? Oct 27 12:02:16 nothing Oct 27 12:02:40 ant_work: 2.6.26 for clamshells should works. 2.6.27 was not even tested Oct 27 12:02:41 hrw: git branch? Oct 27 12:02:48 otavio: master only Oct 27 12:02:56 git branch -a -> both Oct 27 12:03:20 hrw: where I can look at buildbot source? Oct 27 12:03:47 otavio: buildbot.net Oct 27 12:03:50 hrw: ok, thx for clarify Oct 27 12:04:17 03Phil Blundell  07org.openembedded.dev * racd8a62be2 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 27 12:04:27 03Phil Blundell  07org.openembedded.dev * re96dd6b1e5 10openembedded.git/packages/tremor/tremor_20041119.bb: tremor: suppress thumb mode for this package due to inline asm Oct 27 12:05:20 pb__: conf/bitbake.conf:ALLOW_EMPTY_${PN}-dev = "1" Oct 27 12:05:20 conf/bitbake.conf:ALLOW_EMPTY_${PN}-dbg = "1" Oct 27 12:05:30 pb__: that's why they're built Oct 27 12:06:38 hrw: you're using the git fetcher? Oct 27 12:07:39 otavio: yes Oct 27 12:07:54 hrw: what it does by default? Oct 27 12:08:13 otavio: "git fetch git://git.openembedded.net/openembedded org.openembedded.dev" Oct 27 12:09:02 hrw: what is wird is that you have no remote branch Oct 27 12:09:23 otavio: doh. who added those? Oct 27 12:09:53 otavio: indeed Oct 27 12:12:14 pb__: no idea Oct 27 12:12:22 hrw: I'd add a remote Oct 27 12:12:26 otavio: what does the git history say? Oct 27 12:12:29 hrw: so you could merge fromit Oct 27 12:13:47 pb__: found the bug Oct 27 12:14:26 otavio: I just tested with direct call to 'git pull' instead of buildbot git fetcher Oct 27 12:14:56 pb__: I'm fixing it and will test Oct 27 12:15:38 otavio: what was the bug? Oct 27 12:21:45 re Oct 27 12:28:15 hi woglinde Oct 27 12:28:55 hi ant Oct 27 12:30:40 re Oct 27 12:30:58 re thesing Oct 27 12:34:23 03Marcin Juszkiewicz  07org.openembedded.dev * r70b91bf566 10openembedded.git/conf/machine/qemuarm.conf: qemuarm: do not set PACKAGE_EXTRA_ARCHS as this is done in tune-arm926 Oct 27 12:49:50 03Marcin Juszkiewicz  07org.openembedded.dev * r95b53f12c5 10openembedded.git/packages/gtk+/gtk+.inc: gtk+: added missing dependencies (from Poky) Oct 27 12:52:35 moin Oct 27 12:52:41 he zecke Oct 27 12:52:47 someone wants to review my insane.bbclass change? and the busybox.inc change? Oct 27 12:56:21 thesing: again about kexecboot and its if_down, I suppose you should check for /proc/sys/net and if no networking is defined (like for Zaurus) add the "-x" arg e.g. 242 + const char *kexec_exec_argv[] = { NULL, "-e", "-x", NULL}; Oct 27 12:59:14 ant_work: feel free to prepare a patch. Maybe you want to create a kexecboot_git.bb file and get acces to the kexecboot git repository? So you could push patches to kexecboot directly. Oct 27 12:59:20 btw: actually this would apply to kexec-tools upstream (it's a blind call to if_down) Oct 27 12:59:47 zecke: hail zecke Oct 27 13:05:39 uf.. qemuarm builds are now green Oct 27 13:06:39 ~one hour to moment when I could say does OE.dev/Angstrom is comparable in build stability with Poky Oct 27 13:06:40 that's too long, hrw Oct 27 13:06:59 ~one 1 Oct 27 13:09:46 zecke: http://pastebin.ca/1237774 and http://pastebin.ca/1237775 - can you look? Oct 27 13:13:28 hrw: would you port the latest RP patch to udev_124 (removed static rules) from Poky? Oct 27 13:13:59 ant_work: which rev it is? Oct 27 13:14:04 mom Oct 27 13:15:05 hrw: http://svn.o-hand.com/view/poky?rev=5064&view=rev Oct 27 13:15:42 then udev would be 1:1 in sync with Poky Oct 27 13:15:54 (thanks in advance) Oct 27 13:16:10 what for you need it? Oct 27 13:16:45 just to have the same version tested by more ppl Oct 27 13:17:16 you're free to bump version if you think Oct 27 13:17:28 I think that first initscripts/*/devices should be dropped/fixed I think Oct 27 13:17:55 he..I'm finishing my bootloaders stuff...been soon there Oct 27 13:18:11 s/been/I'll be/ Oct 27 13:19:52 hrw: btw have you seen And#oid released the sources? Oct 27 13:20:21 03Phil Blundell  07org.openembedded.dev * r8c5d1369cd 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 27 13:20:22 03Phil Blundell  07org.openembedded.dev * r468cefb618 10openembedded.git/packages/gnupg/ (gnupg-1.4.2.2/long-long-thumb.patch gnupg_1.4.2.2.bb): gnupg: disable eleet asm in mpi when compiling for thumb. Oct 27 13:20:53 hail zecke again Oct 27 13:21:14 pb__: thanks Oct 27 13:21:15 03Phil Blundell  07org.openembedded.dev * r38ac7b17a4 10openembedded.git/packages/glib-2.0/glib-2.0_2.16.4.bb: glib-2.0: correct over-zealous patch application in last change Oct 27 13:21:28 ant_work: I saw but I am still not impressed by that Javalike platform Oct 27 13:22:12 pb__: thanks for your kind mail Oct 27 13:24:03 zecke: about the reverted commits? not at all. I hope you were not too disheartened by the whole incident. Oct 27 13:24:10 hrw: the only nice thing is the menubar à la OSX Oct 27 13:24:36 zecke: I agree with what mickey|bbl wrote, koen's behaviour seems completely unacceptable in this case. Oct 27 13:24:51 mickeyl: good morning Oct 27 13:24:55 pb__: to be honest I'm. :( Oct 27 13:25:33 zecke: can you look at few small bitbake patches? Oct 27 13:25:56 ant_work: and this can be done by few linux apps already Oct 27 13:26:46 he, I did not look at the sources yet..but even KDE has smthg similar Oct 27 13:27:05 morning pb__ Oct 27 13:27:26 hrw: referring to the whole Java platform, me too I'm against Oct 27 13:27:39 the base question is will we look away again, or will there be sanctions this time. Oct 27 13:27:45 hoi mickeyl Oct 27 13:27:56 judging from the responses until now it looks like the former Oct 27 13:27:59 mickeyl: frankly I think there must be sanctions Oct 27 13:28:00 yo woglinde Oct 27 13:28:01 hi mickeyl Oct 27 13:28:10 heya hrw Oct 27 13:29:22 * hrw -> bitbkae. time to push few things Oct 27 13:30:14 good Oct 27 13:30:22 mickeyl: of course, I have no direct power myself to bring any sanctions, but I am heartily sick of koen's bad attitude in general and I think it is long past time that something was done. Oct 27 13:32:01 pb__: i agree. zecke has made a request to prevent r/w on all but koen/org.oe.dev Oct 27 13:32:31 it is, as you say, unacceptable that he treats the oe project like his own personal fiefdom and continues to behave as the worst kind of petty martinet. Oct 27 13:32:41 mickeyl: ok, very good. I would certainly support that. Oct 27 13:32:49 i support this request this request as well Oct 27 13:32:53 although it's a touch measurement Oct 27 13:32:58 and will open up the danger of a complete fork Oct 27 13:33:31 mickeyl: I suggest 'all but koen/*' Oct 27 13:33:34 we'll see what the other core devs have to say here Oct 27 13:33:39 right Oct 27 13:33:41 hrw: sure Oct 27 13:34:15 I'm just very afraid of that. If we do not manage to realize some sanctions, our "threats" will be void for the rest of time Oct 27 13:34:23 oh dear, i sound like a parent Oct 27 13:34:24 *sigh* Oct 27 13:34:32 bbl, need to pick up laf0rge from the train station Oct 27 13:35:02 mickey|bbl: heh Oct 27 13:35:04 * pb__ lunchtime now Oct 27 13:35:05 bbiab Oct 27 13:35:46 the reaction seems exhagerated...you're stopping a major contributor...doh ! Oct 27 13:36:31 ant_work bad behavior is not our goal Oct 27 13:36:45 Hmm, I wish people would grow up Oct 27 13:36:49 rp jupp Oct 27 13:37:04 hi rp btw. Oct 27 13:37:23 morning all Oct 27 13:37:30 I guess this thing is going to have to actually go to the first core team vote Oct 27 13:38:28 as for a lesson, I'd just revert his reverts in public Oct 27 13:39:33 03hrw * r1113 10bitbake/lib/bb/taskdata.py: taskdata: improved error message when BitBake is called with not existing target Oct 27 13:40:01 how about limiting him to koen/* for a fixed period of time (like 1 month?) Oct 27 13:40:03 03hrw * r1114 10bitbake/lib/bb/cache.py: cache: not existing cache file is not out of date Oct 27 13:40:26 03hrw 07bitbake-1.8 * r1115 10/lib/bb/cache.py: cache: not existing cache file is not out of date Oct 27 13:41:34 03hrw 07bitbake-1.8 * r1116 10/lib/bb/taskdata.py: taskdata: improved error message when BitBake is called with not existing target Oct 27 13:41:53 RP: could you take a look at my sanity.bbclass change in holger/random changes? Oct 27 13:43:04 zecke: The mkdirhier looks good Oct 27 13:43:10 thesing: a good pain would be to oblige koen to finish initramfs-kexecboot for *all* machines in OE Oct 27 13:43:17 ;-] Oct 27 13:43:32 ant_work: I know better ways to waste time ;D Oct 27 13:43:32 RP: okay, I will merge that part Oct 27 13:43:39 bbs Oct 27 13:43:40 anyone can ack the busybox change? Oct 27 13:43:42 zecke: Was there another bit? Oct 27 13:44:01 RP: no, just the missing TMPDIR Oct 27 13:44:32 zecke: I suspect there is some magic in poky elsewhere that solves it but your fix is perfectly fine too Oct 27 13:45:02 zecke: RP: we spotted a bug where TMPDIR contains '-tmp' in its name Oct 27 13:47:16 03Holger Hans Peter Freyther  07org.openembedded.dev * rc3d62194db 10openembedded.git/classes/sanity.bbclass: sanity.bbclass: Work if no TMPDIR was created yet Oct 27 13:47:55 RP: easy things. Moving bitbake to git and to the OE servers? Is there agreement? Oct 27 13:51:41 zecke: all changes looks fine for me Oct 27 13:53:56 03Holger Hans Peter Freyther  07org.openembedded.dev * r6177835736 10openembedded.git/packages/busybox/busybox.inc: Oct 27 13:53:56 busybox.inc : Make busybox honor our LDFLAGS Oct 27 13:53:56 We want --hash-style=gnu to be honored, do not unset LDFLAGS Oct 27 13:53:56 force gcc as linker so -Wl does not look like a bad option. Oct 27 14:00:10 my part of weekly changelog can be easily summarized: "(from Poky)" ;D Oct 27 14:04:19 03Holger Hans Peter Freyther  07org.openembedded.dev * ra9d0016a42 10openembedded.git/classes/insane.bbclass: Oct 27 14:04:19 insane.bbclass: import bb to make python happy Oct 27 14:04:19 With bitbake -b this broke, import bb to make the python runtime Oct 27 14:04:19 happy. Oct 27 14:11:08 * zecke gets insane... I see vincent untz, emanuel bassi clones and more in my class... Oct 27 14:11:36 zecke haha Oct 27 14:11:45 whoever this guys are Oct 27 14:16:31 zecke: did hash-style changes speed things up any for you? Oct 27 14:17:15 cbrake: I'm not there yet, everything that could benefit from it is not yet using it... they ignore ldflags (Qt, busybox) Oct 27 14:17:52 cbrake: I pointed pb_ to the Jakub's initial benchmark, you could try that on ARM Oct 27 14:18:41 zecke: ahh, ok. I did a rebuild with it and have not noticed any diff for things like gtk, x, cario, gstreamer, etc, but they could all be ingnoring it as well Oct 27 14:19:30 cbrake: glib/autoconf based in general honor it. I think it is just hard to measure... Oct 27 14:19:46 zecke: nod Oct 27 14:20:13 cbrake: the next thing is disabling hwcaps (we need to merge the poky patch) Oct 27 14:26:32 hi kgilmer Oct 27 14:26:42 morning woglinde Oct 27 14:39:36 bye Oct 27 14:39:45 bye hrw Oct 27 14:46:39 03Phil Blundell  07org.openembedded.dev * r54a436a3b7 10openembedded.git/packages/bluez/ (bluez-utils/sbc-thumb.patch bluez-utils3.inc): bluez-utils: avoid inline asm when compiling for thumb Oct 27 14:47:28 bbl Oct 27 14:58:05 yo kergoth Oct 27 15:01:22 hey Oct 27 15:05:29 hi kergoth Oct 27 15:29:07 i try to build a -native.bb but the svn-checkout creates a different dir than where or example the patches go. which var should i set to control that? Oct 27 15:41:25 * otavio had energy failure; rebuilding oe Oct 27 15:41:32 hm Oct 27 15:41:48 buy some solarpanles Oct 27 15:45:05 which var controls the dir into the svn checkout is done? Oct 27 15:45:57 re thesing Oct 27 15:46:38 bbl Oct 27 15:49:47 I'm trying to create an autotools-based BB recipe with inherit autotools, but even though I should be giving a "--prefix=${D}/usr" to the configure script, it's still complaining that it can't install to /usr/lib during do_install. Is there somewhere else I'm supposed to be telling it where to put things? Oct 27 15:50:09 hu? Oct 27 15:50:26 you shouldnt user --prefix Oct 27 15:50:44 oe is setting this for you and you should not touch it Oct 27 15:52:21 hmm, I was getting the same error before I put that in there, so it was an attempt at a fix. Oct 27 15:54:04 looking at the makefile, it seems that it should be installing everything relative to prefix, so maybe that's not being set correctly somewhere along the line? Oct 27 16:09:17 good morning all! Oct 27 16:12:06 pb_: http://pastebin.com/f6688a114 Oct 27 16:12:15 pb_: the logic is there but it doesn't work Oct 27 16:12:19 pb_: any idea why? Oct 27 16:12:29 RP: ^ Oct 27 16:13:26 hi hubar Oct 27 16:13:27 okay Oct 27 16:13:32 bye for now Oct 27 16:13:37 I moving home Oct 27 16:19:17 question, can I compile, say, apache2 alone without compiling any of the images under package/images/ directory? Oct 27 16:20:29 hubar is there a DEPENDS variable declariation in the apache2 recipe? Oct 27 16:20:45 Yes. Oct 27 16:20:55 hold on. :) Oct 27 16:21:00 hubar: "bitbake apache2" or "bitbake apache" ? Oct 27 16:21:52 DEPENDS = "apache2-native openssl expat pcre" Oct 27 16:22:38 so unless those are already build (and thier deps and so on) the answer is no. Oct 27 16:22:49 oh, images Oct 27 16:22:53 nevermind Oct 27 16:23:38 hubar sorry I misunderstood your question. like RP says just 'bitbake apache2'. Oct 27 16:26:03 hmm, let me try :) Oct 27 16:34:20 hmm, question, apache2 seems to depend on pcre, but there is no pcre directory under package/image/? Oct 27 16:35:01 hubar: packages/libpcre/ Oct 27 16:35:11 is probably what you're looking for Oct 27 16:42:11 * otavio got it Oct 27 16:42:23 * otavio will propose a patch in ML Oct 27 17:10:18 ah hah! seems the autotools package I'm trying to install doesn't properly support DESTDIR. I added in a bunch of $(DESTDIR)s to the makefile and now it installs cleanly. Oct 27 17:25:10 http://arstechnica.com/journals/apple.ars/2008/10/27/codeweavers-software-free-tomorrow-tuesday-october-28 Oct 27 17:26:49 zecke: hail zecke Oct 27 17:26:54 kergoth: heh, cute Oct 27 17:27:01 yeap Oct 27 17:27:02 I wonder how strictly they enforce the "... to all Americans" bit. Oct 27 17:27:42 * kergoth shrugs, can't hurt to try Oct 27 17:27:46 yah Oct 27 17:29:30 kergoth: sed master... can I use "ADDRESS" with s? and how? e.g. replace foo only if it is in the second line... Oct 27 17:29:37 03Julian_chu  07org.openembedded.dev * rdb87d7205c 10openembedded.git/packages/openmoko-projects/assassin_svn.bb: assassin: Remove curl in DEPENDS of assassin Oct 27 17:29:38 03Julian_chu  07org.openembedded.dev * rf848d38a33 10openembedded.git/ (2 files in 2 dirs): Oct 27 17:29:40 assassin: Change the source URI of assassin For some experimental reason, Assassin is going to move to code google. Oct 27 17:29:43 Bump up the reversion of assassin to 217 Oct 27 17:29:45 03Holger Hans Peter Freyther  07org.openembedded.dev * rab618c09f7 10openembedded.git/ (2 files in 2 dirs): assassin-thumbnail: Add tick's assassin-thumbnail recipe Oct 27 17:29:48 03Julian_chu  07org.openembedded.dev * rbb28555555 10openembedded.git/ (3 files in 3 dirs): Oct 27 17:29:51 openmoko: Change name from Diversity-nav to Om-locations Diversity project decided to change the name of diversity-nav to om-locations and move the repository from svn to git. Oct 27 17:29:54 * Remove diverisyt-nav recipe. Oct 27 17:29:56 * Add om-locations recipe. Oct 27 17:29:58 * Change name in tasks-*.bb Oct 27 17:30:00 03Julian_chu  07org.openembedded.dev * r08e6e32fc3 10openembedded.git/ (2 files in 2 dirs): om-locations: Set om-locations to use 0.2 branch of om-locations git. and bump up the revision to 5cdea8da38561348d4cca6726098afb61d8299eb Oct 27 17:30:12 03Julian_chu  07org.openembedded.dev * rcc4d49defb 10openembedded.git/packages/openmoko-projects/om-locations_git.bb: om-locations: Set om-locations to use 0.2 branch of om-locations git. and bump up the revision to 5cdea8da38561348d4cca6726098afb61d8299eb Oct 27 17:30:16 03Julian_chu  07org.openembedded.dev * r9c4503e642 10openembedded.git/packages/openmoko-projects/openmoko-community-repository.bb: openmoko: Change default community repository set http://downloads.openmoko.org/repository/Multiverse/ as a default repository for community Oct 27 17:30:20 03Julian_chu  07org.openembedded.dev * rab53c55e51 10openembedded.git/packages/openmoko-projects/openmoko-community-repository.bb: openmoko: add Openmoko community repository bbfile. It create a configure file in rootfs for opkg. Oct 27 17:30:26 03Julian_chu  07org.openembedded.dev * r841fc87590 10openembedded.git/ (2 files in 2 dirs): Oct 27 17:30:31 om-locations: Switch back to master branch After Om2008.9 released, switch om-locations back to master branch. Oct 27 17:30:33 * Change SRC_URI Oct 27 17:30:35 * Bump up the revision to 28d56a504a41261c363c085a38f81721cdd6683e Oct 27 17:30:37 * Remove om-locations_svn.bb (it is no longer on svn) Oct 27 17:30:39 03Holger Hans Peter Freyther  07org.openembedded.dev * r5a0c6dce01 10openembedded.git/packages/ (images/openmoko-asu-image.bb tasks/task-openmoko-asu.bb): Oct 27 17:30:44 asu: Create ASU/Om2008.8 task and image (to ease removing Qtopia from it) Create Oct 27 17:30:46 a separate task-openmoko-asu and openmoko-asu-image. The image is installing Oct 27 17:30:48 task-openmoko-asu and task-openmoko-qtopia-x11. It should be fairly easy to Oct 27 17:30:50 remove Qtopia for the people that want that. Oct 27 17:30:54 03Julian_chu  07org.openembedded.dev * r81d6c289e4 10openembedded.git/packages/openmoko-projects/tichy_svn.bb: tichy: Apply Charlie's change Oct 27 17:30:59 03Julian_chu  07org.openembedded.dev * r2a8ed0def5 10openembedded.git/packages/openmoko-projects/assassin_svn.bb: Oct 27 17:31:02 Assassin: Add curl to assassin's depends. Tick did a quick hack to assassin for access networking. It's is a temporary solution. Oct 27 17:31:05 After the real solution has been done, this dependency will Oct 27 17:31:07 be remvoed. Oct 27 17:31:11 03Julian_chu  07org.openembedded.dev * r365804d172 10openembedded.git/ (2 files in 2 dirs): diversity: Add diversity-radar, a GPS location based communicative application [sane-srcrevs] Add version of diversity-radar to 405 Oct 27 17:31:15 03Graeme Gregory  07org.openembedded.dev * r888fc72513 10openembedded.git/packages/openmoko-projects/exposure_svn.bb: exposure_svn.bb : tidy recipe and remove AUTOREV that doesnt belong Oct 27 17:31:20 03Julian_chu  07org.openembedded.dev * re8aa71fc8e 10openembedded.git/packages/openmoko-projects/diversity-radar_svn.bb: diversity: remove PACKAGES = "${PN}" from diversity-radar Oct 27 17:31:27 (25 lines omitted) Oct 27 17:32:02 zecke: 2 s/foo/bar/; replaces foo with bar on line 2 Oct 27 17:32:45 hmm, it didn't like the 2 I can try again :) Oct 27 17:33:18 just tested it here Oct 27 17:33:31 cat /etc/init.d/atd | sed -e'3 s/^.*$//;' -- line 3 is empty in the output Oct 27 17:34:42 zecke: could you take a look on the three patches I sent to ml? Oct 27 17:37:38 otavio: when did you send them? Oct 27 17:37:59 * zecke kmail has difficulties with talking to my university mailserver over gprs... Oct 27 17:39:14 zecke: 20min ago Oct 27 17:39:41 zecke: it improves, from my POV, the DISTRO_PR bahaviour Oct 27 17:40:09 otavio: ah okay, there will be more discussion anyway... Oct 27 17:40:12 * zecke runs to gmane Oct 27 17:40:42 zecke: I found few broken cases Oct 27 17:42:08 oh, I will try to read the mail Oct 27 17:45:24 hmm. Oct 27 17:45:39 I ran into an error while bitbake apache2 Oct 27 17:46:27 | In file included from pcre.c:470: Oct 27 17:46:28 | chartables.c:21:27: error: pcre_internal.h: No such file or directory Oct 27 17:46:29 otavio: did you reply to a mail or top post? Oct 27 17:48:01 zecke: top Oct 27 17:48:08 zecke: let me see if I can find it out Oct 27 17:53:49 Hello all Oct 27 17:54:13 hi gre Oct 27 17:54:21 when I launch a python application I wrote, I get "No module named xml.dom". Which package should provide this module? Oct 27 17:55:16 it supposed to be python? I do have python on the system obviously... Oct 27 17:55:19 hubar: that is building apache2 or a dependency? Oct 27 17:55:22 zecke: sent the patches to http://alioth.debian.org/~otavio/OE/ Oct 27 17:55:33 zecke: it seems that having configurable alignment traps is responsible for about another 500ns syscall overhead on gta01. you might want to consider getting rid of that. Oct 27 17:55:43 zecke: it is from bitbake apache2 Oct 27 17:56:05 I think most of the rest of the discrepancy between gta01 and h3600 can, unfortunately, be explained by arm920t suckiness compared to strongarm Oct 27 17:57:51 er, sorry, that was 50ns, not 500ns Oct 27 17:58:05 heh Oct 27 17:59:21 hubar: that is not what I asked. good thing is you are able to help yourself Oct 27 17:59:37 otavio: 0001*.patch, oops Oct 27 18:00:19 zecke: [13:55] hubar: that is building apache2 or a dependency? Oct 27 18:00:30 hubar: yes? Oct 27 18:00:31 zecke: why? Oct 27 18:01:02 otavio: oops in the way that current RDEPENDS are a bit broken... (omitting the DISTRO_PR) Oct 27 18:01:21 zecke: so? should I apply it? Oct 27 18:01:51 otavio: wait, the issue is there, some questions arise: May we remove DEBPV? I think we may not. Oct 27 18:02:03 zecke: it is from buidling srclib/pcre/chartables.c. Oct 27 18:02:22 zecke: and I haven't found a solution. :( Oct 27 18:02:24 zecke: let me look at it again Oct 27 18:02:49 otavio: may we access base.bbclass methods from bitbake.conf (this will break if someone puts A := PR in another .conf file, which is unlikely though) Oct 27 18:03:10 otavio: base.bbclass will be parsed after conf/bitbake.conf is through Oct 27 18:03:24 zecke: humm Oct 27 18:03:37 otavio: Regarding DEBPV I think more of compability... but in general nice finding and I agree with the change Oct 27 18:03:55 zecke: it doesn't looks to be a problem since it looks to be working fine here Oct 27 18:04:04 * otavio is waiting a full build to finish to be sure of that Oct 27 18:04:26 zecke: and _package_revision is at base.bbclass Oct 27 18:04:31 eek, I even put that into package.bbclass Oct 27 18:04:37 The error is following "Entering directory `/home/wweng/tmp/work/i486-linux/apache2-2.2.3-r5/httpd-2.2.3/srclib/pcre'" Oct 27 18:04:45 zecke: yes, this would be better Oct 27 18:04:55 zecke: I see no reason to leave it at bitbake.conf Oct 27 18:05:02 And the very first file it started to compile, pcre.c. Oct 27 18:05:28 zecke: it is indeed bad design to make it there Oct 27 18:05:43 zecke: also because of the ALLOW_EMPTY_* ugly hacks too Oct 27 18:06:22 hubar: and it does not contain a pcre_internal.h anywhere? Oct 27 18:06:54 zecke: where do I look for ? Oct 27 18:06:54 otavio: The source... bbclass. You kill the last three lines but feel free to push that right now Oct 27 18:07:12 hubar: find /home/wweng/tmp/work/i486-linux/apache2-2.2.3-r5/httpd-2.2.3 -name 'pcre_internal.h' Oct 27 18:07:49 zecke: last three lines? Oct 27 18:08:00 otavio: for 0002.patch I think we should move build_package_revision to base.bbclass first (the alternative would be to write inherit package in packaged-staging which might end up in a desaster) Oct 27 18:08:02 zecke: there's a hook that removes an empty like. Oct 27 18:09:14 [ 14:08:05 ] wweng /home/wweng/tmp/work $ find . -name "pcre_internal.h" Oct 27 18:09:15 ./i486-linux/libpcre-7.6-r3/pcre-7.6/pcre_internal.h Oct 27 18:09:16 zecke: right, I can move it there. Oct 27 18:10:55 zecke: [ 14:08:05 ] wweng /home/wweng/tmp/work $ find . -name "pcre_internal.h" Oct 27 18:10:57 ./i486-linux/libpcre-7.6-r3/pcre-7.6/pcre_internal.h Oct 27 18:11:16 zecke: I'm preparing a branch and will give it for you to take a look in few minutes, ok? Oct 27 18:12:59 and from the compilation process I can see, the pcre-7.6 directory is not included anywhere. Oct 27 18:14:36 hubar: thats ok. the build of pcre should put it somewhere in staging dir. Oct 27 18:14:58 hmm let me "find". Oct 27 18:15:17 thesing_: 'pcre_internal.h' does not sound like header that gets staged Oct 27 18:15:38 zecke: oh, no, it is me that sucks. Oct 27 18:15:48 thesing_: [ 14:15:08 ] wweng /home/wweng/tmp $ find . -name "pcre_internal.h" Oct 27 18:15:50 ./work/i486-linux/libpcre-7.6-r3/pcre-7.6/pcre_internal.h Oct 27 18:16:06 thesing_: the header is not put anywhere. Oct 27 18:16:09 pb__: oh? you can not count nano seconds properly? Oct 27 18:16:59 zecke is right. Maybe apache uses some internal stuff from pcre? Oct 27 18:17:06 zecke: no, it turns out that the "dummy" syscall that lmbench uses for timing is actually quite heavyweight when CONFIG_PREEMPT is enabled. Oct 27 18:17:20 which, of course, it is in openmoko Oct 27 18:17:26 otavio: for this change I would be happy to get RP's opinion Oct 27 18:17:45 pb__: oh, hmm interactivity should benefit from that Oct 27 18:18:52 zecke: yeah, I think preempt is probably good news in general (though I am not absolutely certain even of that) Oct 27 18:18:59 it does mean that my benchmark was a bit erroneous though Oct 27 18:19:08 I'm just re-running with a different syscall that isn't affected by preempt Oct 27 18:19:21 Does OE have a working httpd in the repository? (Doesn't need to be apache2) Oct 27 18:20:12 hi pb__ Oct 27 18:20:37 03Otavio Salvador  07org.openembedded.dev * r7eaf16203a 10openembedded.git/classes/sourcepkg.bbclass: sourcepkg.bbclass: use DISTRO_PR value Oct 27 18:20:37 <_diego__> lighttpd Oct 27 18:20:44 zecke: okay, so, the new figures: standard openmoko kernel is about 486ns for null syscall, my modified one is 388ns. Oct 27 18:21:07 <_diego__> or thttpd (very light, but only static pages and simple cgi) Oct 27 18:21:22 old figures were 980ns and 806ns respectively, so you can see that preempt makes quite a difference Oct 27 18:21:28 pb__: are you using RT ? Oct 27 18:21:44 khem: no, why do you ask? Oct 27 18:22:02 Just was curious looking at your numbers Oct 27 18:22:15 are u measuring the syscall path ? Oct 27 18:22:19 yes Oct 27 18:22:29 ah I see Oct 27 18:22:30 this is on openmoko gta01, so s3c2410 cpu Oct 27 18:22:46 arm920 right Oct 27 18:22:52 right, arm920t Oct 27 18:23:05 266MHz in this case Oct 27 18:24:46 Something broke my do_rootfs when I updated to latest git Oct 27 18:25:45 khem: in which way? Oct 27 18:25:47 * pb__ go home now Oct 27 18:25:49 later all Oct 27 18:26:54 later all Oct 27 18:27:06 Can anyone help with the problem? Oct 27 18:29:16 zecke: https://projetos.ossystems.com.br/git/?p=users/otavio/org.openembedded.dev.git;a=shortlog;h=refs/heads/properly-use-distro-pr Oct 27 18:29:21 zecke: take a look at it Oct 27 18:29:29 zecke: i think I've done what you suggeste Oct 27 18:29:31 zecke: i think I've done what you suggested Oct 27 18:30:27 hmm where does bitbake put all the /usr/include files? Oct 27 18:30:29 zecke: Its dumping a whole lot of messages wait a moment I will paste once I get them again Oct 27 18:30:52 hubar: what do you mean? Oct 27 18:32:36 kergoth: In regular system, apache2 looks into /usr/include directory for header files; where does OE put the header files? I can try to put pcre_internal.h there manually first. Oct 27 18:33:44 humm Oct 27 18:34:05 hubar: $STAGING_DIR/usr/include/ Oct 27 18:35:37 hubar: staging is where files go to be accessible to dependent packages. you shouldnt ever have to mess with that in the normal autotools case though.. Oct 27 18:36:27 kergoth: Well, it is obviously not the case for apache2. :( Oct 27 18:37:41 well, it's likely caused by something else, incorrect depends or something, its highly unlikely those packages do anything incorrect when it comes to staging Oct 27 18:37:44 * kergoth shrugs Oct 27 18:38:40 zecke: | Collected errors: Oct 27 18:38:40 | * ERROR: duplicate src declaration. Skipping: Oct 27 18:39:29 zecke: more detailed here http://tinderbox.openembedded.net/public/logs/1766528.txt Oct 27 18:56:47 what URI should I use for the git repo? The one i pulled from last month doesn't work currently (git://git.openembedded.net/org.openembedded.dev.git) Oct 27 18:56:59 apologies if i missed this somewhere... Oct 27 18:58:26 heh - like right in the welcome message there? Nevermind.... Oct 27 19:02:45 03Stefan Schmidt  07org.openembedded.dev * r048cdb22d6 10openembedded.git/conf/machine/ (include/tune-geodelx.inc xo.conf): tune-geodelx.inc: Move in PACKAGE_EXTRA_ARCH from xo.conf Oct 27 19:02:45 03Stefan Schmidt  07org.openembedded.dev * r0ae63176b8 10openembedded.git/packages/linux/linux-xo_git.bb: linux-xo_git: Fix PV. Thanks Koen. Oct 27 19:02:46 03Stefan Schmidt  07org.openembedded.dev * r0e096ad053 10openembedded.git/conf/machine/xo.conf: xo.conf: Add machine config file for the XO laptop Oct 27 19:02:48 03Stefan Schmidt  07org.openembedded.dev * re03189a888 10openembedded.git/ (3 files in 3 dirs): linux-xo: Add GIT XO kernel with fixed SRCREV Oct 27 19:03:54 zecke: If you have a minute the next days please remove the stefan/xo-support branch. Changes are in oe.dev now Oct 27 19:08:55 stefan_schmidt: if you have push access, you can remove it yourself. git push origin :remotebranch Oct 27 19:09:37 kergoth: zecke told me that they disabled branch deleting Oct 27 19:09:43 ahh Oct 27 19:10:19 i dont know what the reasoning was, but the hook could check and make sure you can only remove your own branches, if that's the issue.. Oct 27 19:10:21 * kergoth shrugs Oct 27 19:10:47 kergoth: No idea, perhaps just a fast safeguard Oct 27 19:11:27 does anyone have problems with gstreamer-plugins-bad currently? Oct 27 19:11:35 shoragan: yes Oct 27 19:11:43 :/ Oct 27 19:11:53 and any solutions? Oct 27 19:11:57 shoragan: On our buildhost and on my workstation Oct 27 19:12:05 No package 'gstreamer-check-0.10' found Oct 27 19:12:18 exactly Oct 27 19:12:43 No idea where it comes from. Oct 27 19:12:53 RP: https://projetos.ossystems.com.br/git/?p=users/otavio/org.openembedded.dev.git;a=shortlog;h=refs/heads/properly-use-distro-pr <= please look at it too Oct 27 19:12:53 No time to dig now. Simpsons time. :) Oct 27 19:13:15 otavio: I was just looking at http://alioth.debian.org/~otavio/OE/ Oct 27 19:13:23 otavio: Are those the same patches? Oct 27 19:15:06 03Richard Purdie  07rpurdie/for-comment-zecke * r1f18b01bd2 10openembedded.git/classes/ (5 files): Oct 27 19:15:06 classes/package*.bbclass: Remove the build_package_revision function Oct 27 19:15:06 Based on experience, this function is going to cause massive code readability Oct 27 19:15:06 issues. We can actually clean up the code more without it so do this instead. Oct 27 19:15:16 03Richard Purdie  07rpurdie/for-comment-zecke * r4b64696380 10openembedded.git/classes/packaged-staging.bbclass: Oct 27 19:15:16 packaged-staging.bbclass: Set the extra staging paths to a sane default Oct 27 19:15:16 If either DISTRO_PR and OELAYOUT_ABI staging packages become superseeded Oct 27 19:15:16 therefore add these to the default extra paths variables although distros Oct 27 19:15:18 are free to override this if they know better. Oct 27 19:15:18 otavio: Can you have a look at the for-comment-zecke branch please? Oct 27 19:15:42 hmm, has anyone done builds in osx? Oct 27 19:15:53 stefan_schmidt, http://gitweb.openembedded.net/?p=openembedded.git&a=commit;h=f4fcadb061442a394cbe50041eb6acd9d4125eba Oct 27 19:15:55 kergoth: I helped someone debug them a bit Oct 27 19:16:00 kergoth: It doesn't work at present Oct 27 19:16:17 * kergoth 's using a macbook as his primary machine nowadays, might have to play with it Oct 27 19:16:22 RP: at OE git? Oct 27 19:16:26 otavio: yes Oct 27 19:16:41 otavio: It should solve the packaged-staging problem Oct 27 19:16:53 RP: did you see my chages? Oct 27 19:17:09 otavio: Were the patches I linked to the same as the one in that tree? Oct 27 19:18:22 otavio: What I really don't like is PKG_PV, its not PV anymore :/ Oct 27 19:18:36 EXTENEDEDPV? FULLPV? Oct 27 19:18:52 RP: EXTENDEDPV could be used Oct 27 19:19:12 otavio: Also, why does PF have to change? Oct 27 19:19:56 I suspect the DISTRO_PR check in sanity.bbclass avoids that? Oct 27 19:20:20 RP: Well... it is much easier to understand if PF and final result matches Oct 27 19:20:32 RP: otherwise it is kind of mess Oct 27 19:20:58 otavio: Its pointless adding all this junk to all the paths when you make tmp entirely useless everytime you change DISTRO_PR Oct 27 19:21:51 RP: yes but it is confusing to me, at least, to see: pkg-1.0-r1 and need to check at ipk pkg_1.0_r1.1 Oct 27 19:21:56 RP: it is confusing Oct 27 19:23:33 otavio: People will get used to that, the packaging process is already quite nasty from the point of view since PN can change Oct 27 19:24:03 RP: right however isn't better to avoid that mess when possible? Oct 27 19:24:06 In fact in a way this makes it clearer that the values can get manipulated... Oct 27 19:24:28 RP: I accept that when it is hard to do, we just ignore it, butin this case it is trivial to fix Oct 27 19:25:07 otavio: It buys you nothing except more complex paths IMO Oct 27 19:25:13 and the paths are bad enough already Oct 27 19:25:42 otavio: Did you look over my patches? Do you agree with them? Oct 27 19:26:10 RP: yes, your patches looks sane Oct 27 19:26:31 RP: let me base mine in your ones Oct 27 19:27:00 otavio: Please rename PKG_PV since that name could mean so many things ;-) Oct 27 19:27:16 RP: right; doing it and will ping you Oct 27 19:28:06 otavio: Thanks. I'm going to rebase these and push them since I can't see them making the situation worse :) Oct 27 19:28:53 03Richard Purdie  07org.openembedded.dev * r6f97a58817 10openembedded.git/ (classes/sanity.bbclass conf/bitbake.conf conf/sanity.conf): (log message trimmed) Oct 27 19:28:53 classes/conf: Handle Distro PR change in sanity.bbclass Oct 27 19:28:53 This patch: Oct 27 19:28:53 * Moves the ABI variables into a new abi_version.conf (like Poky). Oct 27 19:28:53 * Renames SANITY_ABI to OELAYOUT_ABI since it has uses outside Oct 27 19:28:57 sanity.bbclass and sanity.conf. Oct 27 19:28:59 * Adds checks to sanity.bbclass so the user is notified when DISTRO_PR Oct 27 19:29:01 03Richard Purdie  07org.openembedded.dev * re86a1f1f30 10openembedded.git/classes/ (5 files): Oct 27 19:29:01 otavio: done, just base your patches on the main tree Oct 27 19:29:04 classes/package*.bbclass: Remove the build_package_revision function Oct 27 19:29:06 Based on experience, this function is going to cause massive code readability Oct 27 19:29:08 issues. We can actually clean up the code more without it so do this instead. Oct 27 19:29:10 03Richard Purdie  07org.openembedded.dev * r8da0dc3451 10openembedded.git/classes/packaged-staging.bbclass: Oct 27 19:29:13 packaged-staging.bbclass: Set the extra staging paths to a sane default Oct 27 19:29:15 If either DISTRO_PR and OELAYOUT_ABI staging packages become superseeded Oct 27 19:29:16 Hmm, my Intel address... Oct 27 19:29:17 therefore add these to the default extra paths variables although distros Oct 27 19:29:19 are free to override this if they know better. Oct 27 19:29:44 ah well, that wasn't entirely intentional but doesn't matter i guess Oct 27 19:30:01 RP: I didn't know you work at intel Oct 27 19:30:16 otavio: I do now, yes Oct 27 19:30:23 OH was aquired by Intel Oct 27 19:32:07 RP: https://projetos.ossystems.com.br/git/?p=users/otavio/org.openembedded.dev.git;a=shortlog;h=refs/heads/properly-use-distro-pr Oct 27 19:33:04 03Cliff Brake  07org.openembedded.dev * rba6b760d53 10openembedded.git/contrib/weekly-changelog-report.py: weekly-changelog-report: remove Merge commit entries Oct 27 19:33:10 otavio: Why is DISTRO_PR not ?= Oct 27 19:33:39 RP: do you want it to be set in environment? Oct 27 19:34:05 otavio: No but I want the distro to be able to override it Oct 27 19:34:20 I guess it should be a weak assignment at the bottom of the file? Oct 27 19:34:21 RP: but the distro file is included later Oct 27 19:34:28 RP: http://tinderbox.openembedded.net/public/logs/1766528.txt my error with latest OE Oct 27 19:34:44 RP: any ideas what could be wrong Oct 27 19:34:53 otavio: So it should be later in the file instead Oct 27 19:35:03 RP: you happen to recall why we use md5sum rather than python's md5 module? Oct 27 19:35:10 khem: Its 2.2MB large so give my connection a minute Oct 27 19:35:11 i'm thinking just not knowing better was the reason Oct 27 19:35:23 or compatbility issues Oct 27 19:35:24 hmm Oct 27 19:35:24 kergoth: I think we fixed that in poky Oct 27 19:35:27 ah Oct 27 19:35:34 kergoth: When ross started playing with osx ;-) Oct 27 19:35:41 cool :) Oct 27 19:35:47 * kergoth fires up git-svn Oct 27 19:36:08 kergoth: I'm pretty sure we added the code to bitbake so its probably just a case of tweaking OE to use it like poky does Oct 27 19:36:18 RP: check it again Oct 27 19:36:21 We badly need a bitbake release so we can change OE for things like that Oct 27 19:37:04 khem: Looks like your opkg.conf file generation is broken Oct 27 19:37:43 otavio: Drop the PF and you can push it ;-) Oct 27 19:38:29 otavio: and fix the commit comment ;-) Oct 27 19:38:52 RP: hmm thx Oct 27 19:38:57 RP: how is it generated Oct 27 19:39:34 khem: package_ipk.bbclass - generate_ipkg_conf Oct 27 19:40:02 RP: since your recent checkin I seem to get "ERROR: Unable to parse conf/bitbake.conf (Could not include required file conf/abi_version.conf)". Where is this abi_version.conf meant to come from? Oct 27 19:40:05 03Otavio Salvador  07org.openembedded.dev * r24db7b9bb0 10openembedded.git/conf/bitbake.conf: Oct 27 19:40:05 bitbake.conf: use base_build_package_revision to ensure DISTRO_PR is used Oct 27 19:40:05 While on it, the variable DEBPV has also been renamed to EXTENDPV since Oct 27 19:40:05 it makes easier to others to read the code. Oct 27 19:40:20 RP: I have not changed this Oct 27 19:40:32 so I wonder what could be wrong Oct 27 19:40:45 I have a patch for opkg to get it compiled with gcc 4.3 Oct 27 19:40:57 pb_: My rebase seems to have lost it. I'l add it now Oct 27 19:41:44 RP: ok, thanks Oct 27 19:42:14 pb_: pushed Oct 27 19:42:14 03Richard Purdie  07org.openembedded.dev * r51955f706a 10openembedded.git/conf/abi_version.conf: conf/abi_version.conf: Add lost from previous commit Oct 27 19:42:25 03Richard Purdie  07org.openembedded.dev * r5c9fa982d6 10openembedded.git/: Merge branch 'org.openembedded.dev' of ssh://git@git.openembedded.net/openembedded Oct 27 19:43:27 RP: right, done it Oct 27 19:47:28 RP: Hmm it was working ok for me Oct 27 19:47:48 khem: Some change must have been made Oct 27 19:47:49 RP: it started to fail recently Oct 27 19:47:53 245f7d129f3dba351b3c8ef284df69a86261904c Oct 27 19:47:59 but I've not been tracking OE that closely :( Oct 27 19:48:22 zecke: what was the above change about ? Oct 27 19:48:35 khem: Is it failing for all images or just sdk/toolchains? Oct 27 19:48:48 RP: I have tried SDK only so far Oct 27 19:49:08 03Marcin Juszkiewicz  07org.openembedded.dev * r218c19c3ec 10openembedded.git/packages/gstreamer/gst-plugins-base_0.10.17.bb: gst-plugins-base: added missing dependencies Oct 27 19:49:09 actually its failing for console-image too Oct 27 19:49:10 03Marcin Juszkiewicz  07org.openembedded.dev * reaffa38a65 10openembedded.git/packages/gstreamer/gst-plugins-base_0.10.17.bb: gst-plugins-base: enabled XVideo output and gnome-vfs plugin Oct 27 19:49:37 khem: That change was to allow distrubutions to force a rebuild of all packages Oct 27 19:50:38 RP: the change log says WorkInProgress: Mangle the DISTRO_PR into the package revisions Oct 27 19:50:43 so may be it is not complete yet ? Oct 27 19:51:09 khem: Its pretty much complete Oct 27 19:51:16 I just pushed some of the other pieces Oct 27 19:51:28 khem: I don't see why it would cause your problem Oct 27 19:52:38 I am just looking at changes done to this file Oct 27 19:52:59 khem: I'd suspect a different one Oct 27 19:53:02 because same was building just file so I think nothing wrong here from my end I believe Oct 27 19:53:05 (different change) Oct 27 19:56:13 I rebuilt from scratch I wonder if someone else is seeing same Oct 27 19:59:06 someone else also saw it here http://de.pastebin.ca/1222949 Oct 27 20:00:03 khem: Ah, check your PACKAGE_ARCHS variable for duplicates Oct 27 20:00:11 khem: I bet this was one of Koen's changes Oct 27 20:00:40 khem: If this is the case, please write a new check for sanity.bbclass Oct 27 20:01:10 I do not have it in my local.conf Oct 27 20:01:40 khem: no, I didn't expect you too Oct 27 20:01:53 khem: bitbake -e | grep ^PACKAGE_ARCHS Oct 27 20:02:32 $ bitbake -e | grep ^PACKAGE_ARCHS Oct 27 20:02:33 PACKAGE_ARCHS="all any noarch arm armv4 armv4t armv5e armv5te armv4 armv4t armv5te omap5912osk" Oct 27 20:02:44 khem: Right, thats bust Oct 27 20:02:56 khem: which machine? Oct 27 20:03:01 omap5912osk Oct 27 20:04:13 khem: I can push a fix Oct 27 20:04:24 ok I can try it right now :) Oct 27 20:04:37 in next 10 mins then I leave Oct 27 20:05:37 khem: try now... Oct 27 20:05:39 03Richard Purdie  07org.openembedded.dev * rc96d6b49a6 10openembedded.git/conf/machine/omap5912osk.conf: omap5912osk: Remove PACKAGE_EXTRA_ARCHS now its set in the tune include file Oct 27 20:08:34 RP: OK Oct 27 20:08:47 RP: I wonder if other machines also need similar fix Oct 27 20:09:08 yeah, probably Oct 27 20:09:20 hmmmm Oct 27 20:09:30 any machine that defines PACKAGE_EXTRA_ARCHS itself and also includes one of the tune-*.inc files is likely to have that problem. Oct 27 20:09:40 right Oct 27 20:09:46 it's a bit of a shame that package.bbclass doesn't just de-dupe it rather than letting ipkg blow up. Oct 27 20:10:55 it is also, admittedly, a bit of a shame that koen checked in this change without correcting the various machine files that broke as a result. Oct 27 20:11:00 oh well Oct 27 20:11:17 :) Oct 27 20:11:38 I did not follow this change so closely Oct 27 20:11:44 since I just joined, may I ask which change this was? Oct 27 20:13:39 pb_: I'll add something to sanity.bbclass that at least checks for it Oct 27 20:13:55 RP: good idea Oct 27 20:15:40 * kergoth tries adding a curl fetcher Oct 27 20:16:26 03Richard Purdie  07org.openembedded.dev * rad6eb13bb4 10openembedded.git/classes/sanity.bbclass: sanity.bbclass: Add check for duplicate entries in PACKAGE_ARCHS Oct 27 20:16:29 kergoth: Did you find poky's md5 updates? Oct 27 20:20:25 just fixed it manually for now, only like 5 uses of it in the classes, just made them use utils.md5_file Oct 27 20:20:37 i'll look at poky in a bit Oct 27 20:43:14 03Phil Blundell  07org.openembedded.dev * rb8fd11a4bf 10openembedded.git/packages/gstreamer/ (2 files in 2 dirs): gst-plugins-bad: disable inline asm during thumb builds Oct 27 20:43:24 03Phil Blundell  07org.openembedded.dev * r859ecccee8 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 27 20:55:49 anyone happen to know why the targetinfo in get_siteinfo_list is all together, rather than split into arch & os info? Oct 27 21:29:54 grr, I wish all these packages didn't include all this stupid inline assembler stuff for things like umull. Oct 27 21:30:05 the compiler can do that perfectly well on its own anyway. Oct 27 21:30:10 * pb_ stabs mpfr Oct 27 21:31:30 pb_: people like control Oct 27 21:31:50 pb_: companies I worked for use -fno-builtins always Oct 27 21:31:54 yeah, it seems, but that way lies gentoo Oct 27 21:32:31 if only they would spend their time on making the compiler better instead. Oct 27 21:32:45 * khem agrees Oct 27 21:33:32 gplv3 has scared companies Oct 27 21:34:22 actually, on the subject of making compilers better, has there been any progress on fixing this ivopts thing that causes such code pessimisation on arm with 4.3 and later? Oct 27 21:34:47 pb_: I think its a middle end problem right ? Oct 27 21:35:07 yeah Oct 27 21:35:13 let me see if I can find the pr number Oct 27 21:35:36 ok Oct 27 21:35:43 basically though, ivopts is doing some unfortunate transformation which defeats a pile of arm instruction patterns like post-inc addressing. Oct 27 21:36:46 I think it's PR 31849 Oct 27 21:41:39 pb_: this is still lurking around it seems Oct 27 21:41:56 I can check it with 4.4 and see if something fixed it Oct 27 21:42:11 ok, that would be good Oct 27 21:42:41 it's a bit frustrating because, in theory, the new section anchor support in 4.x ought to be getting us better code size yet the opposite seems to be true due to other problems. Oct 27 21:43:54 thats right, I think it does good on ppc Oct 27 21:43:58 I had hoped that gcc 4.2.4 would be a good compromise, since that includes the section anchor thing but doesn't have the new dataflow code. but my initial experiments with that version have not yet been terribly encouraging. Oct 27 21:44:20 yeah, ppc and arm are quite similar architectures in a lot of ways Oct 27 21:44:56 section anchor should be a big win for any function which accesses a significant number of global variables. Oct 27 21:49:53 pb_: its still same on 4.4 :( Oct 27 21:49:58 doh Oct 27 21:50:11 pb_: 4.1 is much better Oct 27 21:50:22 yeah, that sucks Oct 27 21:51:10 you should agitate for your company to do something about it :-} Oct 27 21:51:40 my employer doesnt want me to even look at gplv3 code let alone fix it Oct 27 21:51:46 doh Oct 27 21:52:18 * khem is mentally gplv3 contaminated though Oct 27 21:52:35 khem: due patents? Oct 27 21:53:00 I dont know lawyer might Oct 27 21:54:48 khem: I thought you worked at montavista. surely they can't be that anti-gplv3? Oct 27 21:54:58 not anymore :) Oct 27 21:55:01 oh, heh Oct 27 21:56:09 ah yes, the power of google now brings me your linkedin profile. Oct 27 21:56:10 pb_: did you try 4.2.0 or 4.2.1 ? Oct 27 21:56:26 no, not yet Oct 27 21:56:31 do you think those versions are better? Oct 27 21:56:43 pb_: now a days they dont say searching they say googling :) Oct 27 22:28:58 has anyone seen this message - * ERROR: duplicate src declaration. Skipping: src oe-armv5te file:/OE/angstrom-tmp/deploy/ipk/armv5te Oct 27 22:29:42 I'm getting this in do_rootfs of minimal-image... Oct 27 22:33:37 yes, see the logs, or remove any duplicate PACKAGE_ARCH definitions that you might have. Oct 27 22:34:31 03Phil Blundell  07org.openembedded.dev * r84021b1e23 10openembedded.git/packages/mpfr/ (long-long-thumb.patch mpfr_2.3.1.bb): mpfr: crusade against inline assembler Oct 27 22:37:08 khem: do you still need my help? Oct 27 22:37:30 pb_: thanks, looking into it now Oct 27 22:38:19 pb_: I see one PACKAGE_ARCH = "${MACHINE_ARCH}" in task-boot.bb Oct 27 22:38:30 zecke: hail zecke. I think khem has gone. Oct 27 22:38:36 denix: check your MACHINE.conf Oct 27 22:38:42 pb_: thanks Oct 27 22:39:40 pb_: aha, it has PACKAGE_EXTRA_ARCH there... Oct 27 22:40:06 denix: you probably need to remove that. Oct 27 22:41:48 pb_: yeah, it got moved to tune-* few days ago, which I missed. thanks! Oct 27 22:47:41 pb_: SOS Oct 27 22:48:56 hi ant__ Oct 27 22:49:12 pb_: hey, would you please have a look at this? http://rafb.net/p/cAhtv635.html I was building for 'c7x0' but today decided to risk one 'akita' build Oct 27 22:49:56 hmm grep gives this: http://rafb.net/p/7iSavt70.html Oct 27 22:50:35 ant__: does akita need the crazy sharp external toolchain? Oct 27 22:50:42 I forget how the zaurus stuff works. Oct 27 22:50:42 not imho Oct 27 22:50:59 I see crazy gcc 2.9.5 Oct 27 22:51:15 2.95 even Oct 27 22:51:22 rigt Oct 27 22:51:24 right, too Oct 27 22:51:35 there's no gcc 2.95 in oe that I know of Oct 27 22:52:02 bbiab, cat is making holes in my leather chair again Oct 27 22:52:07 ha ha Oct 27 22:56:22 re Oct 27 22:56:28 wb Oct 27 22:56:34 right, cat is banished to the garage Oct 27 22:56:49 luckily she is easily outwitted Oct 27 22:57:01 he..not like wifey Oct 27 22:57:07 heh Oct 27 22:57:25 true enough. luckily wifey is less destructive. Oct 27 22:57:47 mine is like a zombie evening..kid is sucking all stamina Oct 27 22:58:50 back to OE, I'm now trying to build for 'spitz'...let see Oct 27 23:04:18 pb_: same building for 'spitz' Oct 27 23:04:47 ant__: I don't really know much about the zaurus kernel situation. Oct 27 23:05:04 old man kergoth was the oz buildmaster, he should be able to tell you. Oct 27 23:05:16 I see, I suppose there is a wrong commit somewhere Oct 27 23:05:32 or mickeyl probably knows about that stuff too Oct 27 23:05:53 I don't even really know what a spitz or an akita is. Oct 27 23:06:39 I see...I hope 'multimachine' builds are really safe :-/ Oct 27 23:07:03 spitz was the one with HD Oct 27 23:07:05 (3200) Oct 27 23:07:10 akita the one without Oct 27 23:07:13 (1000) Oct 27 23:07:14 IIRC Oct 27 23:07:16 it's been a while :) Oct 27 23:07:19 heh Oct 27 23:07:37 he he .. they are not yet rusty Oct 27 23:07:44 well, my zaurus knowledge stopped with sl-c750 :-} Oct 27 23:08:08 on the SYSTEMS fair we had quite a number of people asking for the Zaurii we had under glass Oct 27 23:08:38 (now that the 'netbooks' are chique) Oct 27 23:09:16 drat I need to kick gcc tomorrow to build libgcc with gnu hash Oct 27 23:09:16 there is some company that released a "zaurus" (pxa270 with bt and wlan) this month Oct 27 23:09:27 pb_: do you think it makes sense to build libgcc that way? Oct 27 23:09:31 gr, now I have some cryptic c++ build error Oct 27 23:09:39 does anybody know what this is about? Oct 27 23:09:49 | /srv/mill/pb/oe/build-fso-thumb/tmp/work/armv4t-angstrom-linux-gnueabi/gcc-4.1.2-r14/gcc-4.1.2/build.arm-angstrom-linux-gnueabi.arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi/libstdc++-v3/include/arm-angstrom-linux-gnueabi/bits/gthr-default.h:103: error: weakref 'int __gthrw_pthread_mutexattr_destroy(pthread_mutexattr_t*)' must have static linkage Oct 27 23:10:07 zecke: libgcc_s, you mean? I doubt it will make much difference. Oct 27 23:10:18 that library doesn't have many symbols to start with. Oct 27 23:10:26 wouldn't hurt though Oct 27 23:10:42 ant__: I think your problem is that bitbake thinks that is has to build an old 2.4 kernel. Nothing else depends on gcc-2.95 AFAIK Oct 27 23:10:45 libgcc_s yes, i will kick gcc anyway Oct 27 23:10:45 thesing: I see most of the strange things are involving 'collie' Oct 27 23:10:51 http://rafb.net/p/7iSavt70.html Oct 27 23:11:01 did you change something recently ? Oct 27 23:11:02 pb_: is virtual in line 103? Oct 27 23:11:08 Does someone has a recipe to build an xorg based image? Oct 27 23:11:18 Most recipes I found are using kdrive Oct 27 23:11:27 I only found beagleboard one Oct 27 23:11:32 zecke: hard to tell, it is wrapped up in many macros Oct 27 23:11:46 I didn't touch 2.4 kernels. Oct 27 23:12:26 otavio: mythfront uses xorg Oct 27 23:13:42 otavio: I'm out of the loop, was RP able to help you? Oct 27 23:14:10 zecke: yes, we've fixed the issues on OE already. All pushed Oct 27 23:14:31 zecke: he fixed some things in a different way then I did and I fixed bitbake.conf Oct 27 23:14:32 ah cool, time to sleep then Oct 27 23:14:36 zecke: it all looks fine Oct 27 23:14:43 pb_: another question...why does resolvconf depend on bash ? Oct 27 23:14:45 http://tinderbox.openembedded.net/packages/resolvconf/ Oct 27 23:15:05 ant__: probably because it has "#!/bin/bash" in some of its scripts Oct 27 23:15:17 * pb_ go to bed now Oct 27 23:15:18 night all Oct 27 23:15:33 nite Oct 27 23:17:07 ant__: collie builds fine. Oct 27 23:20:46 thesing: spitz almost finished, will retry c7x0 soon Oct 27 23:26:20 thesing: doh... entity 'virtual/arm-angstrom-linux-gnueabi-gcc-2.95' for c7x0 too Oct 27 23:26:35 time for a rebuild-from-scratch (tm) Oct 27 23:27:09 if that doesn't work try adding an ASSUME_PROVIDED Oct 27 23:29:56 thesing: before wiping out, I give you the kexecboot-kernel sizes: Oct 27 23:30:17 ? zImage-kexecboot-2.6.26-r6-akita.bin ?1187480?Oct 27 23:52 Oct 27 23:30:38 ? zImage-kexecboot-2.6.26-r6-spitz.bin ?1124516?Oct 28 00:22? Oct 27 23:30:56 zImage-kexecboot-2.6.26-r6-c7x0.bin ?1197196?Oct 27 02:39? Oct 27 23:31:01 this is small enough, right? Oct 27 23:31:07 perfect ! Oct 27 23:32:09 but we exceed the size-limits for ext3 enabled kernels Oct 27 23:32:37 zImage-2.6.26-r5-c7x0.bin ?1336692?Oct 27 00:21 Oct 27 23:32:55 zImage-2.6.26-r5-akita.bin ?1325716?Oct 27 23:31 Oct 27 23:33:19 ^^ these two need some work Oct 27 23:33:25 spitz fits Oct 27 23:33:26 zImage-2.6.26-r5-spitz.bin ?1262248?Oct 28 00:06? Oct 27 23:33:59 anyway, these kernels are meant to be in /boot Oct 27 23:34:08 and nit in Nand Oct 27 23:34:21 *not* in Nand Oct 27 23:44:40 re Oct 28 00:19:11 hmm, woner if i soiuld add a second class for curl and have the supports method check for the existance of the excutable, or make one htpt/ftp class that uses wget or curl Oct 28 01:06:31 * denix is away: I'm not here **** ENDING LOGGING AT Tue Oct 28 02:59:57 2008