**** BEGIN LOGGING AT Sun Dec 18 02:59:56 2005 Dec 18 03:28:03 <_law_> NOTE: package rxvt-unicode-5.9: completed :-) Dec 18 05:01:48 hey Dec 18 05:09:20 morning Dec 18 05:09:55 mickeyl: hey Dec 18 05:10:15 mickeyl: did you ever play with the pythong logging framework? Dec 18 05:11:21 zecke: no, i hadn't a chance yet. Dec 18 05:12:28 mickeyl: that is sad. Dec 18 05:12:39 mickeyl: my mid term bitbake goal is to a) merge RPs fetcher patch Dec 18 05:12:49 mickeyl: b) improve error reporting capabilities Dec 18 05:13:42 mickeyl: and the second thing I want to achieve using the j4log like is to define regions for debugging Dec 18 05:13:57 zecke: sounds like a good plan. any ETA for those things? Dec 18 05:14:06 mickeyl: NO *cry* Dec 18 05:14:40 mickeyl: I need to finish my paper/summary on static code analysis during christmas Dec 18 05:18:33 morning Dec 18 05:19:19 hey reenoo_, how are things? Dec 18 05:20:37 hi mickeyl. pretty good now (winter break, etc.). how about you? Dec 18 05:21:37 busy trying to writing things down fighting against "Schreibblockade" and lazyness ;) Dec 18 05:23:27 hehe. I know the feeling. except, in my case it's because I'm totally exhausted right now. haven't got much sleep during the last couple of weeks... Dec 18 05:25:20 * reenoo_ should be reading about three books right now... Dec 18 05:35:03 * mickeyl dives into some papers about reflective middlewre Dec 18 05:35:07 middleware, even Dec 18 05:35:09 bbl Dec 18 07:05:39 gnu-config-native-0.1cvs20050701-r4 fails to fetch Dec 18 07:06:00 is it known? Dec 18 07:11:58 RP: ping Dec 18 07:14:18 zecke: pong Dec 18 07:14:50 RP: I would apply your fetcher patch now. It will take some time until I have fetched enough sources though to commit ;) Dec 18 07:15:29 zecke: Fair enough :). Its been running here happily for a while so I'm happy it works but someone else testing it is good Dec 18 07:15:34 thanks :) Dec 18 07:16:01 RP: it looks right I do not expect to have any errors Dec 18 07:22:13 RP: did you try the patch with a 'installed' version of bitbake as well? Dec 18 07:22:32 zecke: No, I didn't try that Dec 18 07:23:02 As it doesn't attempt autoloading, it should work but it is untested Dec 18 07:23:18 that is what I'm thinking as well Dec 18 07:23:23 i will try that now Dec 18 07:24:41 RP: what I want to do as well is to put some more stuff in the Fetcher baseclass Dec 18 07:25:02 RP: I think the handling of CVSDATE="now" could be put there, and PRE mirrors and similiar as well Dec 18 07:25:16 zecke: There are several cleanups that can probably be made - I wanted to splut the code up, then worry about improvements Dec 18 07:25:55 zecke: date handling is very SCM specific - git doesn't understand the notion of dates for example Dec 18 07:26:21 but having a now keyword that meant update any cached data would be good Dec 18 07:27:20 git actually needs that - I keep having to tweak bitbake to change the behaviour of the git fetcher in that regard... Dec 18 07:56:29 RP: okay cvs and wget fetchers are working ;) Dec 18 08:00:23 RP: patch applied, thank you! Dec 18 08:01:43 zecke: thanks :) Dec 18 08:02:02 I'm doing some performance work now Dec 18 08:03:00 I still want to develope the parallel processing ideas I have. Just need the time. Maybe over Xmas I might have some... Dec 18 08:04:10 RP: sounds like tree walking fun ;) Dec 18 08:04:14 RP: perfect xmas match Dec 18 08:05:23 #63 is interesting too. I do not know our DAG implementation too good. But the recursion smells fishy... Dec 18 08:10:22 #63 s interesting. I'd guess there are several areas we could improve performance. Has anyone run a python profiler against bitbake? Dec 18 08:11:00 To do the parallel builds, I'm just going to have bitbake parse and dump the information needed and hand over to a separate program :) Dec 18 08:13:00 zecke: Should I mark 438 as fixed? Dec 18 08:13:17 please do so Dec 18 08:18:51 I suspect the bb.data implementation is the bottleneck in most of the cases Dec 18 08:19:08 which is probably obvious as everything is data based Dec 18 08:19:47 yes, I'd say that is a fair assesment Dec 18 08:32:50 RP: bitbake -d -p world spends 50% of the time in getVarFlag Dec 18 08:33:09 RP: and 30% of this time is spend in _findVar Dec 18 08:39:48 zecke: Cosmicpenguin had a link to a tinderbox-like app that looked pretty nice Dec 18 08:40:16 do you have that link handy? Dec 18 08:40:26 sadly not Dec 18 08:42:12 but it was not build bot? Dec 18 08:45:14 hmm in getVarFlag we do a shallow copy, we could pass the reference again Dec 18 08:45:27 that would be a tiny optimisation only Dec 18 08:49:57 zecke: Looks like we need a more efficient data engine.... Dec 18 08:53:55 new number... 25% of the complete time is spent in finding the variable Dec 18 08:54:33 Do we know what's doing the getVarFlag lookups? Dec 18 08:54:38 Maybe we could optimise that? Dec 18 08:54:58 it calls findVar and copy if it found the var Dec 18 08:55:31 RP: and _findVar iterates over the dicts and uses the a in dict method to look the var up Dec 18 08:55:58 RP: the max depth of the iteration is only 1 Dec 18 08:56:25 I'm not sure if a special search tree could help with the performance Dec 18 08:57:58 Its a difficult one... Dec 18 08:58:58 a O(1) lookup of the var would be nice to have Dec 18 08:59:19 and the var could point to the 'depth' of the cow Dec 18 08:59:41 this would make setVar{Flag} less expensive as well Dec 18 09:00:05 O(1) would be nice Dec 18 09:00:21 it would cost memory though :} Dec 18 09:00:50 such is life Dec 18 09:01:06 I hope to find anything more obvious Dec 18 09:01:32 we can save 5% of the time omitting the copy. This was there for debugin purposes anyway Dec 18 09:02:00 some people operated on the reference... which is a bug Dec 18 09:04:18 * zecke gets food now Dec 18 09:06:10 zecke feeding ... :) :) :) Dec 18 09:06:59 cute: _law1 verlaat de chatruimte. ("Powered by OE: www.openembedded.org") Dec 18 09:07:04 so the gaim patch works :) Dec 18 09:19:55 hey folks Dec 18 09:25:28 hey lkcl__ Dec 18 09:31:47 this is OT, but can anyone tell me how i convert a unicode codepoint (as an integer) into a wchar_t? i thought just casting it would work, but it seems not... Dec 18 11:11:32 hi Dec 18 11:12:50 hey Bernardo Dec 18 11:17:23 hi koen Dec 18 11:17:36 the libgpg postinst error bit me again... :( Dec 18 11:17:57 after blasting my tmp dir twice and restarting from scratch Dec 18 11:29:38 * Bernardo wonders why ipkg is trying to create /usr/lib/ipkg... Dec 18 11:32:03 can anyone help me with http://pastebin.ca/34078 ? Dec 18 11:33:56 Bernardo: ipkg wants to keep track of the packages it installs, which is in /usr/lib/ipkg Dec 18 11:34:13 Bernardo: that's why you shouldn't run OE as root Dec 18 11:34:31 koen: but shouldn't it be using tmp/deploy/usr/lib/ipkg? Dec 18 11:34:35 I don't... :) Dec 18 11:34:57 iirc it will use both Dec 18 11:35:39 no idea if it's a bug or feature :) Dec 18 11:36:30 fwiw, it shouldn't touch anything in tmp/deploy Dec 18 11:36:52 tmp/rootfs ? Dec 18 11:37:06 I keep mixing those up Dec 18 11:45:10 Anyone can help me with this? Dec 18 11:45:12 NOTE: Update cvs://anonymous@cvs.savannah.gnu.org/cvsroot/config;module=config;method=pserver;date=20050701 Dec 18 11:45:15 CVS moved to cvs.savannah.[non]gnu.org Dec 18 11:45:17 Check http://savannah.gnu.org/forum/forum.php?forum_id=4168 for details. Dec 18 11:45:20 Anonymous access over SSH is replaced by the (more efficient) pserver. Dec 18 11:45:22 cvs [update aborted]: end of file from server (consult above messages if any) Dec 18 11:45:25 NOTE: Task failed: Fetch failed: config Dec 18 11:46:34 Mardy: $DL_DIR/cvs/gnu-config probably has a CVSROOT entry pointing to the previous location Dec 18 11:46:50 argh! Dec 18 11:46:55 thanks! :-) Dec 18 12:22:06 lkcl__: do you know the status of the blueangel stuff in OE? Dec 18 12:35:53 heya koen: yeh, it mostly works. Dec 18 12:36:02 http://hands.com/~lkcl/blueangel Dec 18 12:36:08 there's a patch there. Dec 18 12:36:19 you need familiar 083 Dec 18 12:37:03 so the monobored pull command needs to include org.embedded.{dev,dreambox,oz354fam083} Dec 18 12:37:21 plus the patch at my web site, which includes adding a couple of files. Dec 18 12:37:47 hmmm.... i'll update it again in a mo. Dec 18 12:38:03 you still have to do ifup usb0 i don't know why yet. Dec 18 12:38:49 oh. it _is_ up. after the 2nd reboot. Dec 18 12:38:51 errr... :) Dec 18 12:39:03 koen left the channel a few minutes ago. would you mind attaching those patches to oe bugzilla? Dec 18 12:39:07 anyway. we're still booting from SD card, from inside wince. Dec 18 12:39:17 burble? oh - okay :) Dec 18 12:39:20 where's that. Dec 18 12:39:40 bugs.openembedded.org Dec 18 12:40:16 hmmm... why isn't that mentioned on the main openembedded.org web site (just curious...)( Dec 18 12:41:23 sounds like a bug >:) Dec 18 12:43:41 oh yeh.... Dec 18 12:45:16 hello Dec 18 12:46:58 is it possible to download the OE.db.bz2 and use it on an NFS sharepoint? I don't have enough space on my "faster" machine, compared to the one that has the storage space, that I would like to build OE on Dec 18 12:51:11 pattieja_, if you dont have enough space for the db i doubt you will have enough to build on Dec 18 12:51:31 its ~ 4-8gb for a standard image build Dec 18 12:54:15 emte, right Dec 18 12:54:50 on my "fast" machine, I have a 4.5GB filesystem with only 1.3GB free space Dec 18 12:55:13 and there's something about NFS mounting OE.db which makes it unhappy about NFS, iirc. Dec 18 12:55:15 I have another machine that I have shared the harddrive from which has 13GB of free space Dec 18 12:55:28 time to get anopther drive perhaps? Dec 18 12:55:32 another* Dec 18 12:55:43 well, you need to spend $2 on e-bay getting yourself a 10gb hard drive, then, don't you :) Dec 18 12:55:47 what's wrong with NFS, though? Dec 18 12:56:00 no strong posix locking Dec 18 12:56:08 besides being slow? Dec 18 12:56:10 and it's going to be dead slow Dec 18 12:56:20 technically, my laptop (which is my "fast" system, 600MHz) has a 10 GB harddrive, but it's split between Win2K and Linux, right now Dec 18 12:56:39 so it would be better to build on the 400MHz system? Dec 18 12:59:47 pattieja_, you dont know anyone with an old faster system? Dec 18 13:00:13 its pretty easy to get over a 1ghz machine for under $100 Dec 18 13:00:30 and the 2.6.8 nfs over xfs OOPS bug of course Dec 18 13:00:40 * koen should update that kernel sometime Dec 18 13:00:49 compiling on machines that slow, i dont know if NFS would seem slow or not Dec 18 13:03:16 heya koen. i did answer your question - you were off with the fairies for a while though. Dec 18 13:03:54 basically: yes, gpe works on blueangel. Dec 18 13:03:59 yeah, I was balancing the christmas tree by eating the candy Dec 18 13:04:13 after the 2nd reboot, usb0 is up, too. Dec 18 13:04:38 oh dear. koen, _what_ have i told you about pretending to be the xmas tree fairy??? you're too _big_ dude. Dec 18 13:04:45 get down, right this minute! Dec 18 13:04:50 :) Dec 18 13:07:01 *stress* - friggin screen initialisation sequence. Dec 18 13:07:34 ~lart p.g.o Dec 18 13:07:35 * ibot does a little 'renice 20 -u p.g.o' Dec 18 13:08:15 frak! where's familiar 0.8.3 in the bugs.openembedded.org list?? :) Dec 18 13:09:46 probably in handhelds.org/~bugzilla Dec 18 13:10:24 http://handhelds.org/~bugzilla/showdependencytree.cgi?id=1334 and http://bugs.openembedded.org/showdependencytree.cgi?id=350 Dec 18 13:11:12 emte, it seems to working quite nicely, so far Dec 18 13:11:18 no complaints or error messages Dec 18 13:12:05 reenoo_: pgo broke your browser again? Dec 18 13:12:35 koen: it breaks everything that's gecko based here Dec 18 13:12:36 pattieja_, i will be supprised if you can compile an image in under a day Dec 18 13:12:44 but its good to know it works Dec 18 13:12:57 * koen happy with his webcore based browser Dec 18 13:13:06 koen: so, is monotone 0.24 required now? Dec 18 13:13:28 yes Dec 18 13:13:42 the static binaries on venge.net work for me Dec 18 13:14:13 hrw has sid binaries at http://handhelds.org/~hrw/ Dec 18 13:16:07 heh Dec 18 13:16:08 well Dec 18 13:16:14 Hi, I have a problem building an opie-image, see http://pastebin.com/469296 Dec 18 13:16:20 emte, one of the issues is that when I tested my disk drive with hdparm -tT on my laptop, it came out to only 10MBps, which is about what my network connection will do to the NFS server, possibly Dec 18 13:16:23 don't really know what to make out of this.... Dec 18 13:16:43 i have a [slow] amd64 it compiles happily in about 3-4 hours, but that's after i'd downloaded everything already and i still have the ~/sources directory. Dec 18 13:16:47 emte, so I figure it doesn't really matter where the compilation happens in this case, except for the NFS warning on the website Dec 18 13:17:04 koen: I'm still stuck on that pseudo OS for ricers... Dec 18 13:17:14 ah, that sucks Dec 18 13:17:15 3000+ (1.8ghz). very cheap. whole machine was only £350. Dec 18 13:18:59 koen: its days are counted though... Dec 18 13:19:11 reenoo_: the 0.24 dep was an unintended side effect of pulling from scratch to get rid of the broken csets Dec 18 13:19:15 /dev/root 9.4G 8.9G 107M 99% / Dec 18 13:19:28 ouch Dec 18 13:21:07 ~lart ubuntu for not checking for available space in /boot Dec 18 13:21:08 * ibot blames ubuntu for not checking for all the evil in the world for available space in /boot Dec 18 13:21:20 well.. I have etch installed on a new disk already Dec 18 13:21:48 about 60% configured too Dec 18 13:21:59 will probably finish the migration next week Dec 18 13:24:19 what should I set the BBFILES variable to in my local.conf file? Dec 18 13:25:43 and what MACHINE should I choose if I just want to build for x86 architecture? Dec 18 13:25:56 koen: hmm... not sure about the 0.24 upgrade path yet.. export the key first or upgrade to 0.24 first? Dec 18 13:25:59 the last time I tried x86, glibc would not build, I think Dec 18 13:26:46 reenoo_: if you have 0.24 installed you can do 'db migrate' on an old db to make it automagically export your key Dec 18 13:26:56 0.24 finally stopped putting the privkey inside the db Dec 18 13:27:25 koen: and it will magically find the right key when I use the new (clean) db? Dec 18 13:29:40 something like BBFILES = /stuff/sources/*/*.bb ? Dec 18 13:29:41 yes Dec 18 13:29:45 maybe? Dec 18 13:30:04 reenoo_: you can do 'ls ~/.monotone/keys' to check Dec 18 13:30:30 koen: ok, thanks. I'll give that a try now Dec 18 13:41:17 bleh. I'm sick of layouting stuff for veroboard style pcbs Dec 18 13:41:32 * reenoo_ wants a cnc machine or something Dec 18 13:42:47 layouting is fun to do the first time Dec 18 13:42:55 after that it gets tedious and annoying Dec 18 13:43:46 well. for _real_ pcbs I don't mind really Dec 18 13:44:03 schematics -> autorouter -> minor fixes -> done Dec 18 13:44:24 that's pretty painless Dec 18 13:44:34 *if* the autorouter works Dec 18 13:44:45 which wasn't the case for orcad9 Dec 18 13:45:07 which also doesn't have a lossless save options for complex layouts...... Dec 18 13:45:43 the one that comes with eagle works well for this rather simple stuff Dec 18 13:46:52 it's just that you can't tell it to use horizontal and vertical connections in a fixed 2.54mm raster Dec 18 13:48:14 ah, I almost forgot about the weird ass measures Dec 18 13:48:22 'mil' vs 'mm' Dec 18 13:49:17 ~lart g++ for major braindamage Dec 18 13:49:18 * ibot pours hot grits down the front of g++'s pants for major braindamage Dec 18 13:49:21 Hi, I've got a problem with opie-image failing at stage_rootfs, http://pastebin.com/469296 Dec 18 13:49:27 ideas anybody? Dec 18 13:54:39 sounds like pH5 messed up dependencies Dec 18 14:00:51 incinerator: would you mind filing a bug at bugs.openembedded.org for that? Dec 18 14:01:12 can do Dec 18 14:01:21 cheers Dec 18 14:02:29 thanks Dec 18 14:03:08 i just realised I wasn't using the most current svn version of bitbake, could that make a difference? Dec 18 14:03:17 no Dec 18 14:03:20 lkcl__: I responed to your patch in bugzilla Dec 18 14:03:21 I'd rather try to build again just in case.... Dec 18 14:03:22 okies Dec 18 14:03:47 it's a bug in OE Dec 18 14:10:17 renoo_, http://bugs.openembedded.org/show_bug.cgi?id=541, thanks Dec 18 16:33:47 koen, are ya there man? Dec 18 16:37:26 morning Dec 18 16:37:37 heya pb. Dec 18 16:37:59 i think... i remember talking to you, like... 18-20 months ago. Dec 18 16:38:30 hi lkcl Dec 18 16:39:30 heh, yeah. you were the Himalaya guy, right? Dec 18 16:39:39 well, you may be interested to know that the work i did on himalaya wasn't wasted (yes, i was) Dec 18 16:39:56 and was taken up by a few people who then moved on to the "blueangel". Dec 18 16:40:03 they managed to get mobile phone calls working. Dec 18 16:40:34 ah, right, very good Dec 18 16:40:42 yehh :) Dec 18 16:40:48 I've seen some blueangel stuff on the mailing list, but I didn't have any idea what it was Dec 18 16:41:16 and - there's some guy kicking around who has a "universal" - aka the jazjar (imate) which does 3G (WCDMA) Dec 18 16:43:18 cute Dec 18 16:44:10 i want one. darn it, i just bought (or, got for free :) the blueangel. oh well, only another 6 months to wait on my phone contract. Dec 18 16:44:44 heh Dec 18 16:47:52 hey pb_ Dec 18 16:49:27 hi reenoo_ Dec 18 16:57:29 hey phil - we got some patches for gpe-contacts and bluez-dbus. any advice? Dec 18 16:57:57 send 'em to the mailing list(s) and cc: them to me. Dec 18 16:58:21 I'll take a look at them, but it might need to wait until I get back to the UK on Thursday. Dec 18 16:58:32 ack. how much active work is going on on those two programs, you know? Dec 18 16:58:50 cp15 is debating whether to create a separate cvs on xanadux, temporarily. Dec 18 16:59:09 gpe-contacts has a small amount. not sure about bluez; potentially quite a lot. Dec 18 16:59:17 okay. cool. thx. Dec 18 16:59:30 bluez.org is (unsurprisingly) the upstream source for the latter. I guess you could check their cvs tree. Dec 18 17:00:18 oh. right. ta. Dec 18 17:03:36 bbl, breakfast time now Dec 18 17:19:28 well, what do you know. I got the same error that I got the previous time I tried this (over 3 months ago, I think) with glibc compilation: Dec 18 17:19:49 how long does it take for an ipaq to recharge ? Dec 18 17:20:07 | ../ports/sysdeps/unix/sysv/i386/signal.S: Assembler messages: Dec 18 17:20:07 | ../ports/sysdeps/unix/sysv/i386/signal.S:31: Error: open CFI at the end of file; missing .cfi_endproc directive Dec 18 17:20:10 i think mine went in panic , although the power indicator said i had like 30% Dec 18 17:20:38 pattieja_: I don't think anyone has touched anything x86 in the last 6-12 months Dec 18 17:20:57 ouch Dec 18 17:21:31 or rather anything that actually went into the main tree Dec 18 17:21:38 k Dec 18 17:21:49 any thoughts what a CFI is? Dec 18 17:22:42 nope, sorry Dec 18 17:23:09 k. thanks Dec 18 17:24:07 what are you building for anyway? does it have to be i386 compatible or would i586 do? Dec 18 17:29:08 in the latter case you could try resurrecting the epia stuff (conf/machine/epia.conf) Dec 18 17:38:58 reenoo_, it's a Compaq iPAQ IA-1 Home Internet Appliance running a K6-2 233 Dec 18 17:39:10 I guess an i586 kernel would work Dec 18 17:40:05 could try starting with the epia stuff then. that's at least known to have worked at one point Dec 18 17:41:58 k Dec 18 17:42:10 bitbaking task-bootstrap with epia as the MACHINE setting Dec 18 17:42:45 I originally was going for opie-image, but the compilation told me that familiar-0.8.3 is no longer maintained in the org.openembedded.dev tree? Dec 18 17:44:00 well. just remove the offending line in familiar-0.8.3.conf ;) Dec 18 17:44:10 k Dec 18 17:44:40 you will need a different kernel+config btw Dec 18 17:44:47 but at least the toolchain should work Dec 18 17:59:06 ibot seen nelson Dec 18 17:59:14 nelson was last seen on IRC in channel #handhelds.org, 12d 3m 57s ago, saying: 'mreimer: thanks! I've added a mention to it to the qmail.org homepage.'. Dec 18 17:59:42 'night all Dec 19 01:31:29 wow... this 'crossdev' program from gentoo is really good! I just got a complete arm toolchain without having to work for it :) Dec 19 01:54:40 morning all Dec 19 01:59:47 cya in a bit Dec 19 02:32:44 morning koen Dec 19 02:46:46 hey RP Dec 19 02:50:50 morning Dec 19 02:53:58 howdy folks Dec 19 02:54:40 hey Dec 19 02:56:14 RP: irda port is /dev/ttyS1? Dec 19 02:57:37 hey hrw|work and zecke|nap **** ENDING LOGGING AT Mon Dec 19 02:59:58 2005