**** BEGIN LOGGING AT Tue Jan 26 02:59:56 2010 Jan 26 04:02:00 I'm trying to do 'bitbake omap3-desktop-image' but I'm getting 2 failures, mplayer_svn.bb and empathy_2.28.2.bb. I've seen posts about the mplayer problem, but no solution. Is there one? I'm pretty new to oe. Jan 26 06:25:49 penguinece: what errors Jan 26 06:26:42 khem, I was getting an "AV_HAVE_BIGENDIAN undefined" error. I found a patch to the mplayer configure file to fix that. now I'm fighting with some undeclared functions/variables in vo_omapfb.c Jan 26 06:31:16 all of the errors refer to functions/variables in x11_common.h, which is correctly #include'd. O.o Jan 26 06:31:43 look for ifdefs Jan 26 06:32:08 it might be needing different versions of packages than what u have Jan 26 06:35:27 there are #undefs in config.h that undefine things that need to be defined :/ Jan 26 06:35:34 khem, Jan 26 06:44:46 Sounds like you need some configure flags and/or "optional" dependencies. Jan 26 08:04:56 good morning Jan 26 08:12:02 hi Jan 26 08:32:11 Hi, anybody has qt embedded 4.6.1 built successfully in oe? Jan 26 08:35:52 and i got "fatal: cannot pread pack file: No such file or directory" when i do git pull, anybody knows how to solve this? Jan 26 08:46:22 morning all Jan 26 08:46:36 gm gremlin[it] Jan 26 08:57:50 hi, I run last night bitbake meta-toolchain-qte but the last step failed Jan 26 08:58:07 I got: * ERROR: Cannot satisfy the following dependencies for task-qte-toolchain-target: Jan 26 08:58:08 * live555 (= 20090602-r1.5) * xpext (= 1.0-5-r1.5) * Jan 26 08:58:44 how could i fix this? i added the libs in task-qte-toolchain-target.bb, but that doesn't help anything Jan 26 09:04:07 steffenmauch: i got almost same error, oe reports can't find locale-en-gdb package Jan 26 09:04:21 then ERROR: Cannot satisfy the following dependencies for task-qte-toolchain-target: Jan 26 09:04:53 leslielg: had you success in fixing it? Jan 26 09:05:24 i added mplayer and gstreamer to qte-toolchain, because i want to use phonon Jan 26 09:05:43 also added it to x11-image and there everything builds fine Jan 26 09:08:04 steffenmauch: not fixed yet Jan 26 09:08:42 steffenmauch: which qte version are you compiling? Jan 26 09:08:49 4.6 Jan 26 09:08:55 4.6.0? Jan 26 09:09:11 i'm not sure which 4.6 version it is Jan 26 09:09:34 did a git pull yesterday and gave it a try Jan 26 09:10:41 i meet an error using qte 4.6.0, the webkit does not response to enter key on a loaded webpage Jan 26 09:11:36 hmm thats not good, but till now, i didn't need webkit :-) Jan 26 10:00:17 good morning florian Jan 26 10:03:33 good morning Jan 26 10:12:48 looks like python 2.6 will be soon default in Debian Jan 26 10:54:57 hrw, 2.6 will be default in debian some time after 3.0 is released ;) Jan 26 11:21:57 heh, predicting a debian realease is only marginally easier than winning the lottery Jan 26 11:25:53 * XorA wishes he could get tin to use proper email address Jan 26 11:26:05 In this case the problem is that the Python maintainer is an idiot. Jan 26 11:26:23 broonie: you have a macro for that sentence :-) Jan 26 11:27:23 What particularly annoys people is that he's already deployed 2.6 in Ubuntu. Jan 26 11:28:16 morning all Jan 26 11:29:06 broonie: in ugly way Jan 26 11:32:20 hi RP Jan 26 12:44:12 logistics question.. I have package archive for multimachine setup. one machine has to use module-init-tools-3.2 so I have "PREFERRED_VERSION_module-init-tools_local = 3.2.2" in that machine's .conf. but if the package for other machines has built 3.4 then it uses that when building the image ... how can I stop that from happening? Jan 26 12:44:28 why you need older? Jan 26 12:44:38 because it uses an older Xen kernel Jan 26 12:44:49 and it breaks with the newer one Jan 26 12:45:54 maybe I can make a depencency on module-init-tools <= 3.2.2 in the kernel recipe? Jan 26 12:46:05 or something like that.. Jan 26 12:47:14 chromium depends on libnspr (/usr/lib/nspr) and libnss (/usr/lib/nss). Should I add these paths to ld.so.conf (in postinst of the libs or sth) or the chromium launcher script? Jan 26 12:47:27 or should I link it with -rpath so the whole path is hardcoded? Jan 26 12:52:04 th1: OE do not have versioned dependencies Jan 26 13:00:35 I hate boost Jan 26 13:00:43 the boost-dev package is broken Jan 26 13:01:12 Crofton|work: missing files? Jan 26 13:03:47 hang on Jan 26 13:03:50 no Jan 26 13:03:57 you can't include it in an image Jan 26 13:04:34 http://pastebin.ca/1766465 Jan 26 13:13:23 hrw, so I guess I have to create another package name and have it PROVIDE module-init-tools Jan 26 13:14:32 Crofton|work: but that is "boost". So boost-dev is depending on boost? Jan 26 13:15:14 apparently Jan 26 13:15:25 I am trying to sort out what is going on the in the packaging Jan 26 13:17:04 its normal for -dev to depend on ${PN} Jan 26 13:17:17 XorA: yes Jan 26 13:17:26 * XorA fails to see the problem Jan 26 13:17:45 RP: XorA: no problem. Jan 26 13:18:00 one just loses if the main package is empty? :) Jan 26 13:18:06 right Jan 26 13:18:12 zecke: yes, that is a problem :/ Jan 26 13:18:13 ah so ALLOW_EMPTY needs to be set Jan 26 13:18:20 for ${PN} Jan 26 13:18:23 boost is packaging stuff into per-library packages Jan 26 13:18:29 linux-libc-headers is such an example Jan 26 13:18:36 and nothing left for boost_ Jan 26 13:18:56 ALLOW_EMPTY is set Jan 26 13:18:59 linux-libc-headers is a bit different, since it's just headers. Jan 26 13:19:07 XorA: I think one should depend on the libboost* libraries Jan 26 13:19:16 pb_: right Jan 26 13:19:19 in the boost case, boost-dev ought to either be depending on all the subpackages where the libs are moved, or it ought to be split into individual -dev packages for each library subpackage. Jan 26 13:19:31 * pb_ stabs nickserv in the eye one more time Jan 26 13:19:39 or it should vanish from the planet Jan 26 13:19:41 stupid freenode Jan 26 13:19:56 split -dev and -dbg packages are an interesting problem :/ Jan 26 13:20:25 yeah, I think the easiest (and most sensible) solution is probably just to make boost-dev depend on all the libraries. Jan 26 13:20:26 RP: they are the problems that you live for :-) Jan 26 13:20:28 * Crofton|work stabs boost on general principles Jan 26 13:20:53 never mind someone converted it to INC_PR, poorly and I had to fix that Jan 26 13:21:02 XorA: No, they really are not. I could live quite happily having never considered them ;-) Jan 26 13:21:09 particularly for sdk type deployments, it doesn't seem all that likely that you would want to install only some small subset of the libs. Jan 26 13:21:26 pb_: agreed Jan 26 13:22:19 right Jan 26 13:22:20 * pb_ goes to start work on his tax return Jan 26 13:22:32 january 31 approacheth :-( Jan 26 13:22:34 pb_: ah yeah must remeber to pay that this week Jan 26 13:22:46 * RP has just been doing that dance Jan 26 13:23:03 I have a time to 1st april ;d Jan 26 13:23:25 grr Jan 26 13:23:28 and this year my accountant will do tax papers for me Jan 26 13:23:43 * RP is having to juggle finances a lot atm due to the house purchase Jan 26 13:23:46 my SA is pretty easy luckilly Jan 26 13:23:51 would one of you read boost-36.inc and tell me what to change ? Jan 26 13:23:57 I iwll do all your taxes Jan 26 13:23:58 RP: moving out from parents? Jan 26 13:24:22 hrw: Yes, once the house is habitable Jan 26 13:24:49 * RP is close to having a functional bathroom which will be a good start Jan 26 13:25:54 ~curse ralink usb wifi Jan 26 13:25:57 May the fleas of a thousand camels infest your most sensitive regions, ralink usb wifi ! Jan 26 13:27:42 hm, not a good start, can't find my P60 or my P11D. Jan 26 13:27:51 * pb_ contemplates tidying his desk Jan 26 13:28:55 RP: ah, very good. we are just about to buy a new house, which it seems will involve living in a caravan for most of the summer while it is rebuilt. doh. Jan 26 13:30:36 pb_: Its even more fun if you do the rebuilding mostly yourself :/ Jan 26 13:31:08 heh Jan 26 13:31:22 There is nothing left of the original bathroom, no pipes, no plaster and no floor Jan 26 13:31:33 I don't plan to do the heavy building myself: my bricklaying skills are rubbish. Jan 26 13:31:54 pb_: Nothing that practise can't fix. I draw the line at plastering Jan 26 13:32:08 RP: The nothing left bit is the easy bit, it's the putting new versions in that's more interesting :) Jan 26 13:32:26 ah, I am getting quite good at plastering now. Jan 26 13:32:47 broonie: You'd be surprised how long turning it into nothing took. You have to do something with all the rubble... Jan 26 13:33:05 git revoke rubble Jan 26 13:33:21 RP: heh, we will have even less left of our original bathroom than that: that part of the building is going to be completely demolished and rebuilt from the ground up. Jan 26 13:33:23 The traditional approach is to leave it in a pile in the garden or something. Jan 26 13:33:24 pb_: Given I was painting the walls, I wasn't prepared to risk it ;-) Jan 26 13:33:59 * XorA has the misfortune of a house to sell :-( Jan 26 13:34:10 broonie: The entire back yard was covered in bricks piled 15 high, piles of earth or deep holes... Jan 26 13:35:43 XorA: selling seems to be the easy bit, as far as I can tell. Jan 26 13:35:56 pb_: not here, has been on the market a year now Jan 26 13:36:05 XorA: ouch :/ Jan 26 13:36:08 * XorA really could do with the money Jan 26 13:36:11 doh, that sucks Jan 26 13:37:41 so anyone wants to buy a 3 bedroom house just outside edin :-) Jan 26 13:37:50 heh Jan 26 13:38:14 Isn't it an option to rent it? Jan 26 13:38:35 florian: not really rental in that area is zero, and my ex wife is still there :-( Jan 26 13:39:25 XorA: oh... this sounds extremely unfortunate Jan 26 13:41:33 XorA: I can see the complications there :/ Jan 26 13:42:24 or if anyone knows a cheap hitman :-) Jan 26 13:43:00 * florian suggests trying to sell house and ex-wife together ;) Jan 26 13:46:17 XorA: that asexual one wife? Jan 26 13:46:28 hrw: that was Liam Jan 26 13:46:33 ah, right Jan 26 13:46:39 * hrw -> shop Jan 26 14:01:50 heh Jan 26 14:02:04 ALLOW_EMPTY was set in a differnet recipe ... Jan 26 14:50:57 morning Jan 26 14:52:27 kergoth: gm Jan 26 14:54:03 can we -R again now? Jan 26 14:54:06 +R sucks a bit Jan 26 15:41:57 yo ulitu1 Jan 26 15:42:06 you're coming to FOSDME? Jan 26 15:42:09 FOSDEM even Jan 26 15:46:08 kergoth: In answer to your question yesterday, 1.10 will be based off master. Master now has the xmlrpc stuff disabled Jan 26 15:46:34 kergoth: and as for passing cooker to the metadata, please no. Not without some kind of public api Jan 26 15:46:43 * kergoth nods Jan 26 15:46:52 okay, will give some thought to api possibilities Jan 26 15:47:16 kergoth: Those structures are all very volatile Jan 26 15:47:31 nod, makes sense Jan 26 15:48:17 * RP won't make FOSDEM :( Jan 26 15:48:43 seems FOSDEM will be quite affair this year Jan 26 15:49:19 RP: doh Jan 26 15:49:25 that's sad Jan 26 15:49:36 :( Jan 26 15:49:39 ya, after some years of degration it might be really good this year Jan 26 15:49:54 no FOSDEM for me either Jan 26 15:49:56 or lrg Jan 26 15:50:05 Hi Jan 26 15:50:16 I just upgraded to latest oe.dev Jan 26 15:50:18 or me Jan 26 15:50:37 oh well Jan 26 15:50:38 which broke configuration of libice-native Jan 26 15:50:46 i'll drink my beer with pb_ then :) Jan 26 15:50:51 if they hadnt moved it :-( Jan 26 15:50:54 checking if xorg-macros used to generate configure is at least 1.2... configure: error: configure built with too old of a version of xorg-macros.m4 - requires version 1.1.0 or newer Jan 26 15:51:27 google said, this problem has been fixed in 10/2009 Jan 26 15:52:58 ulitu1: when can i order your phone? :) Jan 26 15:53:03 :-) Jan 26 15:53:09 how many do you need? Jan 26 15:53:28 as a poor freelance developer... only 1 Jan 26 15:53:37 to port FSO to it Jan 26 15:53:41 and spread the word... Jan 26 15:53:41 that's a little problematic ;-) Jan 26 15:53:45 heh Jan 26 15:53:51 if you want 5000, you can order right now Jan 26 15:53:52 still looking for mass production partner? Jan 26 15:53:57 damn, that#s a bit too much Jan 26 15:54:19 where are those developer samples when you need 'em... Jan 26 15:54:22 I'll keep you in mind though Jan 26 15:54:53 what phone? Jan 26 15:54:59 www.road.de Jan 26 15:55:27 XorA: do you want 5000 pcs ;-) Jan 26 15:55:41 mickey|cafe: I will be there so one more beer please Jan 26 15:56:00 ulitu1: I dont have that many ears :-) Jan 26 15:56:06 lol Jan 26 15:56:30 doesnt happen to be omap based? Jan 26 15:56:34 hrw: excellent! Jan 26 15:56:39 ulitu1: S101 was not released yet? Jan 26 15:56:51 hrw: no, sorry Jan 26 15:56:57 it's pxa270 based Jan 26 15:57:07 eeeuuuuuggggghhhhh! Jan 26 15:57:15 ulitu1: I thought that you did. it has ~3-4 years now, isn't? Jan 26 15:57:36 * ulitu1 hides behind a big rock Jan 26 15:57:58 ulitu1: thanks. i really think supporting freesmartphone.org would be a nice way to boost your audience. after all, the world is still waiting for an open phone, no matter all the false advertising google, nokia, palm, and friends performing Jan 26 15:58:56 mickey|cafe: define "open phone" Jan 26 15:59:11 mickey: I think you are right. Would you really care to work on that? Jan 26 15:59:22 mckoan: 100% free software, relatively competitive hardware Jan 26 15:59:23 if you had a developers sample? Jan 26 15:59:44 ulitu1: of course. i'd have a go at porting the FSO middleware to it Jan 26 15:59:57 ulitu1: are you at FOSDEM? Jan 26 16:00:25 I'll see what I can do. But don't be too excited yet. Jan 26 16:00:32 no FOSDEM Jan 26 16:01:09 mickey|cafe: time to visit berlin again? Jan 26 16:01:14 well Jan 26 16:01:23 with zecke in Asia... Jan 26 16:01:24 mickey|cafe: considering that the killer app for a phone are applications themselves (iPhone teach) an "open phone" will be utopic Jan 26 16:01:33 mickey|cafe: that is going to change 2011 Jan 26 16:01:39 I will be back to berlin Jan 26 16:01:54 mickey|cafe: all applicaztions will be stolen by competitors Jan 26 16:02:31 mckoan: that's fine. it'll increase the developers base Jan 26 16:02:34 mckoan: maybe german security gov office wants some Jan 26 16:02:36 mckoan: depends on who you want to sell to. i agree that for consumer market openness is irrelevant, however for special purposes (training, research, vertical markets) there should still be quite a market Jan 26 16:02:40 mckoan: well, they will be copied anyway. it is just a matter of releasing app x month earlier than competition Jan 26 16:02:56 zecke: sounds good! the two of you? Jan 26 16:03:03 hi zecke Jan 26 16:03:27 mickey|cafe:phone market don't care about small volumes purposes (training, research, vertical markets) Jan 26 16:03:48 to be able to release soon, i need to get over the libice-native configuration ;-) Jan 26 16:03:53 anyone has an idea? Jan 26 16:03:59 mickey|cafe: yes, two of us Jan 26 16:04:01 mckoan: i doubt that vertical markets must be small Jan 26 16:04:04 zecke: great Jan 26 16:04:07 mickey|cafe: that's the reason why every "open phone" try has failed IMHO Jan 26 16:04:14 well Jan 26 16:04:17 there was just one attempt ever Jan 26 16:04:18 seriously Jan 26 16:04:24 and this one failed because of stupid marketing Jan 26 16:04:28 imo Jan 26 16:04:31 maybe theiy failed because they didn't work ?? Jan 26 16:04:37 3miyou mean openmoko? Jan 26 16:04:39 bbiab Jan 26 16:05:05 fundemantal reason for OM failing was the obsession with bubbles in the UI ignoring basic phone functionality Jan 26 16:05:29 XorA: agree Jan 26 16:05:32 yep Jan 26 16:05:43 users are even fine with 2001 gsmmodem inside Jan 26 16:06:50 working for OM was an education in what not to do Jan 26 16:11:31 anyone have a clue about the libice-native configuration problem? Jan 26 16:12:26 ulitu1: what is the problem it was building ok for me on ubuntu 9.10 Jan 26 16:12:42 checking if xorg-macros used to generate configure is at least 1.2... configure: error: configure built with too old of a version of xorg-macros.m4 - requires version 1.1.0 or newer Jan 26 16:13:07 ulitu1: which distro Jan 26 16:13:09 i've seen this error message mentioned in 10 2009 for some other package Jan 26 16:13:28 a custom distro Jan 26 16:13:35 hmm Jan 26 16:13:37 based on ? Jan 26 16:13:59 let's see. it's been a while since i created it Jan 26 16:14:45 minimal Jan 26 16:14:51 your problem seems to be that you have util-macros which is not what it wants Jan 26 16:15:09 ulitu1: sync with the minimal.conf Jan 26 16:15:13 first Jan 26 16:15:42 then see if you have util-macros-native building or not Jan 26 16:15:56 or may be you have it in ASSUME_PROVIDED list Jan 26 16:16:15 then it will use the ones found on your build box which may not be what you want Jan 26 16:16:41 thanks khem. I'll check that Jan 26 16:17:03 XorA: and android was another reason Jan 26 16:19:59 khem: OM had already collapsed before Android Jan 26 16:24:42 hrw: I had the same problem with my scooter battery last evening :-D Jan 26 16:25:22 hrw: 13 years old battery (and scooter) Jan 26 16:27:41 * RP has a battery problem with his 38 year old car :/ Jan 26 16:28:06 38 years old? wow Jan 26 16:28:15 RP: a 38 years old battery would be quite much Jan 26 16:28:40 mckoan: The batteries are more like 10 years old ;-) Jan 26 16:28:58 Twin 6V ones - rather retro Jan 26 16:31:32 RP: which car is it? Jan 26 16:32:40 RP: Twin 6V is not surprising from people driving on the wrong side LOL Jan 26 16:34:52 * kergoth thinks about starting to blog, but specifically about embedded Linux and oe rather than random crap Jan 26 16:34:52 re Jan 26 16:35:00 kergoth: good planm Jan 26 16:35:07 heh :) Jan 26 16:35:24 i figure there has to be some crazy shit I've done with it that most haven't that people might find amusing or frightening Jan 26 16:35:29 or both Jan 26 16:38:30 hmm Jan 26 16:38:32 kergoth: o yes Jan 26 16:38:47 RP: you use volts in UK? Jan 26 16:39:06 I hoped for something strange Jan 26 16:39:42 tomtom hacking part3 Jan 26 16:54:10 have a nice rest of the day Jan 26 17:00:52 hrw: We use volts :) Jan 26 17:01:25 but leftside ones? Jan 26 17:01:45 kergoth: My blog seems to be about off-tarmac motorcycling these days. Seemed like the most interesting thing I do :) Jan 26 17:01:52 hrw: naturally Jan 26 17:02:34 RP: we thought you coded best in mud :-) Jan 26 17:02:47 hehe Jan 26 17:02:57 coling for the brain or something Jan 26 18:01:23 eFfeM_work: 2.6.33-rc1 works ok for you on kirkwood? Jan 26 18:01:23 re Jan 26 18:11:04 kergoth: I don't support you still have my GIT tree as a remote do you? I just noticed another tiny shadow issue I sorted in http://git.openpandora.org/cgi-bin/gitweb.cgi?p=openembedded.git;a=commit;h=70d5f0947e444cccf34008047beb3a7af436c7a3 Jan 26 18:13:23 DJWillis: pushed. Jan 26 18:21:24 anyone familiart with the la file for libstdc++ we install with the native sdk? Jan 26 18:21:46 we install libstdc++.la in /usr/lib, but the paths in the .la file say /lib Jan 26 18:21:53 how do we fix this? Jan 26 18:22:07 I edited by hand and am testing on the beagle atm Jan 26 18:23:53 crap, still an issue Jan 26 18:30:00 kergoth: thanks Jan 26 19:22:52 re Jan 26 19:32:58 Crofton|work: The sdk gcc recipes need work to fix that Jan 26 19:33:04 Crofton|work: I think Poky's work Jan 26 19:34:51 Crofton|work: There may be more than one .la file that needs fixing Jan 26 19:43:23 hmm Jan 26 19:43:52 * kergoth thinks about doing an external python tool that uses the bb python package to import recipes from collections into a project, rather than using the newcollection.inc approach Jan 26 19:45:26 Crofton|work: libstdc++-dev is installed on the target ? Jan 26 19:50:14 kergoth: Is this your crazy bootstraping? Jan 26 19:51:20 nah, thats collections.inc. newcollection is a .inc/task that looks at __depends and deploys a recipe, filespath files, its classes, etc into a directory of your choosing. handy for an initial population of a poky or mvl6 like more minimal set of metadata. also used it recently when syncing the latest oe changes into mvl6 Jan 26 19:52:26 kergoth: interesting :) Jan 26 19:53:20 * RP swears rootfs generation time has sprialled out of control, but why :/ Jan 26 19:54:03 RP: yes I observed that too on native-sdk-image it takes like 25mins Jan 26 19:54:09 yikes Jan 26 19:54:15 on a decent p4 3ghz machine Jan 26 19:54:47 oh, anyone have any suggestions for what to pick up as a new development / build box? core 2 quad or something the way to go? Jan 26 19:55:06 i7 Jan 26 19:55:19 or i5 if your budget allows Jan 26 19:55:43 hmm, i7 are a bit pricier than 2, i assume there's a nice performance boost there Jan 26 19:56:09 kergoth: ddr3 makes difference Jan 26 19:56:11 yes i5 is good on power too its a nice mix Jan 26 19:56:31 but i5 are dualcore only? no quads Jan 26 19:57:15 hmm Jan 26 19:57:35 thinking about doing more working from home, or freelancing a bit at some point, having a decent dev machine would be good Jan 26 19:58:03 I have core2 quad q6600 but it has ~2 years now Jan 26 19:58:29 * XorA needs a bigger monster these days Jan 26 19:58:33 yeah ddr3 is big boost Jan 26 19:58:51 * XorA has the same as hrw Jan 26 19:58:57 kergoth: generally - go for ddr3 system. socket 1156 (i3/i5) or 1366 (i7) or maybe something from amd with ddr3 (if such exists) Jan 26 19:59:11 okay, thanks, that helps :) Jan 26 19:59:14 * kergoth jots that down Jan 26 19:59:33 * kergoth 'll probably just keep an eye out on dealnews/fatwallet/etc, too lazy to build one nowadays Jan 26 19:59:46 well, not lazy, just a matter of priorities Jan 26 19:59:47 heh Jan 26 20:00:03 same here, also when you order it you can send it back when the ram fails Jan 26 20:00:13 kergoth: gigabyte has now mainboards which gives 1.5A for each usb port Jan 26 20:00:13 and when seems to be a norm these days Jan 26 20:00:21 heh Jan 26 20:00:40 hrw: I had a gigabyte board that gave 12A from USB :-) Jan 26 20:00:50 hrw: USB +5V was direct from PSU Jan 26 20:00:56 XorA: awesome Jan 26 20:01:10 hrw: awesome until usb device cooks itself Jan 26 20:01:18 sure Jan 26 20:01:32 that reminds me schurig's 3.3V usb dongles Jan 26 20:02:37 for their boards (which were 3.3V) they ordered some usb dongles without 5V->3.3V converters. too bad that non-engineers tried to use them in pc... Jan 26 20:03:03 * XorA discovers why OOO.org is so slow Jan 26 20:03:34 * hrw reboots sheeva Jan 26 20:03:37 21:00:37 up 51 days, 23:07, 1 user, load average: 1.12, 1.09, 1.09 Jan 26 20:04:13 kergoth: quad core Nehalems rock Jan 26 20:04:17 bb after reset Jan 26 20:05:42 kergoth: which opkg version are you using? Jan 26 20:06:14 RP: where is our OE discount on Intels, we burn enough of them out :-) Jan 26 20:06:33 XorA: :) Jan 26 20:07:45 XorA: two weeks of Poky burns out cheap (non-Intel) SSDs Jan 26 20:08:10 so Intel donated you Extreme SSD ? Jan 26 20:08:13 RP: in mvl6 right now we're using srcrev 197, it seems, but post-oe-sync we'll be up to 446 Jan 26 20:08:16 heh Jan 26 20:08:23 meh.. is the week over yet? Jan 26 20:08:33 kergoth: I'm wondering if uptream has fixes for the speed... Jan 26 20:08:40 ah Jan 26 20:08:52 kergoth: I'm on 215, upstream is 518 Jan 26 20:09:00 heh, they've certainly been busy Jan 26 20:09:17 XorA: I don't have an Intel SSD Jan 26 20:13:47 kergoth: I'm wondering if its installing the packages multiple times Jan 26 20:13:56 kergoth: Looking at the logs it worries me Jan 26 20:14:03 yeah, it shows the messages many times Jan 26 20:14:04 very odd Jan 26 20:14:08 * kergoth never got around to investigating Jan 26 20:14:45 hm, what is wrong if I try to rebuild console-image from packaged staging and I get errors like Cannot find package task-base (in log.do_rootfs) Jan 26 20:15:15 eFfeM: task-base*.ipk doesn't exist yet something depends on it Jan 26 20:15:21 hmm.. Jan 26 20:15:52 * kergoth as an experiment, does a dump of rodata in the binaries in staging to see how many have a string for the full path to staging Jan 26 20:15:56 RP, figured that out but I would expect if I start with an empty tmp it gets rebuild if it is missing Jan 26 20:16:15 kergoth: Look at the logs from insane - crazy amounts :( Jan 26 20:16:21 :( Jan 26 20:16:57 133 out of 323 binaries in the native staging bindir.. this makes me sad Jan 26 20:23:13 kergoth: Thats better than I expected Jan 26 20:32:38 has anyone seen "tar: .: file changed as we read it" during kernel's package_stage? Jan 26 20:33:22 the only tar there is to archive all modules Jan 26 20:34:04 I've only ever seen that message with TMPDIR on nfs, which I guess is not your situation. Jan 26 20:35:14 yeah, I've heard of NFS caching the metadata/attributes... no, I'm on ext4 Jan 26 20:38:20 denix: I am seeing different issue in do_package_update_index_ipk for console-image Jan 26 20:39:22 http://paste.lisp.org/display/93953 Jan 26 20:39:31 RP, checked if task-base exist but I have pstage/angstromglibc/staging-task-base-beagleboard-angstrom-linux-gnueabi_1.0-r87_i686-linux.ipk Jan 26 20:39:51 and even if I do a bitbake of it and then a bitbake of console-image the error stays Jan 26 20:41:25 basically this is where it became to go wrong: http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-January/016668.html Jan 26 20:44:02 khem: mine just says "function staging_packager failed" and gives the above tar message Jan 26 20:45:27 denix: so I am not alone Jan 26 20:54:25 what comes first Makefile.in or Makefile.am Jan 26 20:54:34 .am Jan 26 20:57:08 .am is input to automake Jan 26 20:57:57 khem: thought so, now where is Makefile.in getting the silly values :) Jan 26 20:58:15 like @xxxx@ Jan 26 20:58:34 they are filled in by autoconf Jan 26 20:58:45 when generating the Makefile at last Jan 26 21:05:57 am I doomed if I am using python from build box instread of python-native ? Jan 26 21:06:29 khem: python won't build Jan 26 21:07:34 RP: ok what are safe -native packages that I can use from build box Jan 26 21:08:05 right now a load of stuff gets baked for build box Jan 26 21:08:36 hello: what package do I need to rebuild so that the initscript links (/etc/rcX.d links) created with update-rc.d are regenerated in my image Jan 26 21:09:13 update-rc.d is run by the postinst of the particular package, for its init script. it isn't one package for all the links Jan 26 21:11:42 ok, then if I rebuild the package that inherits update-rc.d and defines INITSCRIPT_NAME and INITSCRIPT_PARAMS, the links should be created ? Jan 26 21:16:27 khem: Offhand, I'm really not sure Jan 26 21:16:42 khem: quilt is probably fine, as is git Jan 26 21:17:02 khem: That should stop quite a few dependencies creeping in Jan 26 21:17:41 kergoth: opkg is installing things multiple times afaict Jan 26 21:17:47 ugh Jan 26 21:19:55 kergoth: Looks like it checks "is installed" when it should also check "is unpacked" :/ Jan 26 21:20:05 * RP mutters Jan 26 21:21:11 grr, why is gnuradio trying to use libstdc++ from /lib? Jan 26 21:21:25 well libstdc++.la Jan 26 21:21:41 Crofton|work: If the .la files were corrupt, the corruption can spread to other .la files :/ Jan 26 21:21:45 Can't say, Crofton :-) Jan 26 21:22:18 ah Jan 26 21:24:16 We should fix the libstdc++.la paths Jan 26 21:25:00 * kergoth still thinks we should kill .la files wherever they aren't absolutely necessary Jan 26 21:25:31 kergoth: If I hadn't seen them actually serve a purpose on darwin I'd agree... Jan 26 21:25:56 well, no need to keep them around on non-darwin build/target, then :) Jan 26 21:26:05 kergoth: right :) Jan 26 21:26:05 just more complexity, more possible points of failure Jan 26 21:26:15 Would be a lot easier I admit... Jan 26 21:26:27 I just want to be able to try and compile gnuradio on the beagle .... Jan 26 21:26:38 A noble goal. Jan 26 21:26:38 Crofton|work: Wipe out all the .la files ;-) Jan 26 21:27:10 will that break anything? Jan 26 21:27:44 I am going to try a clean clone of the source first Jan 26 21:28:26 Crofton: It will most likely not break anything but no 100% guarantee Jan 26 21:28:42 Crofton: Assuming this is a linux target Jan 26 21:28:43 it is only a sd card Jan 26 21:28:55 shouldn't, the main place they're really needed is for linkers that don't support shared libs linking to other shared libs.. and to deal with odd install path issues, but the latter is better dealt with via pkg-config, and likely isn't an issue if everything is installed int he same place Jan 26 21:29:34 Hmm, so the package always says its not installed. opkg is so so so broken Jan 26 21:30:18 * kergoth tries to recall how the postinsts are handled.. remembers its via package state, but not which states Jan 26 21:30:32 kergoth: its not that :/ Jan 26 21:30:37 damn Jan 26 21:56:02 can a task RDEPEND on another task? Jan 26 22:01:13 Crofton: task as in do_foo ? Jan 26 22:02:27 kergoth: Its some kind of recursion nightmare in opkg :/ Jan 26 22:02:34 ugh, that sucks Jan 26 22:03:01 localedef get half way through installing then triggers other stuff which triggers itself to install again and there are no guards against the recursion Jan 26 22:03:20 If I understand the crazy debug right anyway Jan 26 22:04:00 hi RP Jan 26 22:04:09 hi woglinde Jan 26 22:04:24 I am trying to extend task-sdk-native Jan 26 22:04:28 heh, some days it feels like we need a mini apt & dpkg, and say fuck opkg Jan 26 22:04:36 I think I can RDEPEND on it and add packages Jan 26 22:04:45 kergoth: right :/ Jan 26 22:04:51 so I do not ave two ipk's with the same content Jan 26 22:05:13 * RP wants to cry at this debug. Explains why its so slow though Jan 26 22:28:35 hmm, 4270b7062d5e385282d2ad2b636f4535f9219b72 in bitbake seems to cause a regex exception for me, even not specifying --ignore-edeps.. Jan 26 22:28:41 * kergoth reverts to check Jan 26 22:32:23 * RP wonders if chunks of stdout are repeating somehow in the logs Jan 26 22:34:29 kergoth: may be name and target should be validated before calling re.search Jan 26 22:51:40 yeah, i had to revert that to do builds at all based on the current 1.10 branch. will have to test current master Jan 26 22:51:54 damn Jan 26 22:52:08 * kergoth pokes at the code Jan 26 22:52:11 why the disk virtio for windows now suckz so badly on my computer Jan 26 22:53:33 kergoth: Ignore the current 1.10 branch, please use master... Jan 26 22:54:17 * kergoth nods, was just thinking that that would be better.. was procrastinating merging the few remaining mvl6 specific patches into master, but oh well :) Jan 26 22:54:42 kergoth: I just removed that branch, just to make this more clear ;-) Jan 26 22:54:49 good Jan 26 22:55:36 Something is corrupting the opkg log output :( Jan 26 22:57:22 When we fork for the gzip process it breaks Jan 26 23:08:17 kergoth: How annoying - after all that it was the logfile that was screwed up :/ Jan 26 23:09:45 :\ Jan 26 23:38:28 khem: hmm, I disabled PARALLEL_MAKE and BB_NUMBER_THREADS and it no longer fails on tar in the kernel staging... Jan 27 00:59:24 Anyone familiar with the zd1211 WiFi firmware/driver? Jan 27 00:59:48 I'm trying to invoke it under Angstrom, on a Beagleboard, to no avail Jan 27 01:00:59 robtow, you're trying to modprobe the driver, and it fails to load? Jan 27 01:01:29 grg - modprobe zd1211rw fails Jan 27 01:01:36 fails how? Jan 27 01:01:49 FATAL: Module zd1211rw not found. Jan 27 01:02:22 well its either not there or not in the modules.dep file Jan 27 01:03:43 hmm Jan 27 01:04:10 wtf ... my name appears as a contributor to the zd1211 driver... http://lwn.net/Articles/186311/ Jan 27 01:04:16 i don't remember that Jan 27 01:05:39 grg - no zd1211 in modules.dep, altho there is a zd1211 dir in /lib/firmware. Jan 27 01:06:54 robtow, is there a zd1211 module in your /lib/modules/`uname -r` tree? Jan 27 01:08:26 No, there is not. Jan 27 01:08:49 root@beagleboard:/lib/modules/2.6.29-omap1# ls Jan 27 01:08:49 kernel modules.dep modules.isapnpmap modules.seriomap Jan 27 01:08:49 modules.alias modules.ieee1394map modules.ofmap modules.symbols Jan 27 01:08:49 modules.ccwmap modules.inputmap modules.pcimap modules.usbmap Jan 27 01:09:32 So what do I need to do to get the module? Jan 27 01:10:44 install the opk for that module i suppose Jan 27 01:12:03 opkg list "*zd1211*" Jan 27 01:12:09 the opkg install it Jan 27 01:12:16 s/the/then/ Jan 27 01:15:21 opkg install zd1211-firmware Jan 27 01:15:21 Package zd1211-firmware (1.4-r0.1) installed in root is up to date. Jan 27 01:15:54 the list also show a -dev and a -dbg Jan 27 01:18:38 Hmmm. When I try to install the -dev or the -dbg I get this: Jan 27 01:18:51 * Failed to download http://www.angstrom-distribution.org/feeds/2008/ipk/glibc/armv7a/base/zd1211-firmware-dev_1.4-r0.1_armv7a.ipk. Jan 27 01:18:51 error detail: Couldn't connect to server Jan 27 01:18:51 * Failed to download zd1211-firmware-dev. Perhaps you need to run 'opkg update'? Jan 27 01:21:10 robtow, the firmware is the code that is uploaded by the driver to the device. you still need the actual driver Jan 27 01:22:38 robtow, it looks like you have an angstrom specific support issue Jan 27 01:22:46 perhaps you should try #angstrom ? Jan 27 01:23:01 thanks I shall Jan 27 02:37:54 Hmmm. "bitbake zd1211" fails...... see http://pastebin.com/m27e3a558 Jan 27 02:40:50 robtow: well, you are qualified to send a fix then. ;) Jan 27 02:41:12 zecke.... thanks for the encouragement! Jan 27 02:42:07 robtow: and mainline linux has this driver included anyway Jan 27 02:42:51 zecke - it's not included in the Angstrom I'm currently using; my problem is to get it on a Beagleboard. Jan 27 02:43:52 robtow: well, then change the defconfig of your kernel, zd1211 support is in 2.6.18+ Jan 27 02:44:34 uname -r in the Angstrom says "2.6.29-omap1" Jan 27 02:44:59 There are zd1211 files in /lib/firmware. Jan 27 02:45:18 robtow: 03:43 < zecke> robtow: well, then change the defconfig of your kernel, zd1211 support is in 2.6.18+ Jan 27 02:46:38 robtow: as grg already said, the firmware and the kernel module are two different things? having the files in /lib/firmware doesn't mean your kernel supports it. Jan 27 02:46:46 s/\?/./ Jan 27 02:47:31 Right. Jan 27 02:49:25 robtow: recipes/linux/linux-omap-2.6.29/beagleboard/defconfig should contain the defconfig for the beagleboard, just enable your driver and rebuild the kernel Jan 27 02:49:41 thanks zecke Jan 27 02:50:07 robtow: and actually it is set to =y... which means it is compiled into the kernel already. :) Jan 27 02:50:46 zecke - I thought it was, which puzzles me why I'm not getting the functionality. Jan 27 02:51:42 robtow: what does dmesg say? does it recognize your usb module? does it try to bind a driver? Jan 27 02:51:55 robtow: you can still enable the ZD1211 debug messages in the kernel. Jan 27 02:53:25 dmesg output: Jan 27 02:56:41 http://pastebin.com/m6848a050 Jan 27 02:58:15 gm zecke Jan 27 02:59:28 <----is puzzled and dismayed. :-( **** ENDING LOGGING AT Wed Jan 27 02:59:57 2010