**** BEGIN LOGGING AT Thu Nov 09 02:59:57 2006 Nov 09 06:48:23 hello in case there is someone already awaken Nov 09 06:49:31 good morning (from someone who isn't quite awaken yet) :) Nov 09 07:09:13 good night, kergoth Nov 09 07:45:14 morning Nov 09 08:10:53 hi Nov 09 08:13:46 hi Genesis ! Nov 09 08:28:26 morning Nov 09 08:33:38 good morning all Nov 09 08:33:40 03dirk 07org.oe.dev * r00fbddaa... 10/ (3 files in 2 dirs): file: add 4.18 version Nov 09 08:35:05 morning koen and XorA Nov 09 08:35:22 hey XorA & valLongfield Nov 09 08:35:53 morning koen Nov 09 08:36:13 hey greentux Nov 09 08:38:30 03koen 07org.oe.dev * rb362de21... 10/ (3 files in 2 dirs): disapproval of revision '00fbddaa13e002e339a062c328b76d88c57a345e' Nov 09 08:39:46 wow ... wget expects 1h download for fakeroot from openzaurus.org ! Nov 09 08:47:15 koen: why dissaprove, DEFAULT_PREFERENCE = -1000000 should do Nov 09 08:48:48 they are broken, why increase the crap level of the metadata? Nov 09 08:50:23 hey lrg Nov 09 08:50:36 hey koen Nov 09 08:59:59 morning all Nov 09 09:02:09 hey ade|desk Nov 09 09:07:13 hi Nov 09 09:07:27 hey obergix[work] Nov 09 09:08:20 valLongfield: we need to get rid of it Nov 09 09:09:08 koen: from what Mickeyl was saying FIC phone has BT (but no Irda) Nov 09 09:09:27 03koen 07org.oe.dev * r3d7e2cab... 10/ (1 packages/geoclue packages/geoclue/geoclue_svn.bb): geoclue: add geoclue, a dbus service to find out where you are via various means like gps or hostip.net Nov 09 09:09:39 koen@bitbake:~/OE/monotone/org.openembedded.dev$ grep FEATURES conf/machine/fic-gta01.conf Nov 09 09:09:40 MACHINE_FEATURES = "kernel26 apm alsa bluetooth usbgadget" Nov 09 09:10:31 hrw|work: that's what I recall as well, but the linuxdevices article hinted at bluetooth not being in the first model they release Nov 09 09:10:43 but their OE tree lists bluetooth :) Nov 09 09:12:53 http://www.vanille-media.de/site/index.php/2006/11/08/congratulations-madam-its-a-phone-2/ Nov 09 09:12:59 koen: read this Nov 09 09:13:13 any more questions? Nov 09 09:13:27 :) Nov 09 09:14:21 quad band as well Nov 09 09:16:46 sign me up for a discount on the black model :) Nov 09 09:16:58 * koen wonders if it syncs to OSX Nov 09 09:18:12 funambol sounds like vegan food Nov 09 09:28:10 hey zecke Nov 09 09:28:28 Morning Nov 09 09:28:43 hey sirfred Nov 09 09:33:44 ho ho ho Nov 09 09:34:12 * zecke searches a flat... Nov 09 09:34:36 zecke: Laughing about the cheap prices? Nov 09 09:34:52 sirfred: no, it is more expensive than hotel mother ;) Nov 09 09:35:16 ;) Nov 09 09:39:28 we form a geek 'wg' :} Nov 09 09:39:56 wg? Nov 09 09:42:04 good morning Nov 09 09:42:20 hey Bernardo Nov 09 09:45:50 wow hrw|work you drop from irc a lot lately, your wires ok ? Nov 09 09:47:17 ade|desk: company where I work has 10Mbps (soon upgraded to 100Mbps) wire but microdrops are often.. Nov 09 09:48:58 I though you might be using the same kind of setup I do Nov 09 09:51:41 zecke: was ist wg? Nov 09 09:52:45 cgi::irc running at a webserver outside company firewall Nov 09 09:52:45 wohn gemeinschaft ~ flat share Nov 09 09:52:56 ah.. Nov 09 09:59:58 hrw|work: conecto.pl? one step beyond Nov 09 10:00:10 to a native speaker, is one step beyond positive? Nov 09 10:01:41 zecke: any new bonsai improvements? Nov 09 10:02:06 the new term starts next week so I have some spare time this week Nov 09 10:09:10 NOTE: package file-4.18: completed Nov 09 10:09:31 koen: I'm busy fighting MFC :( Nov 09 10:09:37 XorA: file-native too? Nov 09 10:10:20 hrw|work: yes Nov 09 10:10:29 nice Nov 09 10:11:04 hrw|work: the stuff that doesnt work in 4.17/4.18 was added for 4.17/4.18, so if I take two of the magic files from 4.16 it works Nov 09 10:11:12 hrw|work: I shall now commit Nov 09 10:11:39 * XorA gueses a lack of QA from author Nov 09 10:11:48 XorA: I also have a batch of gnome 2.16 stuff, but I'l write the no-docs.patches first before committing :) Nov 09 10:12:45 XorA: send him a note about it, so file 4.19 maybe will get better Nov 09 10:12:49 XorA: Hi. Did you find time to test the mplayer driver? Nov 09 10:13:01 sirfred: sorry been snowed under Nov 09 10:13:12 XorA: No problem. Just curious Nov 09 10:13:15 ~lart samsung kernels Nov 09 10:13:16 * ibot makes a balloon animal out of samsung kernels Nov 09 10:16:20 03xora 07org.oe.dev * rfa302963... 10/ (5 files in 3 dirs): Nov 09 10:16:20 file, file-native 4.18 : replaced the magic files for filesystems and dump Nov 09 10:16:20 for the ones for 4.16. The stuff that doesnt compile in 4.18 is new so it Nov 09 10:16:20 looks like a lack of QA on part of this release that they dont work. Nov 09 10:17:33 hrw|work: did you have time to check the problems I am having with arm-linux-strip in the oz354x branch? Nov 09 10:18:45 nope Nov 09 10:44:25 ooooh Nov 09 10:44:32 Xora-mail on LAK Nov 09 10:45:56 * koen also notices people doing a port for netbook (non-pro) Nov 09 10:57:59 hi all Nov 09 10:58:48 hey florian_kc Nov 09 11:04:23 morning Nov 09 11:23:54 is svn.o-hand.com working for anyone? Nov 09 11:25:38 ade|desk: it used to Nov 09 11:26:00 can't ping it Nov 09 11:26:10 maybe the right address is monotone.o-hand.com? Nov 09 11:26:50 hi Dirk Nov 09 11:29:49 zecke: well their http://projects.o-hand.com/matchbox/get_involved.html indicates svn.o-hand.com .... makes me wonder is they updated subversion and killed apache svn ness Nov 09 11:30:42 ade|desk: well svn.o-hand.com seems to be down Nov 09 11:50:51 hi guys... I have a few python module recipes laying around that may be worth adding to the oe repo Nov 09 11:51:16 pylirc, pythondaap and pyzeroconf Nov 09 11:51:33 attach them to the bugtracker Nov 09 11:51:43 tkp: -> bugs.oe.org Nov 09 11:51:44 koen: k Nov 09 11:51:52 and wait for mickeyl the python master Nov 09 11:52:08 interesting: http://bugs.oe.org/ Nov 09 11:52:17 bugs.openembedded.org Nov 09 11:52:23 ( I know ) Nov 09 11:52:35 enter it as a bug? Nov 09 11:52:44 yes and attach the files Nov 09 11:52:54 'attach', not 'paste inline' Nov 09 11:53:05 np Nov 09 11:56:08 done :) Nov 09 11:56:31 If I'm currently using a non premptive kernel can I just rebuild my kernel with premptive support enabled or do I also need to rebuild my cross tools and rootfs? Nov 09 11:56:47 kernel Nov 09 11:57:51 hrw|work, thanks. Nov 09 12:00:53 Gerrath: I've noticed you do have to rebuild your kernel modules and update them on your rootfs. Nov 09 12:02:15 cbrake, yeah I figured that since I think compiling prem. changes the kernel version. The last time I tried enabling it I started to get kernel dumps and thought I might need to recompile something else as well. I'll give it another try. Nov 09 12:02:33 cbrake, so modversion won't match, I think :-) Nov 09 12:04:59 Gerrath: Yeah, it should work. Fixes a lot of problems when you are trying to make a system responsive :-) Nov 09 12:05:23 cbrake, good, are you using it on your LPD board? Nov 09 12:06:04 Gerrath: no, just the compulab board so far. It is still running 2.6.16, so the kern version is different. Nov 09 12:08:54 Gerrath: eventually, I'll move to something newer. 2.6.18 supposedly has additional real-time improvements. Nov 09 12:11:02 cbrake, I was thinking of pushing the logicPD patch to 2.6.18 . Nov 09 12:11:53 I expect it will just work, I think I even tried it once. Nov 09 12:12:58 cbrake, I might play with it this weekend :-) Nov 09 12:13:23 Gerrath: sounds good Nov 09 12:35:05 hey mallum Nov 09 12:35:30 hey koen Nov 09 12:35:43 mallum: congrats on the little one Nov 09 12:35:57 http://entertainment.newsforge.com/entertainment/06/11/01/1413204.shtml?tid=23&tid=132 Nov 09 12:36:10 hi Crofton Nov 09 12:37:07 i.mx21, nice Nov 09 12:37:16 still not a mx31 though Nov 09 12:37:56 hello; I'm having a bit strange problem - if I use bitbake (angstrom)-bootstrap-image, creating image fails with | | mkfs.jffs2: opening directory /home/mmp/arm/oe_root/build/tmp/rootfs/etc/ipkg/arch.conf: Not a directory Nov 09 12:38:10 but if I run mkfs.jffs2 directly, it just works correctly... . Nov 09 12:38:22 hi mallum Nov 09 12:38:43 Crofton: what an ugly piece of hardware :-) Nov 09 12:39:11 hey florian_kc Nov 09 12:39:18 koen: thanks :) Nov 09 12:41:58 ah, but you can open it without destroying it :) Nov 09 12:42:49 Chumby, gotta catch them all Nov 09 12:45:46 XorA: http://xkcd.com/c178.html Nov 09 12:46:20 koen: :-D Nov 09 12:51:02 ljp: why was the Greenphone SDK widthdrawn? Nov 09 12:54:06 XorA: dbus 1.0 should get released today Nov 09 12:55:47 koen: cool Nov 09 12:59:59 ah, hal and dbus dont like each other Nov 09 13:00:06 s/ah/bah/ Nov 09 13:00:13 koen: it is tagged, build it and test it? Nov 09 13:00:25 XorA: what is the exact problem? Nov 09 13:00:41 mmp: undefined reference to `dbus_connection_disconnect' Nov 09 13:01:36 XorA: didn't I patch that out for networkmanager? Nov 09 13:02:50 XorA: s/dbus_connection_disconnect/dbus_connection_close/g Nov 09 13:02:59 XorA: see http://www.openembedded.org/repo/org.openembedded.dev/packages/networkmanager/files/dbus-api-fix.patch Nov 09 13:04:00 koen: this is hal Nov 09 13:04:44 XorA: I suspect hal is still using the 0.6x api instead of the 0.9x api Nov 09 13:05:00 I patched it out in a load of gnome stuff as well, I just forgot about it Nov 09 13:06:13 hal 5.8 needs polkit whatever that is Nov 09 13:08:33 XorA: http://ewi546.ewi.utwente.nl/tmp/dbus-initscript.diff Nov 09 13:08:51 XorA: that's the diff between our script and ubuntu's 0.93 script Nov 09 13:14:01 hey Nov 09 13:14:56 03koen 07org.oe.dev * r6af5be50... 10/ (3 files in 3 dirs): dbus 0.94: fix initscript, closes #1563 Nov 09 13:18:50 03koen 07org.oe.dev * r103ed846... 10/ (5 files in 3 dirs): Nov 09 13:18:50 dbus: add 0.95 cross version Nov 09 13:18:50 * cross.patch isn't needed anymore Nov 09 13:18:50 * dbus-uuidgen is now run in the initscript instead of the postinst Nov 09 13:19:08 XorA: more toys to play with ;) Nov 09 13:43:31 03xora 07org.oe.dev * r051aed61... 10/ (4 files in 3 dirs): hal_0.5.7.bb : fix for dbus 1.0 API changes Nov 09 13:44:24 mallum: little one ?? Nov 09 13:48:34 ade|desk: I became a daddy Nov 09 13:48:52 mallum: oh congratulations !!!!! Nov 09 13:49:30 mallum: size, m or f etc ? Nov 09 13:49:54 female, 2 months old now :) Nov 09 13:49:54 eeps more mini geeks Nov 09 13:49:57 mallum: congrats Nov 09 13:50:05 thanks Nov 09 13:50:17 mallum: what paternity benefits did you give yourself ? Nov 09 13:50:35 hehe Nov 09 13:51:21 hello lrg Nov 09 13:51:35 so what was that big mickey announcement after all? Nov 09 13:51:38 hey valLongfield Nov 09 13:52:30 lrg: I sent you an email an hour ago Nov 09 13:52:46 hrw|work, well I just compiled my kernel with PREEMPT support and now I get occasional kernel dumps. I will try to decipher the dump but I wondered if you can think of anything else that may need recompiled or other options that may need set in the kernel config other than PREEMPT enabled? Nov 09 13:53:02 valLongfield: got it, will reply soon Nov 09 13:53:12 lrg: cool thank you very much Nov 09 13:53:28 I've been running slugos preempt for a year now without problem. Nov 09 13:54:06 ade|desk: not enough :( Nov 09 13:54:30 interesting, if I go into /media/cf (compact flash mounted) and do an ls the kernel dumps.. if I do another ls every thing is fine, if I then unmount the CF then remount it and do an ls on it then the kernel dumps again.. Nov 09 14:00:04 Gerrath: check oops content Nov 09 14:00:14 Gerrath: maybe cf driver is not preempt save? Nov 09 14:10:28 hrw|work, I suspect it is the preempt save missing from the the CF.. I need to read up on what preempt save is and how to implement it :-) Thanks again for the help. Nov 09 14:21:11 /data/build/koen/OE/build/tmp/angstrom/work/armv4t-angstrom-linux-gnueabi/gcc-cross-4.1.1-r8/gcc-4.1.1/libgfortran/generated/_mod_i8.F90:42: internal compiler error: Segmentation fault Nov 09 14:21:14 * koen stabs gcc Nov 09 14:23:23 koen: I get that is I dchroot into 32bit but forget to use linux32 Nov 09 14:23:52 this is 32bit... Nov 09 14:27:57 hrw|work, lol sorry I understand.. you meant preempt safe, I guess I'm having a slow day :-/ Nov 09 14:33:14 ops Nov 09 14:49:14 elemental function specific__mod_i8 (p1, p2) Nov 09 14:49:14 integer (kind=8), intent (in) :: p1, p2 Nov 09 14:49:14 integer (kind=8) :: specific__mod_i8 Nov 09 14:49:14 specific__mod_i8 = mod (p1, p2) Nov 09 14:49:15 end function Nov 09 14:49:24 * koen wonders why gcc ICEs on that Nov 09 14:53:00 ping lrg Nov 09 14:56:28 hey greentux Nov 09 14:56:34 hi lrg Nov 09 14:56:47 lrg: remember, we had also some headphone probs with the zaurus... Nov 09 14:56:58 greentux: yep, we had a few Nov 09 14:57:10 lrg: can i check something in the /proc/asound ? Nov 09 14:57:20 lrg: i updated to recent version Nov 09 14:57:52 greentux: what are you checking ? Nov 09 14:58:08 sound plays with headphones, but not with internal speakr Nov 09 14:59:42 greentux: what version ? Nov 09 14:59:49 1.0.13 Nov 09 15:00:08 hda_codec: Unknown model for ALC262, trying auto-probe from BIOS... Nov 09 15:00:13 perhaps the problem Nov 09 15:01:39 greentux: this is not zaurus so I can't really help, sorry :( Nov 09 15:02:14 unless thats the new secret SL-4000 Nov 09 15:02:43 XorA: yesss. the size is nearly the same Nov 09 15:03:06 03koen 07org.oe.dev * rfa332122... 10/ (1 packages/gcc/gcc_4.1.1.bb): gcc 4.1.1: make $JAVA and $FORTRAN actually have an effect on $LANGUAGES Nov 09 15:03:45 XorA: sneak in distcc into mickeyl's phone-image, so we have a huuuuuge native cluster over gprs ;) Nov 09 15:04:11 hehe Nov 09 15:05:59 hi, I am trying to build for arm using openembedded, I have a prebuilt toolchain; I followed the getting started guide and especially the "Using a prebuilt toolchain to create your packages" guide; somehow when it comes to building the ncurses package do_configure failes with an error, the log says "checking for C compiler default output file name... configure: error: C compiler cannot create executables"... any idea what I could have done wrong? Nov 09 15:06:18 Jin^eLD rule of thumb Nov 09 15:06:26 take a look at the config.log of the specific package Nov 09 15:06:39 and see what it tried to execute and normally you see an error as well Nov 09 15:06:46 e.g. unknown march/mcpu Nov 09 15:07:31 another rule of thumb: most prebuilt toolchains suck Nov 09 15:07:32 zecke: was doing that but to be honest I could not locate the exact position of the failure, there are plenty of failures there that seem to pass ok; ok let me take another look Nov 09 15:07:57 koen: I have to compile for a system that has some sort of prebuilt linux on it Nov 09 15:07:57 Jin^eLD: feel free to pastebin it Nov 09 15:08:12 Jin^eLD: and greppig for "default output file name" can help as well Nov 09 15:09:49 oh crap :) I was looking for at wrong config.log Nov 09 15:10:28 that explains why I could not find the error heh Nov 09 15:12:01 http://images.ucomics.com/comics/ft/2006/ft061109.gif Nov 09 15:22:37 Foxtrot is in re-runs right now Nov 09 15:25:20 ok thanks :) my -mtune switch was wrong :> Nov 09 15:29:28 morning Nov 09 15:29:47 koen: Which version of gcc should be used for eabi? Nov 09 15:30:17 florian_kc: see the angstrom configs Nov 09 15:31:50 koen: ok, thanks Nov 09 15:36:47 Hello; I have a question about Angstrom - I just built my image, flashed it - kernel initializes successfully, but I noticed that there are strange parameters in /boot/params (like init=/linuxrc) and with init=/sbin/init I can't get it boot - it just stops after mounting / Nov 09 15:39:49 although it seems that it doesn't panic - when I insert/remove SD card, it generates output in the console - it just doesn't continue to boot Nov 09 15:43:43 I can't do much research about it; I can just flash the PDA, don't have any better way to alter the root filesystem (I don't have CF jacket) Nov 09 15:43:51 and the deviceis h5550 Nov 09 15:44:51 mmp: you could redirect the console in the bootloader parameters (to ttyS0) Nov 09 15:46:10 florian_kc: I already did (h5550 doesn't have bootloader support for LCD); but the things I posted are the only I can see. I can't say what happens, but it simply stops after running (trying to run?) /sbin/init Nov 09 15:47:16 and running init=/bin/ash fails with 'attempting to kill init', so I suppose there could be something wrong... . Nov 09 15:47:45 mmp: ash is busybox? Nov 09 15:48:08 mmp: if yes then do 'init=/bin/busybox sh' Nov 09 15:48:13 it should be... Nov 09 15:53:03 mmp: hrm nasty... sounds like for some reason it doesn't want to start any binary Nov 09 15:53:23 i suffer from the same effect with my a780 Nov 09 15:56:14 florian_kc: for now it seems so; I have no idea what is happening inside the silver box:) Nov 09 16:01:03 hrw|work: I just tried to run busybox; I'm not sure whether I can give arguments to init:( I tried to start 'vi' instead of init (I'm not sure whether 'vi' is part of busybox too), it seems to hang the same way as /bin/init... . Nov 09 16:03:46 mmp: 'init=/bin/busybox sh' - exactly in this way Nov 09 16:05:21 lalalala Nov 09 16:06:10 hrw|work: it still stops and does nothing more... hmm, this is _strange_ Nov 09 16:10:59 hi ar Nov 09 16:17:00 hi Nov 09 16:17:14 hrw|work, florian_kc: Thanks a lot, this seems to be quite complicated issue; maybe e-mail to h5400-port (and reply:) will give some clue... Nov 09 16:36:04 mmp: yw Nov 09 16:41:48 RP: wow, using the ringbuffer and the microcode packets to copy from host memory to the screen, my font demo goes from 17645.417236 cps to 47858.279669 cps Nov 09 16:46:24 It's addictive to see the zaurus to display text this fast. Nov 09 16:47:14 sirfred: what can make use of that accell? kernelfb, X, freetype? Nov 09 16:48:26 koen: This is mostly oriented to the framebuffer. I mean, simple two monochrome fonts where the card can make colour expansion. Nov 09 16:49:00 koen: The real gain of speed , given that the card bandwidth seems to be slow, is that I'm directly feeding the monochrome fonts, and the card is making the colour expansion to 16bpp Nov 09 16:49:40 bye Nov 09 16:49:42 sirfred: mplayer subtitles? Nov 09 16:50:03 koen: This is what the patch for the kernel framebuffer does, but feeding directly the HOST_DATA registers of the card is painfully slow. What I've found is that using the card command processor we can do it very fast. Nov 09 16:50:32 koen: Could be, if we find a way to blend them into the YUV surface. Nov 09 16:50:53 koen: That doesn't seem to be trivial. Because the imageon doesn't actually process planar surfaces. Nov 09 16:51:40 koen: In fact, the mplayer driver is processing the three planes separatedly, the imageon only knows that they are 8bpp planes. The only thing it seems to be able to do is directly overlaying it. Nov 09 16:52:18 koen: The bad thing about this font speed improvement, is that the microcode is needed. Nov 09 16:53:31 I don't know what legal issues could arise about using the libaticore extracted microcode. Nov 09 16:57:46 I have an oe setup here which does not try to download and build anything... i fails in the quilt staging then. Nov 09 16:59:27 * florian_kc feels a little bit clueless Nov 09 17:01:07 someone help Nov 09 17:02:35 i have changed the kernel version from .17-hh3 to .18-hh0 in the angstrom-2007.conf file, and in the linux-handhelds_cvs.bb file. But it still makes the 17-hh3 kernel. Wich files have i forgot to change? Nov 09 17:03:37 florian_kc: I had to "ASSUME_PROVIDED" quilt-native to build Nov 09 17:04:10 installing hrw's openembedded-essential packages also helped me Nov 09 17:04:46 Unfortunately, I'm now stuck building openzaurus for my akita both using the oz354x and the dev branch Nov 09 17:04:58 goxboxlive: because there's no .18-hh0 kernel Nov 09 17:05:03 goxboxlive: you maybe have a preferred kernel in your machine configuration Nov 09 17:05:19 koen: Yes there is Nov 09 17:05:59 Powerbook-2:~/Projects/OpenEmbedded/org.openembedded.dev/packages/linux koen$ ls | grep hh | grep "2.6" Nov 09 17:05:59 LAB-kernel-2.6.13-hh2 Nov 09 17:05:59 LAB-kernel_2.6.13-hh2.bb Nov 09 17:05:59 linux-handhelds-2.6_2.6.16-hh5.bb Nov 09 17:06:00 linux-handhelds-2.6_2.6.16-hh7.bb Nov 09 17:06:00 linux-handhelds-2.6_2.6.17-hh2.bb Nov 09 17:06:01 Powerbook-2:~/Projects/OpenEmbedded/org.openembedded.dev/packages/linux koen$ Nov 09 17:06:04 no, there isn't Nov 09 17:06:28 florian_kc: No there is no in the machine file. Nov 09 17:06:48 oen: I am building form linux-handhelds-2.6_cvs.bb Nov 09 17:08:21 that doesn't change the fact that there is no .18-hh0 kernel in OE Nov 09 17:08:21 goxboxlive: ah... oe doesn't know about this then. you need to set the preferred kernel to the cvs version. Nov 09 17:08:37 florian_kc: I have done that in the angstrom-2007.conf file. Nov 09 17:09:03 koen: why is it building .17-hh3 then? There is no hh3.bb file either Nov 09 17:09:21 hh3 is specified in the _cvs file Nov 09 17:12:55 hey do13 Nov 09 17:13:02 koen: ok, but i have changed that to hh0 (also 2.6.18) Nov 09 17:13:30 Hi! Nov 09 17:13:34 so i guess i need to make a .18.bb file then. Nov 09 17:13:36 hi Nov 09 17:14:08 psokolovsky: Man you have been active today. My gmail where filled with over 1000 new mails :) Nov 09 17:14:45 goxboxlive, well, so that issue is that you build from _cvs. That's hacky thing, not suitable for general use. So, yiu either shouldn't use it, or should know how to handle such issues ;-) Nov 09 17:15:16 goxboxlive, that happens on every upgrade ;-) Nov 09 17:15:35 but the fact that it was active yesterday, doens't mean it's already suitable for use today Nov 09 17:15:47 psokolovsky: Well you know that cr2 is comitting to HEAD so to involve the latest kernel drivers i have to build from cvs. Nov 09 17:16:07 stay away from it until it is added to OE.dev, or submit stuff to add it (tested) ;-) Nov 09 17:16:47 goxboxlive, then you should learn how to handle that stuff. I've been trying to make it awfully easy ;-) Nov 09 17:17:45 psokolovsky: Of course i will submit, but i have allready changed angstrom.conf file and linux-handhelds-2.6_cvs.bb so what's left then. Nov 09 17:21:09 03pfalcon 07org.oe.dev * rb45598ac... 10/ (1 packages/linux/linux-handhelds-2.6_cvs.bb): Nov 09 17:21:09 handhelds-2.6 cvs: HH.org CVS upgraded to 2.6.18, follow. Nov 09 17:21:09 * -hh0 stuff is *never* suitable for general use. Use at your own risk, Nov 09 17:21:09 don't ask questions, report results and send patches. Nov 09 17:21:16 goxboxlive, ^ Nov 09 17:22:08 umm. there's "instead" before "results", of course %) Nov 09 17:22:15 psokolovsky: "don't (ask, report, patch)" or "don't (ask), do (report, patch)" Nov 09 17:22:17 ? Nov 09 17:22:37 do report ;-) sorry ;-( Nov 09 17:22:54 that's why I don't like to upgrade in hurry ;-) Nov 09 17:25:37 * mrz80 is away: Metaphysically I'm still here; physically, I'm elsewhere Nov 09 17:26:29 psokolovsky: That was the same as i have done in the linux-handhelds-2.6_cvs.bb file. thanks anyway. Nov 09 17:27:00 goxboxlive, then, why so much noise? ;-) Nov 09 17:27:40 upgrading the version also means that it was QAed that kernel at least build for bunch of machines, if not runs w/o issues. Nov 09 17:27:46 that's not the case for now. Nov 09 17:28:28 So, could wait w/upgrade version too. Now people might think it's indorsed to use, and it's not the case ;-) Nov 09 17:28:28 psokolovsky: It wasent ment as noise, it's just my bad enlish. But, it has configured now, thx anyway. I'll shut up. Nov 09 17:28:40 np Nov 09 17:31:58 psokolovsky: any userspace changes needed for updating to .18? Nov 09 17:32:08 psokolovsky: congrats on gettting a hx4700 btw Nov 09 17:32:30 koen, so far, I didn't run userspace yet apart from dropbear ;-) Nov 09 17:32:33 psokolovsky: One more question :-) The same thing happens. All the modules is being written as 2.6.18-hh0 and the zImage. But the kernel-image ipk file is beeing named 2.6.17-hh3. Why is that? I have cleaned the linux-handhelds-2.6. Nov 09 17:32:34 koen, thanks! Nov 09 17:33:07 goxboxlive, I have no idea where -hh3 comes from at all. but need to check Nov 09 17:33:14 ok Nov 09 17:35:34 psokolovsky: I know opie will break over the corgi-bl, but (puts angstrom hat on) that's unsupported anyway Nov 09 17:37:45 * lrg -> train home Nov 09 17:37:54 lrg: enjoy Nov 09 17:56:25 koen: iirc, I ran 2.6.17 which was already on corgi-bl w/opie, and backlight work. iirc, there's a patch for 2.6 backlight classdev in opie's CVS HEAD. Nov 09 17:57:22 goxboxlive, so, I have no idea. there' clearly 2.6.18-hh0 in CVS HEAD Makefile. where -hh3 comes from, can't imagine. maybe you have it in some OE recipe/config? Nov 09 17:57:37 sirfred: That sounds rather interesting :) Nov 09 17:57:40 morning all Nov 09 17:59:19 Hi, RP Nov 09 18:00:08 hi RP Nov 09 18:00:10 bbl Nov 09 18:00:24 hey RP Nov 09 18:03:00 hmm.. I installed a couple of packages on the target platofrm and I have a problem with shared libraries; my package can not find shared libs that I also compiled and installed; can it be that this rpath thing has something to do with it? Nov 09 18:03:16 RP: Anyway, as the ringbuffer is placed in the internal card memory, the bandwidth I'm getting is more or less the same than copying directly into the mmapped memory. Of course, colour expansion is a gain with monochrome bitmaps because we're sending 1/16 data Nov 09 18:03:59 do I need to set some extra parameters when building packages? Nov 09 18:05:00 sirfred: I guess that can be useful on the console although its probably not so useful anywhere else? Nov 09 18:05:10 RP: I agree. Nov 09 18:05:30 RP: The framebuffer console should fly, but we need to upload the firmware to the card. Nov 09 18:05:51 RP: Perhaps it doesn't worth the effort Nov 09 18:06:39 RP: Well, another thing that could really be this useful for, is to host surfaces with non standar trajectories. Nov 09 18:07:21 RP: For example, thinking about video rendering, what I need to do now, is to copy the original image to the internal memory and then, rotate it using a blt with a rotated trajectory. Nov 09 18:08:27 Actually, three blts, one for any plane of the YUV image. If I was able to apply a rotation trajectory while host-copying, I should need only a surface , not two. Nov 09 18:08:28 But what I got was zaurus hangs, trying to do so. ;-) Nov 09 18:09:06 ...any idea who sets this "rpath" option when doing oe_runconf ? Nov 09 18:09:25 It seems that using a host data source is not compatible with some values in the DP_CNTL register Nov 09 18:11:37 sirfred: I hope you're keeping notes on all this - I wish I had when I was experimenting :) Nov 09 19:28:40 * mrz80 is back (gone 02:03:04) Nov 09 20:10:24 anyone with DVB-T in berlin? Nov 09 20:11:11 Good night. Nov 09 20:12:18 hi zecke Nov 09 20:13:33 RP: hey :) Nov 09 20:13:40 gosh I hate myself and T-Systems :( Nov 09 20:14:04 I forgot to turn my own DVB-H signal on and the berlin trial system looks like it crashed Nov 09 20:14:26 he koen Nov 09 20:14:42 hey woglinde Nov 09 20:14:48 zecke: working late? :) Nov 09 20:14:55 he rp Nov 09 20:15:03 RP: hehe, yes semi private project Nov 09 20:15:07 I want to get it done :) Nov 09 20:15:17 RP: we have this huge 42" LCD in the office Nov 09 20:15:44 RP: and now I want to have the live DVB-H signal on it 12 videos in a 4x3 mosaic Nov 09 20:16:20 RP: so my task for this night is implement the mpeg2 transport stream format Nov 09 20:16:33 but I have no test data as a foreign server looks like it crashed :( Nov 09 20:17:44 lo all Nov 09 20:17:45 zecke: :-( Nov 09 20:18:05 heya RP is there a way to have hostap firmware update work on poodle ? Nov 09 20:26:15 kantlivelong: It is possible and should work as far as I know Nov 09 20:26:40 RP: how would i enable it in the config so i can compile (im new to this bitbake thing) Nov 09 20:28:47 kantlivelong: Enable it in the defconfig and recompile the kernel... Nov 09 20:29:01 RP: where? Nov 09 20:29:08 RP: sorry to be a pain Nov 09 20:29:48 kantlivelong: packages/linux/linux-openzaurus-2.6.xx/defconfig-poodle Nov 09 20:29:58 RP: ah ok i see Nov 09 20:30:22 RP: i want nvram right? Nov 09 20:31:53 kantlivelong: Yes, but keep in mind its at your own risk Nov 09 20:32:13 RP: only risk is user not hardware/software right? Nov 09 20:32:43 Flash the wrong firmware and who knows what could happen... Nov 09 20:32:57 RP: well ive done it before Nov 09 20:33:19 RP: may i ask how to get bitbake to recompile? it just says attempted builds 0 Nov 09 20:33:49 kantlivelong: -c compile -f but you probably want to reconfigure (-c configure -f) Nov 09 20:35:40 RP: ok thanks.. :) Nov 09 20:38:28 03koen 07org.oe.dev * r864008cd... 10/ (1 site/arm-linux-gnueabi): site/arm-linux-gnueabi: add test for abstract sockets Nov 09 20:38:34 03koen 07org.oe.dev * r44673bf7... 10/ (3 files in 3 dirs): dbus 0.95: re-add cross.patch for people that don't have the ac_cv test result in their site files Nov 09 20:54:02 RP: how's the mini-holiday coming along? Nov 09 20:55:07 koen: Its different to work :) Nov 09 20:55:41 koen: Been laying floorboards today ;-) Nov 09 21:00:47 RP: do you still chew pencils? Nov 09 21:02:15 zecke: I'm trying to give it up ;-) Nov 09 21:02:29 RP: I just found myself chewing on one while reading ETSI... Nov 09 21:02:39 BAN PAPER from your office :) Nov 09 21:02:44 No Paper No Pencil Nov 09 21:02:52 right Nov 09 21:03:01 inspect your style after a week Nov 09 21:03:08 styli* Nov 09 21:03:17 I hope you won't start chewing the corners of your laptop Nov 09 21:03:50 * RP tries to resist Nov 09 21:04:19 luckily I'm more in favor of cats than on dogs, otherwise I could imagine chewing the laptop Nov 09 21:05:07 zecke: I had a look at those duplicate methods btw, we can't declare python functions in .inc files :-( Nov 09 21:10:27 RP: oh, there is a bug somewhere in the parse_py as fn is wrong Nov 09 21:10:40 RP: I would say it is possible to define them in .inc files Nov 09 21:10:59 my cat used to steal Palm stylii, and hide them, and chew on them in secret Nov 09 21:11:00 RP: BBHandler is used for .inc and it should know which file it is currently looking at (it should really know) Nov 09 21:11:15 raduga: that is a good excuse for losing styliii Nov 09 21:11:29 raduga: I will use that. There must be a cat at nite in the office stealing all styliiii Nov 09 21:11:30 he leaves the Zaurus' alone though Nov 09 21:11:59 I see zecke is a proud zaurus owner and is pooing on basic latin grammar as a result Nov 09 21:11:59 zecke: his favorite trick now, is to sneak into the medicine cabinet, and raid the q-tips Nov 09 21:12:03 koen: ETSI rocks they define 1 bit sometimes as most sign. bit first Nov 09 21:12:09 and leave one or two of them lying around... everywhere Nov 09 21:12:13 koen: and some times as least signifcant bit.. Nov 09 21:12:27 raduga: hehe Nov 09 21:12:27 in his water dish, in the toilet, in my shoes Nov 09 21:12:27 zecke: I'd agree, we should allow .inc files. We need to find the bug :-/ Nov 09 21:12:39 once, even on my dinner plate ;) Nov 09 21:12:43 RP: defining the methods should work! we do that already Nov 09 21:12:56 RP: the source is just wrongly reported Nov 09 21:13:28 zecke: unlike PDA users... he prefers lighter, flimsier things to chew on Nov 09 21:13:51 so, q-tip is better than stylus :) Nov 09 21:14:21 i guess there are people that like flimsy, lightweight stylii too Nov 09 21:14:24 zecke: I want to clean up these method errors :) Nov 09 21:14:58 zecke: want to have a stab at moving tinderbox this weekend? Nov 09 21:17:11 RP: check why include does not set fn properly ;) Nov 09 21:17:24 RP: maybe add a this_file_name string which points to the current filename Nov 09 21:17:37 RP: I could write a patch you could test? Nov 09 21:18:11 * zecke ditches the OS class tomorrow Nov 09 21:19:15 I'm a moron (I guess) Nov 09 21:20:01 oh no hmm Nov 09 21:22:22 zecke: I'm open to patches :) Nov 09 21:22:42 RP: there is methodpool.insert_method Nov 09 21:22:50 RP: which uses fn as the filename Nov 09 21:28:31 zecke: We seem to use modulename rather than fn? Nov 09 21:29:35 I will assign this bug to me now :) Nov 09 21:36:17 RP: I hate this 'parser' Nov 09 21:37:17 zecke: I looked at it and I felt ill :-/ Nov 09 21:38:04 I always knew phil and chris a true perl hacker :) Nov 09 21:39:18 koen: oe.dev build dbus-glib-0.72-r0 fails. mreimer said you may know the fix/workaround. Nov 09 21:41:28 yep, use angstrom-2007.1 as a distro Nov 09 21:43:12 RP: ~135 I see Nov 09 21:43:20 appweb.inc as basename and fn Nov 09 21:46:11 EEEK! Nov 09 21:46:33 aaaah Nov 09 21:46:46 * koen solves the dbus-daemon troubles Nov 09 21:47:32 RP: Do you want to know the cause? Nov 09 21:47:32 * koen hails rebooting Nov 09 21:47:38 koen: ok, thanks Nov 09 21:53:10 zecke: Go on then... Nov 09 21:56:30 RP: We parse appweb.inc Nov 09 21:56:37 and save the python code in __body__ Nov 09 21:56:43 then we are done with parsing Nov 09 21:56:54 and we check if we are __inpython__ Nov 09 21:57:08 and then we send "" as text to close the code Nov 09 21:57:12 well this does not close the code Nov 09 21:57:26 and __body__ gets filled with more code from the next file ;) Nov 09 21:58:04 sending None instead of "" changes something Nov 09 21:59:13 ARRRGGGG Nov 09 21:59:48 zecke: ick :-/ Nov 09 22:00:00 yes, it is so horrible Nov 09 22:00:05 ERROR: expected string or buffer while parsing /space/hacking/embedded/oe/org.openembedded.dev/packages/appweb/appweb_1.2.0.bb Nov 09 22:00:12 this is what I get now and I have no clue... Nov 09 22:01:45 RP: now if I only would get a proper exception... Nov 09 22:02:14 zecke: Where is the exception blocked? Nov 09 22:02:22 I don't know :) Nov 09 22:02:51 RP: http://rafb.net/paste/results/FRp9Pz64.html Nov 09 22:04:02 zecke: feeder must require a string (none is not accepted by it) Nov 09 22:04:13 it is Nov 09 22:04:23 hmm Nov 09 22:06:06 RP: eek that is so disgusting :( Nov 09 22:06:28 * koen scratches head Nov 09 22:06:35 I abuse the lineno now Nov 09 22:06:50 zecke: This code makes me feel ill :-/ Nov 09 22:07:05 zecke: I note you can get false line numbers with this code :-/ Nov 09 22:07:05 RP: we are returning to that point Nov 09 22:08:19 eek Nov 09 22:08:20 zecke: The python function regexp would appear to accept blank lines Nov 09 22:08:36 anyone know if dpkg is the only provider of start-stop-daemon? Nov 09 22:08:37 RP: http://rafb.net/paste/results/qvsRlu76.html Nov 09 22:08:48 furlongm: sorry, no idea Nov 09 22:09:07 RP: yes but blanks lines are legal in python and needed to separate two methods/classes/objects Nov 09 22:09:32 zecke: Yes, which explains why "" is not going to end python functions... Nov 09 22:09:45 ZECKES_EOF does end lines Nov 09 22:09:59 and I place 100$ koen will be the first to break this amount of lines... Nov 09 22:10:24 zecke: Better still, we factor the EOF part of feeder into a separate function Nov 09 22:10:47 RP: NO! Nov 09 22:11:00 I WON'T SANITIZE THIS CODE, AM I SCREAMING? Nov 09 22:11:18 RP: crash two days in berlin and we will fixup the parser once and for all Nov 09 22:11:41 zecke: Its a tempting offer :) Nov 09 22:11:45 or if your time permits, I could crash in new castle Nov 09 22:12:26 I'd happily show you around but I can't offer accomodation, at least not atm/yet ;-) Nov 09 22:12:47 ah searching for a flat/house as well? Nov 09 22:13:05 yes... Nov 09 22:13:30 I need more space... Nov 09 22:13:50 RP: hehe, move to berlin and join our geek flat sharing community Nov 09 22:14:36 ERROR: Could not include required file classes/nylon_image.bbclass while parsing /space/hacking/embedded/oe/org.openembedded.dev/packages/images/nylon-image-extended.bb Nov 09 22:14:41 ah, hmmmm Nov 09 22:14:50 include, required files, bbclass we need to improve error reporting Nov 09 22:14:53 it's the session-bus being broken Nov 09 22:15:29 zecke: The only word missing in inherit :) Nov 09 22:15:44 zecke: I have fixes for some of these parse errors I can push... Nov 09 22:19:02 RP: http://rafb.net/paste/results/50fnCF81.html Nov 09 22:19:31 that is my final version, but I can't build ATM Nov 09 22:19:45 zecke: Feel free to just commit :) Nov 09 22:19:49 I'll test... Nov 09 22:19:56 RP: please take a close look ;) Nov 09 22:20:43 zecke: What am I looking for? Nov 09 22:20:52 source of illnes Nov 09 22:21:54 03zecke123 * r630 10bitbake/lib/bb/parse/parse_py/BBHandler.py: (log message trimmed) Nov 09 22:21:54 lib/bb/parse/parse_py/BBHandler.py: Commit the parsed method at EOF Nov 09 22:21:54 Make sure to commit the pending __body__ when we are Nov 09 22:21:54 __inpython__. Currently sending '' lead to simply appending Nov 09 22:21:54 it to the body. Nov 09 22:21:54 We create a special LINE Number to indicate the EOF which Nov 09 22:21:56 will lead to closing/comitting the code. Nov 09 22:22:29 okay back to ETSI Nov 09 22:23:04 something is very wrong with dbus >0.92 Nov 09 22:23:09 tmbinc: do you have a clue about mpeg2 ts streams? Nov 09 22:23:16 koen: and <= 1.0 Nov 09 22:23:18 zecke: thanks :) Nov 09 22:25:20 aaaah Nov 09 22:25:28 the env vars don't get setup Nov 09 22:26:15 zecke: i have to :/ Nov 09 22:27:51 tmbinc: hehe, the transport_packet.continuity_counter + payload_unit_start_indicator show me that I have more data? Nov 09 22:28:07 what does "eval `dbus-launch --auto-syntax --exit-with-session`" do? Nov 09 22:28:35 tmbinc: I still have little clue how e.g. the DSM-CC/MPE section fits in 184-pointer_field-1 :} Nov 09 22:28:56 zecke: define "more". .. a transport stream is endless ;) Nov 09 22:29:03 the CC is just a running number Nov 09 22:29:11 to detect missing and duplicate packets Nov 09 22:29:13 zecke: That appears to fix the problem :) Nov 09 22:29:25 tmbinc: yes. E.g. I have a UDP package of size 1000 bytes Nov 09 22:29:39 the PUSI shows that the beginning of a "payload" packet is contained in this TS packet Nov 09 22:29:41 tmbinc: which is inside one DSCM-CC section Nov 09 22:30:01 tmbinc: so in the first section I get the header Nov 09 22:30:05 tmbinc: sorry Nov 09 22:30:07 again Nov 09 22:30:16 hi Nov 09 22:30:19 tmbinc: so in the first ts_packet I'm likely to get the datagram_section Nov 09 22:30:47 tmbinc: and then I see the more ts_packets following on the pid with continuity_counter+1%X Nov 09 22:30:58 MPE is using "section"-syntax Nov 09 22:31:05 so you have sections encapsulated in TS packets Nov 09 22:31:13 tmbinc: and I will just treat it as more data, or will I have to look at the table_id of each TS packet? Nov 09 22:31:22 (each UDP packet is one section) Nov 09 22:31:29 but sections are split across ts packets Nov 09 22:32:08 tmbinc: yes, now I wonder how to determine this 'split'. I know I have more data to go Nov 09 22:32:19 so for example one TS packet might look like | 47 4x xx xx | yy | ... yy bytes of section x-1 | section x ... | section x+1 | Nov 09 22:32:22 tmbinc: but how do I know that the next packet for this PID has the more data Nov 09 22:32:40 so the rule is: "when PUSI is set, the first byte after the 4 byte header is a pointer, otherwise, there is no pointer" Nov 09 22:33:14 tmbinc: okay but will be data[0] == table_id Nov 09 22:33:15 the pointer points to the begin of the first starting section Nov 09 22:33:24 or just the remaining data from the section Nov 09 22:33:24 inside the section, the first byte is the table_id, yes Nov 09 22:33:48 tmbinc: always, even if it is continued (section on more than one ts packet) Nov 09 22:33:52 the usual case is "47 4x xx xx 00 tt ss ss ..." (where tt is tableid, ss ss the size) Nov 09 22:34:10 ngr Nov 09 22:34:21 running everything by hand works Nov 09 22:34:21 when the section is larger than one packet, it will continue in the next packet Nov 09 22:34:30 without PUSI being set Nov 09 22:34:48 so PUSI indicates that a section is *starting* in this packet Nov 09 22:34:48 tmbinc: yes, it will just continue... no special indication besides PUSI=0 Nov 09 22:34:59 yes and no pointer (as PUSI=0) Nov 09 22:35:28 and no table_id as the section is still open.. okay thanks I will give it a try Nov 09 22:35:33 * zecke hacks away on dvbtraffic Nov 09 22:35:53 03rpurdie 07org.oe.dev * r2604f3e9... 10/ (1 packages/images/nylon-image-base.bb): nylon-image-base: Correct the nylon-image class name to remove parsing errors Nov 09 22:35:54 03rpurdie 07org.oe.dev * r3dcc8c26... 10/ (5 files in 2 dirs): gcc: Factor get_gcc_fpu_setting() into a common .inc file (removing method errors) Nov 09 22:35:59 03rpurdie 07org.oe.dev * r72f7d821... 10/ (9 files in 2 dirs): glibc: Remove duplicate python function declarations Nov 09 22:37:23 koen: You can remove the patch to methodpool.py now ;-) Nov 09 22:37:28 RP: does .dev require bitbake trunk now? Nov 09 22:37:35 RP: or change it to an fatal error Nov 09 22:37:47 koen: no, the cleanups apply equally to old and new Nov 09 22:38:01 RP: not to old as old is broken :( Nov 09 22:38:14 zecke: there was once a brilliant poster (available as PDF and paper) from tektronix about MPEG-2 transport stream :/ Nov 09 22:38:15 RP: this is probably a backport candidate Nov 09 22:38:18 but i can't find it anymore Nov 09 22:38:51 and it's out of stock Nov 09 22:39:02 zecke: agreed Nov 09 22:39:02 it had a visual representation of the fields Nov 09 22:39:04 tmbinc: that is not my issue. I just wonder how to make this robust Nov 09 22:39:33 tmbinc: I will have to buffer the data until the Section is complete, and I do not want to go out of memory Nov 09 22:40:00 tmbinc: and I want to append the data the right way (avoid duplicates by looking at the continuity_counter) Nov 09 22:40:02 zecke: btw why don't you want to use the driver for that? Nov 09 22:40:09 what you are doing is a software demux Nov 09 22:40:22 and the driver should already be able to do that, possibly with hardware acceleration Nov 09 22:40:32 (not on dreamboxes, i agree, because we are too lame ;) Nov 09 22:40:42 but it's transparent for the user Nov 09 22:40:46 tmbinc: My qanu USB stick can't Nov 09 22:40:54 is it supported in linux? Nov 09 22:40:58 tmbinc: I would love to set 0x2000 as PID filter but I can't Nov 09 22:41:06 tmbinc: usb dongle... Nov 09 22:41:21 it doesn't work with the Nova-T... Nov 09 22:41:21 do you really need to filter on more than one pid? Nov 09 22:41:22 Gah. One parse error left in task-bootstrap.inc with task-pivotboot.bb :} Nov 09 22:41:39 (agreed, section filtering doesn't work for 0x2000) Nov 09 22:41:52 tmbinc: I thought it should work? Nov 09 22:42:17 you can only get raw TS packets on 0x2000 Nov 09 22:42:19 tmbinc: what I'm doing now is a dvbnet on steoreids without killing the kernel Nov 09 22:42:32 zecke: yes but one pid is usually enough Nov 09 22:42:37 so just filter that pid, and use section filtering Nov 09 22:42:45 if you need more than one pid, open more than one filter Nov 09 22:42:48 tmbinc: not if you run 12 h264 decoders :) Nov 09 22:42:51 if you need more than ~32 pids, you are screwed :) Nov 09 22:42:55 oh, DVB-H? Nov 09 22:42:59 ;) Nov 09 22:43:15 hmm 32 pids could have been enough Nov 09 22:43:28 but still, 12*2 pid filters (especially section) is better than 1x 0x2000 Nov 09 22:43:32 tmbinc: I believed the Nova only allowed ~4 filters Nov 09 22:43:40 which nova? PCI? or USB? Nov 09 22:43:44 PCI Nov 09 22:43:52 03rpurdie 07org.oe.dev * recb8ad71... 10/ (1 packages/tasks/task-pivotboot.bb): task-pivotboot.bb: Use task-bootstrap.inc to avoid method errors Nov 09 22:44:02 tmbinc: well with the qanu device I'm targeting I get the complete TS stream anyway Nov 09 22:44:03 PCI nova use software filtering, and can filter arbitrary Nov 09 22:44:41 tmbinc: I start to understand DVB-{T,H} :) Nov 09 22:44:48 the Nova doesn't know anything about DVB or MPEG, it just busmasters the received data into PCI Nov 09 22:44:55 (thinking that it's video data ;) Nov 09 22:45:02 lol Nov 09 22:45:03 the software filter then picks up the data Nov 09 22:45:15 which is the reason why 0x2000 initially only worked on Novas - because they are so stupid Nov 09 22:45:26 every "intelligent" dvb card usually can't filter 0x2000 without tricks Nov 09 22:45:33 lol Nov 09 22:46:03 I only have dumb hardware, at least it doesn't try to kill me :) Nov 09 22:46:51 tmbinc: now you say that, I have a bad soft demuxer Nov 09 22:47:24 just open a section filter on the specific pid (not 0x2000), and read(), and you have *1* section per time. :) Nov 09 22:49:01 tmbinc: too late, I have four more hours Nov 09 22:49:11 tmbinc: now I can even check if the ts stream is standard compliant Nov 09 22:49:19 tmbinc: and I can read ts streams from file ;) Nov 09 22:49:39 zecke: you could throw them trough the DVR playback device ;) Nov 09 22:49:46 but yes, checking for compliance is a reason Nov 09 22:50:09 03rpurdie * r631 10bitbake/lib/bb/parse/parse_py/ (BBHandler.py ConfHandler.py): Improve errors messages for required vs. inherited files Nov 09 22:50:16 zecke: what are you doing, btw? :) Nov 09 22:50:27 tmbinc: hacking :) Nov 09 22:50:37 sure, but what? :) Nov 09 22:50:57 tmbinc: we have a huge 42" LCD and display 4x3 h264 videos Nov 09 22:51:22 tmbinc: from our internal multicast network. now I wanted to make it show everything on air :) Nov 09 22:51:45 tmbinc: so you end up somewhere with a 42" LCD where you have DVB-H reception.... ;) Nov 09 22:52:05 tmbinc: so what I'm doing is dumping every IP packet from the tsstream through a tun device Nov 09 22:52:59 tmbinc: ah now I remember why didn't investivagate PID filterting further. I would have to look up the PIDs from the PMT/NAT/INT... Nov 09 22:53:22 he cool Nov 09 22:53:43 tmbinc: can dvr do bandwidth limitation? e.g. playback at a specific bandwidth Nov 09 22:53:58 tmbinc: well I have a patch against dvbsnoop doing this for one PID... Nov 09 22:54:07 but it is a hack Nov 09 22:54:27 you can watch the adaption fields Nov 09 22:54:29 and throttle Nov 09 22:54:39 oh Nov 09 22:54:44 sections only -> no adaption fields :/ Nov 09 22:54:49 then it's hard Nov 09 22:57:33 tmbinc: either I will read this section about bandwidth in the ISO MPEG2TS or I will just use a QoS pipe of linux... Nov 09 22:59:02 i've seen hardware which records 188bytes ts packets together with 4byte timestamps Nov 09 22:59:44 I will see such hardware... :} Nov 09 22:59:45 0015 1 p/s 0 kb/s 2 kbit Nov 09 22:59:45 1fff 4902 p/s 899 kb/s 7372 kbit Nov 09 22:59:46 2000 4904 p/s 900 kb/s 7375 kbit Nov 09 22:59:52 I hate T-Systems... Nov 09 23:08:32 03rpurdie 07org.oe.dev * r9ec3e521... 10/ (1 packages/tslib/tslib/usec_fix.patch): tslib: drop old patch Nov 09 23:08:37 03rpurdie 07org.oe.dev * re7fed00e... 10/ (3 files in 3 dirs): openswan: Add patch to workaround install flags causing build failures Nov 09 23:08:43 03rpurdie 07org.oe.dev * r29a0527b... 10/ (1 conf/machine/hx2000.conf): hx2000.conf: Set PREFERRED_PROVIDER for virtual/kernel correctly Nov 09 23:43:09 zecke: could you prod woglinde to apply the msp stuff? Nov 09 23:43:50 koen: didn't he? Nov 09 23:44:03 nope Nov 09 23:51:45 zecke: I've just had a sudden insight into #1294 - we're missing a load of expandKeys calls in bitbake :-( Nov 09 23:55:58 oh, they are quite cheap Nov 09 23:56:59 five cents each? Nov 09 23:57:19 a quarter of a pound of frikandel Nov 09 23:57:57 Anyhow, bedtime for me. 'night all Nov 09 23:58:08 later Nov 09 23:58:25 wtf... EOVERFLOW? Nov 09 23:59:06 'night RP Nov 10 00:13:10 03koen 07org.oe.dev * rfce83af6... 10/ (1 packages/eds/eds-dbus_svn.bb): eds-dbus: merge in some fixes from poky Nov 10 00:13:15 03pfalcon 07org.oe.dev * r858ba3be... 10/ (1 packages/linux/linux-handhelds-2.6_2.6.17-hh3.bb): Nov 10 00:13:15 linux-handhelds-2.6 2.6.17-hh3: Add latest hh.org 2.6.17 tag made before Nov 10 00:13:15 upgrading to 2.6.18. **** ENDING LOGGING AT Fri Nov 10 02:59:56 2006