**** BEGIN LOGGING AT Tue Dec 26 03:00:00 2006 Dec 26 06:01:09 03mwester 07org.oe.dev * r315f8d94... 10/ (1 packages/linux/ixp4xx-kernel_2.6.19.bb): ixp4xx-kernel: bumped SVN to include the dsmg600 fixes for Openslug Dec 26 07:44:12 hi Dec 26 11:52:36 morning folks Dec 26 11:53:05 hey mickey|35 Dec 26 11:53:42 mickey|35: ...464 is your mobile number? Dec 26 11:53:49 yeah, thanks for your SMS! Dec 26 12:05:24 stupid gprs Dec 26 12:09:19 bleh Dec 26 12:09:27 how much sustained bandwidth do you get via gprs? Dec 26 12:09:35 * koen|gprs isn't looking forward to the pile of work waiting at home Dec 26 12:09:47 mickey|35: ~6kbyte/s Dec 26 12:09:57 d'oh Dec 26 12:09:59 that few? Dec 26 12:10:00 can peak at 40kbyte/s Dec 26 12:10:02 s/few/little/ Dec 26 12:10:08 yep Dec 26 12:10:22 bummer, i was thinking about web radio, but seems i can forget that over gprs Dec 26 12:10:25 the biggest bottleneck is with the bluetooth link Dec 26 12:10:47 lots of scripts set up the rfcomm link at 9600baud or something Dec 26 12:10:54 mickey|35: biggest problem there would be packet jitter Dec 26 12:11:00 in theory you should get 160kbyte/s Dec 26 12:11:15 mickey|35: On gprs/edge, there's a fair bit of lag/jitter.. umts improves things slightly, but there's still issues Dec 26 12:11:28 i see. ya can only cure that by prebuffering a lto Dec 26 12:11:34 lot, even Dec 26 12:11:57 Yeah, you'd want 20+sec buffer, in my experience Dec 26 12:12:31 mickey|35: I hope the neo supports PAN as well as DUN Dec 26 12:12:59 piling on layers and layers off ppp is a bit backwards IMO Dec 26 12:13:29 koen: do we need special hardware for it to support PAN profile or is it just software? Dec 26 12:13:36 software Dec 26 12:14:23 'pand --master' or something like that Dec 26 12:15:32 mickey|35: http://bluez.sourceforge.net/contrib/HOWTO-PAN Dec 26 12:15:41 (a bit dated, but still usefull) Dec 26 12:15:58 pand is in now in bluez-utils by default Dec 26 12:18:38 cool. need to take a look Dec 26 12:19:40 and add 'test gtk 2.10.x/pango 1.15.x/cairo 1.3.10' to your TODO as well :) Dec 26 12:24:12 hehe, yeah Dec 26 12:25:30 mickey|35: I also had a chat with Ross and he said sapwood is just like the original 'pixmap' theme engine Dec 26 12:25:40 so it isn't maemo-specific at all Dec 26 12:26:10 so if moko-libs is heavy on pixmaps, sapwood could be a big help Dec 26 12:26:25 (we were discussing the integer pixops patch) Dec 26 12:28:12 koen: ah, excellent. so sapwood is a drop-in for pixmap? Dec 26 12:28:16 * koen|gprs looks at the new gnome-vfs-dbus in poky Dec 26 12:28:44 'yes' Dec 26 12:29:03 you need to start the sapwood server, but it's a fork of the pixmap gtk engine Dec 26 12:29:53 cool, it looks upstream gnome-vfs has the dbus port Dec 26 12:30:13 oh. a client/server architecture to cache the pixmaps process-independent Dec 26 12:30:14 interesting Dec 26 12:31:03 mickey|35: sapwood also caches the the pixmaps after processing, so it should save a zillion scale operations as well Dec 26 12:31:09 cool! Dec 26 12:31:20 let's hope they fixed some alphachannel problems of the pixmap engine as well Dec 26 12:31:23 if not, we need to :/ Dec 26 12:31:41 mickey|35: http://www.openembedded.org/repo/org.openembedded.dev/packages/sapwood/sapwood_svn.bb Dec 26 12:31:47 cool. thanks Dec 26 12:32:28 I forgot to bring a book, so I got bored enough to add some new stuff Dec 26 12:32:35 heh, good Dec 26 12:33:28 did you see the ALP stuff yet? Dec 26 12:33:44 didn't run anything, but i browsed through the source a bit Dec 26 12:33:58 nothing UI related, so I stopped after a quick glance Dec 26 12:34:05 libsqlfs looks pretty interesting Dec 26 12:34:34 yes that's very interesting Dec 26 12:34:40 i'd like to see some benchmarks though Dec 26 12:34:45 could imagine it being _slow_ Dec 26 12:35:23 it's in OE :) Dec 26 12:35:32 excellent :) Dec 26 12:35:43 http://www.openembedded.org/repo/org.openembedded.dev/packages/alp/hiker_0.9.bb Dec 26 12:36:01 I am developing a fondness of nicely autotooled applications Dec 26 12:36:03 I'm most interested in their "optimized" gtk+ Dec 26 12:36:12 koen: guess what, me too! Dec 26 12:36:40 SRC_URI = "blah" ; inherit autotools Dec 26 12:36:48 or even 'inherit gnome' Dec 26 12:37:09 i converted everything qmake in openmoko to autotools, but retained the qmake infrastructure for development purposes Dec 26 12:37:17 best of both worlds Dec 26 12:37:23 yeah Dec 26 12:40:10 I wonder how the 'general public' will like OE when they start developing for openmoko Dec 26 12:40:22 yeah, i wonder about that as well Dec 26 12:40:38 * koen leaves train and walks home Dec 26 12:40:38 later Dec 26 12:43:05 someone have info about where is possible to buy 'openmoko' cells ? Dec 26 12:43:29 on sale early 2007 directly from fic.com.tw Dec 26 12:43:57 ahhh is a fic device !!! Dec 26 12:44:15 cheers hrw|gone1xmas Dec 26 12:44:20 hrw|gone1xmas: thanks for your SMS! Dec 26 13:09:34 morning all Dec 26 13:10:11 hey RP Dec 26 13:11:04 koen: I think I have a handle on the mystery loadkeys problem. Looks like its some kind of cache cohernecy problem... Dec 26 13:11:38 worsened by the faster zlib code? Dec 26 13:11:48 koen: yes, that just shows it up Dec 26 13:12:00 Inserting flush_cache_all() into the jffs2 readpage code makes all the problems disappear Dec 26 13:12:09 which makes me think the zlib code isn't at fault Dec 26 13:12:27 right Dec 26 13:13:18 Italso explains why things work second time around. One context swicth would be enough to flush the caches Dec 26 13:16:57 flush_cache_all() makes it probably a lot slower, right? Dec 26 13:18:23 Yes, although compared to the zlib decompression time, probably not a lot Dec 26 13:19:20 The thing is, I don't think I've found the root of the problem yet. I can add the cache flush before the zlib decompressor is called and the bug disappears Dec 26 13:19:49 So there is something bad in the cache before the decompressor. I just don't know what Dec 26 13:21:04 This is looking like a VFS problem :-/ Dec 26 14:12:47 RP: congrats on that Dec 26 14:15:35 XorA: I'm still not sure what the real problem is though :-/ Dec 26 14:16:12 RP: but at least you now proved there is Dec 26 14:20:07 XorA: Maybe :) Dec 26 14:20:33 RP: some people also reported that it doesn't happen with glibc 2.3 Dec 26 14:21:10 koen: It certainly doesn't occur with poky oabi... Dec 26 14:22:55 hi zecke Dec 26 14:23:16 hey Dec 26 14:23:23 hmm I should copy stuff... now? Dec 26 14:23:50 copy which stuff? Dec 26 14:23:56 BitBake (tagging) Dec 26 14:24:07 if it occurred on glibc 2.3.5 OZ build would be fscked Dec 26 14:24:35 XorA: Yes, poky uses the same toolchain and is also fine Dec 26 14:24:53 zecke: I guess now is as good a time as any Dec 26 14:25:29 RP: if you feel like working on GUI stuff, feel free to move trunk to bitbake-1.8 Dec 26 14:25:53 zecke: you mean 1.9? Dec 26 14:26:15 Current trunk is 1.7? next stable would be 1.8? Dec 26 14:26:23 and trunk would get 1.9 to make GUI work Dec 26 14:26:36 or do you think it is not worth stabilising BitBake Trunk? Dec 26 14:27:21 zecke: I can't see any harm in stablising it as it stands before starting on the GUI work Dec 26 14:27:36 okay let me wake up first Dec 26 14:27:44 zecke: We(I) really do need some pressure to sort out the remaining bugs ;-) Dec 26 14:28:20 zecke: Also, I'm waiting on mickey before I start on the GUI atm ;-) Dec 26 14:28:42 (not that its holding me up as I have a ton of other things to look at too :-/) Dec 26 14:29:03 yeah, I have used the last two days to finish internal stuff as well... Dec 26 14:29:58 zecke: I've been looking at this zlib+jffs2+glibc2[45] random crash bug. Looks like a cache coherency bug :-/ Dec 26 14:46:50 morning Dec 26 14:47:18 RP: eek :( Dec 26 14:47:40 RP: debugging sounds like fun Dec 26 14:48:03 zecke: I know nothing about tracking this kind of problem down :-/ Dec 26 14:48:35 yes, I would be lost as well. I would disable the cache completely (I,D if it is possible) Dec 26 14:48:57 disable zlib compression for jffs2 (if there is a no-op compression) Dec 26 14:48:57 zecke: I've proven that if flush the cache at point X, it all works Dec 26 14:49:33 okay that does sound evil Dec 26 14:49:40 zecke: It appears data from one terminated process is appearing in a different process too... Dec 26 14:50:05 I would unpack the emergency set, setup the vodoo shrine and call Phil/Russell for help :} Dec 26 14:50:53 cachepolicy=uncached or cachepolicy=writeback in CMDLINE? Dec 26 14:51:19 zecke: Filling memory with 0xff in one process then watching libraries fail to link due to memory corruption with random 0xff characters isn't good :-/ Dec 26 14:51:34 koen: Whatever the PXA default is. If you change it, the PXA crashes iirc Dec 26 14:51:59 iirc pxa use writethru Dec 26 14:52:03 * koen tries to remember Dec 26 14:52:20 koen: writethrough would be my guess too Dec 26 14:52:20 RP: argh, aye, aye, aye Dec 26 15:00:00 anyone around this morning? Dec 26 15:02:10 ibot: botmail for mmp: what's the status of 2.6 on h5xxx? Does usb work yet? Dec 26 15:02:27 was just wondering if anyone has successfully gotten lirc in 3.5.4 to work on collie Dec 26 15:02:44 i keep getting segfaults when i try to load the lirc_sir module Dec 26 15:02:56 what is 3.5.4? Dec 26 15:03:01 the release Dec 26 15:03:19 openzaurus 3.5.4 running on collie Dec 26 15:03:34 with the lirc & lirc-modules packages Dec 26 15:05:59 circut: you want #openzaurus Dec 26 15:06:10 idling there as well Dec 26 15:06:24 didnt know if this might be a development thing Dec 26 15:06:30 * XorA presses the on switch on the collie Dec 26 15:06:41 hey XorA Dec 26 15:08:22 collie should die/2.6ify Dec 26 15:08:26 hey koen Dec 26 15:08:32 XorA: I told ssvb you were our mplayer guru Dec 26 15:08:51 koen: ssvb? Dec 26 15:09:09 koen: ok, thanks Dec 26 15:09:18 no 2.6 support for collie? Dec 26 15:09:32 XorA: https://garage.maemo.org/plugins/scmsvn/viewcvs.php/trunk/?root=mplayer <- ssvb's work Dec 26 15:10:02 koen: ah yeah Ive been tracking that on and off, I think they really should be or maybe are the same people currently doing arm optimisations in mainline tree Dec 26 15:11:49 ssvb: are the libmpeg2 optimizations in mplayer svn? Dec 26 15:12:37 koen: yes, there are some libmpeg2 optimizations in maemo mplayer svn, but they were not submitted upstream yet Dec 26 15:13:38 koen: i'm not the author of these libmpeg2 improvements, i only integrated these fixes Dec 26 15:13:50 ssvb: I know Dec 26 15:14:09 ssvb: the original author didn't leave a clean patch, so I never applied that in OE Dec 26 15:14:28 ssvb: koen I have been tracking both lists and if I get time I mean to try and benchmark a random svn build for mpeg4 speed Dec 26 15:15:26 n770 is armv4t? Dec 26 15:16:00 v5te Dec 26 15:16:06 arm926-ejs Dec 26 15:16:33 circut: confirmed, loading the modules oopses the kernel Dec 26 15:17:02 circut: feel free to file a bug on bugs.openembedded.org Dec 26 15:17:15 * RP -> back later Dec 26 15:17:39 XorA: thats fine, collie uses 2.4.18-smth Dec 26 15:18:41 XorA: can we bribe you to convert http://josejx.net/collie/ to C code ? Dec 26 15:19:54 koen: it doesnt need bribes it needs time, I need to do something with Zaurus remote, and my personal life has been hectic since the day I got it Dec 26 15:20:51 * RP knows this feeling... Dec 26 15:22:02 XorA, all the optimizations for armv5te (dsp instructions) were done according to arm926 instructions timings, this code is most likely not optimal for xscale (you may even notice performance degradation in the worst case) Dec 26 15:23:25 XorA, is it possible to run some benchmarks for video decoding in mplayer 1.0rc1 on zaurus (-benchmark option) using different idct Dec 26 15:23:27 ? Dec 26 15:24:09 ssvb: good to know, and I havent tried Dec 26 15:24:40 mplayer needs (something like) liboil Dec 26 15:25:35 XorA: i'll probably post some more information with detailed instructions about benchmarking in oesf.org forum this evening Dec 26 15:26:06 ssvb: cool Dec 26 15:29:29 * XorA can see handcrafted assembler for different incarnations of arm becoming a pain in the arse Dec 26 15:29:35 roll on everything having mx31 Dec 26 15:29:53 yes! Dec 26 15:29:59 or mx31+ Dec 26 15:30:36 ciao Dec 26 15:30:55 koen: what happened to your poppler patch? Dec 26 15:32:04 XorA: didn't apply it yet Dec 26 15:32:09 XorA: need to split it out Dec 26 15:32:34 XorA: and I'm waiting for the irex dudes to submit other patches as well Dec 26 15:35:18 koen: I think mickey|35|family would be very interested in those patches if they produce real speed increase Dec 26 15:35:29 does that mean he has 35 families? Dec 26 15:35:59 XorA: it means he turned 35 and his family is burying him in gifts Dec 26 15:36:52 didnt realise micky was so old :-) Dec 26 15:37:02 sssssht! Dec 26 15:37:15 ;) Dec 26 15:40:41 XorA: I agree with you over 32 it's old ... Dec 26 15:43:00 :-D Dec 26 16:06:16 * mwester thinks "32 is OLD??? Good grief!" Dec 26 16:07:12 * mwester shakes his head and mutters something about "young whippersnappers" :p :D Dec 26 16:08:09 huhu Dec 26 16:08:29 whats interesting is in the newer ages, people tend to not want to become old, but there where times where getting old was good Dec 26 16:13:47 XorA|gone: my next commit will make poppler loads faster Dec 26 16:17:17 hmm what can poppler do ? Dec 26 16:17:37 render pdf documents Dec 26 16:17:41 ah Dec 26 16:19:40 as to say, x11vnc does a much cleaner job , compared to fbvncserver Dec 26 16:22:14 doing better as fbvnc isn't that hard Dec 26 16:22:22 thanks to XorA|gone for that hint Dec 26 16:22:53 koen, fbvnc has a strange input device approach, which never worked for me Dec 26 16:23:01 right Dec 26 16:23:37 XorA told me about x11vnc and now heavens came done to these parts .. Dec 26 16:24:18 i heard the name before but funny enough i did stick to fbvnc and its input hell Dec 26 16:29:22 03koen 07org.oe.dev * rc1c4d0a6... 10/ (4 files in 2 dirs): Dec 26 16:29:22 poppler: cleanups: Dec 26 16:29:22 * create a poppler.inc Dec 26 16:29:22 * create a poppler-fpu.inc to set --enable-fixedpoint when needed Dec 26 16:29:22 * enable zlib Dec 26 16:29:50 happy xmas around from saxonia... Dec 26 16:32:50 greentux: likewise Dec 26 16:33:04 guys i need and info, exist a tool to graph che calling graph of functions in a project (set of sources) ??? Dec 26 16:33:50 yo Dec 26 16:38:16 gremlin[it]: oprofile? Dec 26 16:39:04 it work with sources ?! Dec 26 16:39:42 is a kernel modules ... i'm not able to (easyly) figure out which functiona are API and which are internal :( ... Dec 26 16:47:22 03koen 07org.oe.dev * r885825fa... 10/ (1 packages/evince/evince_0.6.1.bb): evince: add missing DEPENDS on libxt Dec 26 16:53:22 XorA|gone: https://bugs.freedesktop.org/show_bug.cgi?id=7808 and https://bugs.freedesktop.org/show_bug.cgi?id=7910 Dec 26 17:00:34 03koen 07org.oe.dev * rf3f3d16d... 10/ (1 packages/intltool/intltool_0.35.2.bb): intltool: update to 0.35.2 Dec 26 17:03:42 hi all Dec 26 17:05:19 hey mr_nice Dec 26 17:08:03 hi koen Dec 26 17:10:55 koen: I am bit ot but I think you are the right person to ask - it seems that the angstroem distribution has dropd the simpad support is that right? Dec 26 17:15:53 mr_nice: gcc doesn't support strongarm yet for EABI Dec 26 17:16:12 mr_nice: but I have been thinking about doing an OABI feed for strongarm machines Dec 26 17:16:26 like collie, simpad, h3800, etc Dec 26 17:16:51 koen: thx for the info Dec 26 17:17:31 * koen -> food Dec 26 17:26:37 * RP returns Dec 26 17:29:54 today`s pull now came up for some strange error Dec 26 17:30:13 Sorry, the Familiar Linux... blablabla ?? Dec 26 18:21:01 koen: nice poppler improvements, we really should start tracking this stuff that should go into OE but isnt yet Dec 26 18:21:43 XorA|gone: not all patches apply cleanly, that's why they are not in yet Dec 26 18:22:42 XorA|gone: I sent the same links to the irex dudes to have them do benchmarks Dec 26 18:23:29 XorA|gone: benchmarking between the 'splash' and 'cairo' backend would be usefull as well Dec 26 18:24:30 * koen needs way more time Dec 26 18:25:02 :-D Dec 26 18:25:37 koen: Ill have to start noting them as sometime I get a spare 5 mins, ideal for massaging patches, not ideal for dev work Dec 26 18:26:18 the first patch applies as-is, the second bug entry has 3 patches of which the second one fails with 3 hunks Dec 26 18:26:37 :-D always the way Dec 26 18:27:06 the patches are from http://blog.kowalczyk.info/software/sumatrapdf/develop.html Dec 26 18:27:23 that dude now has commit access to poppler cvs, but lacks the time to commit it Dec 26 18:27:32 according to his posts on the poppler list Dec 26 18:28:27 * koen so wants a http://www.irextechnologies.com/products/iliad/specs Dec 26 18:40:23 koen: The display on the iliad is ratehr cool :) Dec 26 18:41:11 I know that a quick hack 2.6 kernel port would be straighforward too... Dec 26 18:58:05 RP: irex has a 2.6 port, but they lack a driver for the flash Dec 26 18:58:33 koen: Ah, things have changed since I last looked then... Dec 26 18:58:42 not a public port Dec 26 18:58:51 RP: I'm persuading them to attend fosdem Dec 26 18:59:40 they are using OE and want to release they patches soon(TM) Dec 26 18:59:53 where their 'soon' differs from mine Dec 26 18:59:59 indeed :-/ Dec 26 19:00:10 The GPL doesn't use soon in its terms Dec 26 19:00:18 FOSDEM could be interesting... Dec 26 19:00:21 * RP -> food Dec 26 19:00:28 ~bon apetit Dec 26 19:00:35 ~bon appetit Dec 26 19:00:37 i heard bon appetit is smacznego. Guten Appetit. Eet Smakelijk. God Appetitt. Buon Appetito. Buen apetito Bom Apetite. buen apetito Dec 26 19:02:15 Приятного аппетита. also ;-) Dec 26 19:06:02 bd2: русский? :) Dec 26 19:06:29 yup ;-) Dec 26 19:06:44 me 2 ;) Dec 26 19:07:12 nice to meet Dec 26 19:07:51 yes, I think you are the first russian I met here Dec 26 19:08:20 brb, food :> Dec 26 19:17:17 dobryj wjetschir Dec 26 19:23:07 hehe.. good evening ;-) Dec 26 19:31:07 re Dec 26 19:33:04 hey greentux :) Dec 26 19:55:48 hey Jin^eLD: gdje ty zhivesh? Dec 26 19:58:24 * chouimat|ibook is away: nap Dec 26 19:58:54 w awstrii, a tak rodom iz moskwy :> Dec 26 19:59:01 greentux: where are you from? Dec 26 19:59:27 saxonia, germany Dec 26 19:59:39 <- vienna, .at Dec 26 20:00:35 panimaju :) Dec 26 20:52:06 03rwhitby 07org.oe.dev * r2710d036... 10/ (5 files in 4 dirs): slugos: Added NAS100d image support Dec 26 21:06:06 Which version of monotone do I need for OE? Dec 26 21:06:11 minimum, that is? Dec 26 21:09:00 sorry, nevermind, just found it in documentation. Dec 26 21:11:34 http://www.openembedded.org/repo/org.openembedded.dev/packages/xorg-xserver/xserver-kdrive/no-serial-probing.patch Dec 26 21:12:11 can this patch also prevent kdrive to be pointed to a symlink of ttyS0 ? Dec 26 21:18:18 rob_w: probably Dec 26 21:18:41 rob_w: recompile kdrive with -DBREAK_MY_SERIAL_CONSOLE or without that patch Dec 26 21:20:18 it locks up so strangely Dec 26 21:23:43 rob_w: http://www.openembedded.org/bonsai/view/rev/12819/ Dec 26 21:25:42 koen, too bad , the bug # link is dead on this site Dec 26 21:25:58 works over here Dec 26 21:26:59 slow, but it works Dec 26 21:27:06 slow , eh Dec 26 21:27:27 feels like it is far away in the galaxy Dec 26 21:28:11 "xserver-kdrive messes up the serial console when probing for mice on ttyS[01]" Dec 26 21:30:28 maybe i am messing up something too Dec 26 21:40:07 ok iam recompiling .. lets see Dec 26 22:05:05 it still locks up when pointing tslib to /dev/ttyS0 .. Dec 26 22:31:03 * chouimat|ibook is away: Gone away for now. Dec 26 22:31:13 * chouimat|ibook is back. Dec 26 22:33:03 NOTE: build 200612261452: completed Dec 26 22:33:06 bloomin eck Dec 26 23:08:22 hey Bernardo **** ENDING LOGGING AT Wed Dec 27 03:00:00 2006