**** BEGIN LOGGING AT Fri Nov 17 02:59:57 2006 Nov 17 07:41:38 hi Nov 17 08:30:10 good morning all Nov 17 08:58:23 morning all Nov 17 08:58:34 hey RP Nov 17 08:58:55 RP: Jorn's benchmarks just got mentioned on the cairolist :) Nov 17 08:59:21 http://lists.freedesktop.org/archives/cairo/2006-November/008438.html Nov 17 09:01:52 morning Nov 17 09:02:02 hey hrw|work Nov 17 09:03:07 koen: yay :) Nov 17 09:08:24 good morning Nov 17 09:16:02 http://www.linuxdevices.com/news/NS5744003201.html Nov 17 09:16:48 350$ for i.mx Nov 17 09:16:56 imx21 Nov 17 09:17:32 [10:30] suihkulokki: confirmed.. zune has a MX31 - 533Mhz arm11 Nov 17 09:17:32 [10:31] suihkulokki: I guess they are scratching their head how apple can make a mp3 player with a 80(?)Mhz arm7.. Nov 17 09:18:02 zune has a mx31 , hum interessting :( Nov 17 09:18:55 "ARM9-based i.MX31 SoC" Nov 17 09:18:57 milo looks like nice hacking device.. 1G flash, keyboard, wifi Nov 17 09:19:01 http://lists.openmoko.org/pipermail/community/2006-November/000081.html is a nice read as well Nov 17 09:19:02 i believe mx31 is arm11 Nov 17 09:19:19 Genesis: linuxdevices.com people are dyslectic Nov 17 09:19:23 oki :) Nov 17 09:19:32 Genesis: they keep writing 'GPE' with a 'Q' Nov 17 09:20:02 lol they put a mx21 block diagram Nov 17 09:20:16 t's a mx21 oki Nov 17 09:20:27 that makes a bit difference lol Nov 17 09:20:48 koen, : ouch :( Nov 17 09:21:26 I need to subscribe there Nov 17 09:21:51 i work on xinerama in kdrive , someone has already test that ? i think i need xorgconf Nov 17 09:22:23 --enable-xorgcfg Nov 17 09:23:10 The best quote from rwhitby: "and you get the same thing on SlugOS/OpenEmbedded, but you had it six Nov 17 09:23:10 months earlier than you could do it on Debian." Nov 17 09:24:15 it's optimistic Nov 17 09:25:21 RP: didn't poky have a modified version of the vincent OGL-es lib? Nov 17 09:28:20 I know someone who ported vincent to 770, but according to him the upstream never did anything with the patches he sent Nov 17 09:30:08 koen: http://ogl-es.svn.sourceforge.net/viewvc/ogl-es/branches/linux_nokia_770/ogles/ Nov 17 09:31:14 Packaged contents of vincent into /data/build/koen/OE/build/tmp/angstrom/deploy/glibc/ipk/libvincent0_svn-r0_armv5te.ipk Nov 17 09:31:20 I've built that branch Nov 17 09:34:04 http://svn.o-hand.com/view/misc/trunk/ogles/README?rev=119&view=auto Nov 17 09:59:45 morning all Nov 17 10:04:25 hey ade|desk Nov 17 10:15:14 mallum: good morning! Nov 17 10:16:37 hey koen Nov 17 10:17:00 anyone else seeing build problems during the do_rootfs stage of bulding slugos-image-1.0-r38 (trying to build "debianslug"-image) ... i am getting two missing packages in the log file ... Cannot find package kernel-module-nfs. and Cannot find package kernel-module-firmware-class. Nov 17 10:17:01 mallum: are you still working on the ogl-es stuff? Nov 17 10:17:35 koen: yes, but not too much time lately Nov 17 10:17:36 Caelian|work: didn't rwhitby fix that last night? Nov 17 10:18:24 mallum: someone added a recipe for n770 branch of vincent svn, and I'm debating wether to add that or ogl-es from poky Nov 17 10:18:28 koen, i saw a commit that claimed something similar to that nature ... but i just updated and the requirements are still in the openembedded/conf/{debianslug|openslug}.conf files Nov 17 10:18:41 koen Nov 17 10:18:49 koen: the poky one is from that also Nov 17 10:18:58 openembedded/conf/distro/ i meant Nov 17 10:19:01 Caelian|work: it's a bug anyway to RDEPEND on kernel modules, it should RRECOMMEND them Nov 17 10:19:06 mallum: I know Nov 17 10:19:10 koen: I just tweaked it a little Nov 17 10:19:11 mallum: the readme states that :) Nov 17 10:19:25 :) Nov 17 10:19:48 o-hand svn is 5 months more recent that the n770 branch Nov 17 10:22:10 are they based on same code or different ports? Nov 17 10:30:49 koen: looks like eabi on uclibc is getting some attention, well ARM in general, even paul brooks is getting in on the act w.r.t thumb too Nov 17 11:15:04 quiet day? Nov 17 11:22:12 ade|desk: cool Nov 17 11:24:50 hm. time to change names of imap folders.. inbox/mailing lists/zaurus/{oe, oz, angstrom, linux kernel} need remove of zaurus Nov 17 11:35:26 greetings ! Nov 17 11:36:14 could somene please check if the svn for owmnr is working ? svn co https://svn.digital-opsis.com/public/owmnr Nov 17 11:39:21 Ifaistos: it works Nov 17 11:40:15 koen : Thanks ! i moved it in the local svn in the office, and was hard to check as i am on the same router Nov 17 11:47:29 heh.. another hard to fetch by automat source Nov 17 11:47:45 svn fetcher hangs on https Nov 17 11:48:27 svn: PROPFIND z '/public/owmnr': Server certificate verification failed: certificate has expired, certificate issued for a different hostname, issuer is not trusted (https://svn.digital-opsis.com) Nov 17 11:49:35 btw - someone know office machine (printer/scanner in one) working good with linux? Nov 17 11:52:58 hey likewise Nov 17 11:55:02 koen: hey Nov 17 11:55:04 hi all Nov 17 11:55:19 RP: the linux-nokia770 recipe you added to poky is bad :( Nov 17 11:55:32 RP: merge the one from OE to poky :) Nov 17 11:56:12 huhu Nov 17 11:58:14 koen: bad in what way? Nov 17 11:58:40 RP: no eabi switch, not using DEPLOY_DIR_IMAGE, starting at r4, etc Nov 17 11:59:01 improper COMPATIBLE_HOST Nov 17 11:59:29 koen: Ah, right. It was just a quick hack to get something working for someone ;-) Nov 17 11:59:30 I think I didn't touch SRC_URI, DESCRIPTION and LICENSE Nov 17 12:00:40 koen: The one in OE won't build - KERNEL_IMAGETYPE is wrong ;-) Nov 17 12:01:15 in nokia770.conf, right? Nov 17 12:01:34 koen: No, in the .bb file :-/ Nov 17 12:02:01 koen: I'll sync a hybrid Nov 17 12:02:13 (a version bump to 2.6.16 too) Nov 17 12:02:34 I'll sync nokia770.conf Nov 17 12:02:40 RP: did you use this: http://ogl-es.svn.sourceforge.net/viewvc/ogl-es/branches/linux_nokia_770/ogles/ as basis for the poky ogles port? Nov 17 12:02:57 koen: ah, CIA is dead and I didn't see the commit... Nov 17 12:03:16 suihkulokki: I think mallum did Nov 17 12:04:09 ok Nov 17 12:07:42 * koen kicks CIA-4 Nov 17 12:07:42 ow Nov 17 12:24:28 * NAiL kicks CIA-* Nov 17 12:42:14 RP: "install -m 0644 ${WORKDIR}/defconfig ${S}/.config" needs to get removed from the n770 recipe as well Nov 17 12:52:38 hey all Nov 17 12:56:43 hey lrg Nov 17 13:00:32 lrg: did you hear that the zune is using a i.mx31 as well? Nov 17 13:00:59 koen: sounds like overkill to me Nov 17 13:01:12 koen: yep Nov 17 13:01:22 IvoC: ipods run on a ~100MHz arm7 SoC Nov 17 13:01:36 IvoC: same screen resolution Nov 17 13:14:23 lrg: you there ? Nov 17 13:14:35 Spyro: I just had to ask lrg about you ;D Nov 17 13:14:56 Spyro: yep I here now :) Nov 17 13:15:01 I'm Nov 17 13:15:38 hi Nov 17 13:15:41 remember me ? Nov 17 13:15:53 Spyro: yep, how are things ? Nov 17 13:15:57 not bad... Nov 17 13:16:04 Im working on the eseries port again Nov 17 13:16:18 cool Nov 17 13:16:19 come to the point of using asoc for the touchscreen Nov 17 13:16:25 and theres no wm9705 driver! Nov 17 13:16:44 Spyro: my laziness Nov 17 13:16:50 lrg: I'll write it Nov 17 13:16:55 just gimme specs :) Nov 17 13:17:24 Spyro: Aren't they on the Wolfson website? Nov 17 13:17:36 RP: dunno, are they ? Nov 17 13:17:46 Spyro: yep they are Nov 17 13:17:55 sweet Nov 17 13:18:51 where? (being dense here) Nov 17 13:18:53 Spyro: I didn't add it initially because it didn't have the fine grain PM control like the WM9712 (on tosa) and I'm very busy these days :-/ Nov 17 13:18:59 * lrg looks Nov 17 13:20:22 Spyro: http://wolfsonmicro.com/products/WM9705/ Nov 17 13:20:29 koen: Better still, that kernel .bb needs to rm ${S}/.config Nov 17 13:21:06 * lrg -> coffee Nov 17 13:21:21 ta Nov 17 13:24:26 Spyro; thanks Nov 17 14:11:53 morning Nov 17 14:13:45 hi chouimat Nov 17 14:16:47 hey pH5 Nov 17 14:16:55 hey koen Nov 17 14:17:55 I just had a problem with kernel module dependencies again. Nov 17 14:18:15 module-init-tools-cross 3.2.2 didn't install ...-depmod-2.6, only ...-depmod Nov 17 14:18:40 but kernel.bbclass wants to run depmod-2.6 Nov 17 14:19:58 "ln -s arm-angstrom-linux-gnueabi-depmod arm-angstrom-linux-gnueabi-depmod-2.6" solved the problem - is this a bug in module-init-tools-cross_3.2.2 or in kernel.bbclass? Nov 17 14:20:10 * koen has no idea Nov 17 14:20:31 * pH5 neither Nov 17 14:21:28 I assume this has to do something with builds for machines that have (had?) both 2.4 and 2.6 kernels. Nov 17 14:21:47 I guess Nov 17 14:23:16 #524 in bugtracker Nov 17 14:24:57 hi Philip Nov 17 14:25:06 gm Nov 17 14:26:07 hrw|work: thanks, so it's module-init-tools-cross' fault Nov 17 14:26:20 strange, the "mv ${bindir}/${TARGET_PREFIX}depmod ${bindir}/${TARGET_PREFIX}depmod-2.6" is in the recipe. I wonder why that didn't work.. Nov 17 14:51:56 I wait for a day when zecke will discover 'no wrap lines' option in mailer Nov 17 14:52:20 haha Nov 17 15:00:15 ~change 158 eur to pln Nov 17 15:00:20 158.00 Euro (EUR) makes 601.699 Zloty (PLN) (from http://www.xe.com/) Nov 17 15:00:38 ~change 158 eur to cad Nov 17 15:00:44 158.00 Euro (EUR) makes 231.904 Dollar (CAD) (from http://www.xe.com/) Nov 17 15:04:49 ~change 2200 eur to rur Nov 17 15:04:54 2,200.00 Euro (EUR) makes 707,538.42 Dollar (ZWN) (from http://www.xe.com/) Nov 17 15:05:05 ~change 2200 eur to rub Nov 17 15:05:10 2,200.00 Euro (EUR) makes 6,014,271.49 Ruble (BYR) (from http://www.xe.com/) Nov 17 15:07:59 russia need denomination Nov 17 15:08:23 :) Nov 17 15:08:34 ~change 1 cad to rub Nov 17 15:08:39 1.00 Dollar (CAD) makes 1,862.75 Ruble (BYR) (from http://www.xe.com/) Nov 17 15:08:44 nice Nov 17 15:14:58 ~change 599 cad to usd Nov 17 15:15:04 599.00 Dollar (CAD) makes 523.876 United States Dollar (USD) (from http://www.xe.com/) Nov 17 15:15:19 ~change 1 usd to gbp Nov 17 15:15:24 1.00 United States Dollar (USD) makes 0.527510 Pound (GBP) (from http://www.xe.com/) Nov 17 15:21:45 wow, is that all? Nov 17 15:22:19 :) Nov 17 15:39:30 hi Nov 17 15:42:13 hrw|work: How did the fetchall builds end up? Nov 17 15:42:18 hi sirfred Nov 17 15:43:00 RP: good question - I didnt checked today Nov 17 16:17:43 hrw|work: hey, I have the image now, created following your instructions yesterday Nov 17 16:17:49 hrw|work: but I would like to trim it down somehow Nov 17 16:18:23 it has a lot of stuff that is not really needed; and also, I would probably need to customize the stuff in etc, some settings for the board seem to be required Nov 17 16:18:30 * NAiL pokes CIA-4 Nov 17 16:18:42 Jin^eLD: what you want to remove? Nov 17 16:18:56 hrw|work: sshd for example Nov 17 16:19:02 Jin^eLD: 1. remove unneeded. 2. polish confs Nov 17 16:19:22 Jin^eLD: something more? Nov 17 16:19:32 Jin^eLD: dropbear is by default - you have to edit task-base Nov 17 16:19:32 hrw|work: which confs? I did not specify ssh to be added anywhere... Nov 17 16:19:37 aah, ok.. Nov 17 16:19:49 Jin^eLD: 2. /etc/ == confs Nov 17 16:19:50 well, I did not look at the rest yet - my problem right now is that the image does not work with the board Nov 17 16:20:09 oh, so "confs" is not "various configs" but a dedicated file Nov 17 16:20:09 I see :) Nov 17 16:26:18 * cbrake observes that the MMC driver in 2.6.16 must not be preemption safe. Nov 17 16:33:09 cu Nov 17 16:43:59 a+ Nov 17 16:47:53 kergoth: good morning :) Nov 17 16:47:55 hey Nov 17 16:48:43 hi hergoth Nov 17 16:48:46 kergoth Nov 17 16:48:50 hi chouimat Nov 17 16:52:37 lrg: you here ? Nov 17 16:55:03 Spyro: yep Nov 17 16:56:59 lrg: is it worth adding wm9705 to the wm9712 driver in you opinion ? Nov 17 16:57:18 or should I just create another one based on wm9712 ? Nov 17 16:57:30 Spyro: I would create a new one. Nov 17 16:57:44 I didnt think they were that different... Nov 17 16:58:01 In terms of PM and audio paths they are Nov 17 16:58:05 right Nov 17 16:58:22 is the basic AC97 stuff handled in a common core ? Nov 17 16:58:28 yes Nov 17 16:58:30 cool Nov 17 16:58:54 Any objections to changing to CACHE = "${TMPDIR}/cache/${MACHINE}" in bitbake.conf? Nov 17 16:59:24 Spyro, RP: I've now got some hosting... http://opensource.wolfsonmicro.com/ Nov 17 16:59:34 The starts when you then put MACHINE ?= "x" in local.conf :) Nov 17 16:59:42 magic starts... Nov 17 16:59:50 lrg: Yay :) Nov 17 17:00:19 RP: yay, about 10 months late, but it now exists :) Nov 17 17:00:25 lrg: cool Nov 17 17:00:48 lrg: You got there in the end though :) Nov 17 17:00:49 the reasme in the asoc folder says 'wm9705 ts' - whats that ? Nov 17 17:00:57 *readme Nov 17 17:01:05 touch screen Nov 17 17:01:25 RP: we also have an OE mirror running as well :) Nov 17 17:01:27 touchscreen driver but no ac97 driver ? Nov 17 17:02:21 lrg: Things are coming together :) Nov 17 17:02:23 Spyro: not sure what you mean Nov 17 17:02:34 RP: at last :) Nov 17 17:02:43 lrg: theres no wm9705 C97 driver is there ? Nov 17 17:02:49 C97 = AC97 Nov 17 17:03:05 lrg: Its the same here. I'm pulling various bits of poky together, finally :) Nov 17 17:03:14 RP: :) Nov 17 17:03:46 lrg: what use is the wm9705 ts driver without a wm9705 AC97 driver? Nov 17 17:03:49 Spyro: the touch drivers are separate, thay attach to ASoC via the AC97 bus Nov 17 17:04:07 right but how can anyone have tested the wm9705 ts driver ? Nov 17 17:04:23 Spyro: it can also attach to the generic AC97 Nov 17 17:04:28 oh cool... Nov 17 17:04:32 *tries this Nov 17 17:08:24 so what is 'SoC audio for intel PXA2xx' ? Nov 17 17:08:24 the PXA AC97 driver? Nov 17 17:08:24 does that replace 'AC97 driver for the intel PXA2xx chip' ? Nov 17 17:08:29 Spyro: yep Nov 17 17:08:36 ok Nov 17 17:09:10 so I need that plus the WM9705 AC97 ts driver and I get touchscreen support (but no audio) Nov 17 17:09:22 Spyro: yes Nov 17 17:09:26 zecke, i have fixed my parser problems, but won't be able to commit until Monday Nov 17 17:10:22 Spyro: a good place to look is at the tosa code. It will be similar config to yours Nov 17 17:10:27 cool Nov 17 17:10:43 (yes I gatheredthat, similar chipset) Nov 17 17:13:54 * lrg -> pub Nov 17 17:16:47 lrg: take care Nov 17 17:17:45 lrg: Enjoy :) Nov 17 17:17:51 * RP will be copying shortly :) Nov 17 17:18:35 03justinp 07org.oe.oz354x * rf35709c1... 10/ (3 files in 2 dirs): emotion: remove sed line causing build problems Nov 17 17:18:47 03justinp 07org.oe.oz354x * r7d818438... 10/ (1 classes/efl.bbclass): efl.bbclass: add includedir to -dev package Nov 17 17:19:43 03justinp 07org.oe.dev * r7d8c3952... 10/ (4 files in 4 dirs): evidence: move to noworking, it has not been updated in a long time Nov 17 17:20:34 03rwhitby 07org.oe.dev * ra3369ab4... 10/ (1 packages/images/slugos-image.bb): slugos-image: Removed kernel modules that are now compiled in Nov 17 17:20:43 03lenehan 07org.oe.dev * r88e0d852... 10/ (3 files in 3 dirs): Nov 17 17:20:57 ah, cia wakes up Nov 17 17:21:34 cherokee: Add an initscript for the cherokee http server so it'll start on Nov 17 17:21:43 boot, so it'll be stopped when removing it etc. Nov 17 17:22:34 03lenehan 07org.oe.dev * reb1aa548... 10/ (1 packages/cherokee/cherokee_0.5.3.bb): Nov 17 17:23:34 cherokee 0.5.3: Specify the configuration files so that local changes Nov 17 17:23:39 don't get wiped out during an upgrade. Nov 17 17:24:34 03lenehan 07org.oe.dev * r08136d50... 10/ (1 packages/cherokee/cherokee_0.5.3.bb): cherokee 0.5.3: Pass through oe-stylize.py. Nov 17 17:25:34 03koen 07org.oe.dev * r7a331798... 10/ (6 files in 4 dirs): linux-nokia770: add 2.6.16, from poky, but enhanced greatly by me :) Nov 17 17:25:39 03koen 07org.oe.dev * rd2013714... 10/ (1 conf/machine/nokia770.conf): nokia770.conf: add armv4t as extra ipkg arch Nov 17 17:27:16 03koen 07org.oe.dev * r24729db7... 10/ (1 packages/linux/linux-nokia770_2.6.16-osso15.bb): linux-nokia770: fix defconfig install Nov 17 17:27:22 03zecke123 * r677 10bitbake-parser/ (ast-test.py ast.py parse-one.py): Add the test cases I use... Nov 17 17:27:29 03koen 07org.oe.dev * rfeb7ef88... 10/ (1 packages/linux/linux-nokia770_2.6.16-osso15.bb): linux-nokia770: remove old config before adding a new one Nov 17 17:29:38 03zecke123 * r678 10bitbake-parser/ (. ast.py ast_defines.py evaluator_data.py): Nov 17 17:29:44 -AST Defines hold the Node number Nov 17 17:29:51 -evaluator_data will evaluate the AST Nov 17 17:30:34 -now use a mixin to 'inject' the eval method Nov 17 17:30:35 (35 lines omitted) Nov 17 17:34:06 hey mreimer Nov 17 17:34:11 hey koen Nov 17 17:34:25 mreimer: haven't noticed any problems with cairo yet Nov 17 17:34:46 koen: good. neither have I. did you recompile gtk 2.10.6 without the pangoxft patch? Nov 17 17:35:04 koen: and does it feel any faster to you? Nov 17 17:35:08 no, since I assume xft will always be faster as cairo Nov 17 18:45:50 hi fooks: is this any interesting? http://www.maxfair.com.tw/products/pdf/TS101_Comparison_of_New_Competitors.pdf Nov 17 18:46:15 hey likewise Nov 17 18:46:33 Jin^eLD: hey! Nov 17 18:48:04 are those all nas devices? Nov 17 18:50:05 yes Nov 17 18:50:28 I was looking if the RDC 2881 devices run Linux, but they don't AFAIK Nov 17 18:50:38 do you know ReadyNAS? Nov 17 18:50:44 no Nov 17 18:52:01 http://www.readynas.de/index_e.html Nov 17 18:52:14 they were shipping with my upnp server, but that was about a year ago Nov 17 18:52:34 I think they have Twonky now, we did not yet release a newer version :) Nov 17 18:52:40 but afaik it is quite powerful Nov 17 18:54:02 what exactly does your upnp server provide? I'm interested in adding upnp support to a device of ours, but I do not know enough about upnp Nov 17 18:54:52 likewise: UPnP is a standard where devices can find each other and figure out their capabilities and so on; when you have a UPnP mediaserver you also need a UPnP mediarenderer Nov 17 18:54:57 those two work together.. Nov 17 18:55:12 the server has your content, it organizes it in a certain way Nov 17 18:55:23 parses metadata, id3 tags, thumbnails, whatever Nov 17 18:55:48 it can also do transcoding and so on, actually there are tons of cool things that can be done; but the whole purpose is - provide content via a standardized interface Nov 17 18:56:05 so when you have a mediarenderer implementation, you can browse stuff on the server and stream it Nov 17 18:56:55 in addition, when you have control point applications, you can control the devices over the net, for example saying: ok, this renderer - play this stuff from that server Nov 17 18:58:19 likewise: what device are you working on? Nov 17 18:59:04 Jin^eLD: wow... Nov 17 18:59:24 Jin^eLD: We have a PVR-alike device that records up to one year of a television channel Nov 17 18:59:43 one year?! Nov 17 18:59:50 :-) Nov 17 18:59:56 add solar collectors and drop those devices over some islands Nov 17 18:59:56 :) Nov 17 19:00:03 4x3 pixels ;) Nov 17 19:00:11 heh Nov 17 19:00:21 koen: almost right :-) Nov 17 19:00:23 likewise: is it capable of playback or does it just record the media Nov 17 19:00:41 ? Nov 17 19:00:42 video is downscaled to a quarter, i.e. 368x288 Nov 17 19:01:14 Jin^eLD: playback as well, and HTTP streaming. RTSP/RTP streaming is planned Nov 17 19:01:49 well. actually you could do both on that kind of device - a UPnP server and a UPnP MR implementation Nov 17 19:01:57 MR is cool because you could stream from other servers then Nov 17 19:02:07 likewise: HD capable? Nov 17 19:02:12 server is nice if you want to distribute the recorded stuff via your home network Nov 17 19:02:20 When the London MTV building was on fire (yesterday) we tried logging into the recorders but the power was already dead. Nov 17 19:02:39 koen: the new ones will be yes, we are designing a PCIe card for that. Nov 17 19:03:24 * koen doesn't have any HD capable devices anyway Nov 17 19:03:28 * likewise neither Nov 17 19:03:48 my TV can do this half HD thing as far as I know, but I do not have a player yet Nov 17 19:04:04 Jin^eLD: 720p? Nov 17 19:04:11 koen: I think so Nov 17 19:04:22 I never tried any HD stuff Nov 17 19:04:23 my tv is a 19" CRT Nov 17 19:04:28 and aging (badly) Nov 17 19:04:41 720p will probably be the best HD standard. There is word that broadcaster want to go from 1080i to 720p to get rid of "i"... Nov 17 19:04:58 heh Nov 17 19:05:02 go for 1080p :) Nov 17 19:05:15 koen: yep Nov 17 19:06:22 likewise: well, about your PVR... adding a UPnP server would be very easy I think Nov 17 19:06:24 koen: they want to stash as many channels as possible in the satellite uplinks (38 Mbit/s or such) Nov 17 19:06:51 Jin^eLD: are you coming to FOSDEM by any chance? :-) Nov 17 19:06:54 ~10 channels? Nov 17 19:07:04 likewise: I do not think so :) Nov 17 19:07:08 HDTV with H.264 ~= 7 Mbit/s Nov 17 19:07:30 Currently HDTV with MPEG-2 ~= 12 Mbit/s Nov 17 19:08:17 so either 5 or 3 channels per uplink Nov 17 19:08:17 I saw a funny presentation on 3dtv a while back Nov 17 19:08:17 philips is having problems because the '3d' is a depthmap Nov 17 19:08:24 the depthmap and regular picture get h264 compressed independantly Nov 17 19:08:28 Jin^eLD: too bad. I would like to talk about the UPnP idea. Nov 17 19:08:38 likewise: where is FOSDEM? Nov 17 19:08:41 .be Nov 17 19:08:48 .be(lgium) Nov 17 19:08:58 koen: i misread that as 'deathtrap' Nov 17 19:08:58 and when is it? Nov 17 19:09:15 Jin^eLD: feb'07 Nov 17 19:10:02 I'll think about it, I have a couple of friends in .be Nov 17 19:10:19 but right now I have no idea what will be going on in february :) Nov 17 19:10:22 http://fosdem.org/2007/ Nov 17 19:10:23 * suihkulokki wants to fosdem.. but needs a good excuse Nov 17 19:10:40 suihkulokki: OE shirt? Nov 17 19:10:46 :-) Nov 17 19:11:23 koen: you think that will convince my boss will pay the trip there? ;) Nov 17 19:11:32 anyone knows the "Brecis 2600 170Mhz" processor?? Nov 17 19:11:59 suihkulokki: finland? Nov 17 19:12:54 any cool FOSS things in .fi? Nov 17 19:13:14 no regular meetings.. perhaps its a time to start one Nov 17 19:13:46 likewise: I think brecis is mipsish Nov 17 19:13:58 likewise: well, if you happen to be somewhere around austria - let me know :> but we could also talk online if you have any specific upnp questions Nov 17 19:14:20 Jin^eLD: where is your server at? Nov 17 19:14:28 I mean, the package? Nov 17 19:14:30 likewise: my server is at http://mediatomb.org/ Nov 17 19:14:42 but that version is very old, about a year passed since the last release Nov 17 19:14:56 new version is supposed to appear this year :) Nov 17 19:16:10 Jin^eLD: what's the jslib dependency for? Nov 17 19:16:17 aaah Nov 17 19:16:22 we allow users to create a custom layout Nov 17 19:16:28 I think I tried to add that to OE a while ago :) Nov 17 19:16:44 for example structures like: Music / Artists / Album and so on Nov 17 19:16:55 is it a hard dependency, i.e. is it needed? Nov 17 19:17:03 koen: forget 0.8.1 ;) when the new one is out I will submit a .bb fiole Nov 17 19:17:04 file Nov 17 19:17:16 likewise: well, without it you will not have the nice structure Nov 17 19:17:21 I think koen meant jslib Nov 17 19:17:30 no, mediatomb Nov 17 19:17:35 ah, ok, sorry Nov 17 19:17:39 js is already in OE ;) Nov 17 19:17:58 http://www.openembedded.org/filebrowser/org.openembedded.dev/packages/js Nov 17 19:17:58 Jin^eLD: nice structure in some upnp presentation layer? Nov 17 19:17:59 likewise: with js the user can customize the import script to create a layout that s/he desires Nov 17 19:18:12 likewise: the structure how the content will be presented to the mediarenderer Nov 17 19:18:37 so on your MR device you can browse into a music container, there you will see subcontainers listed by artist, by date, genre and so on Nov 17 19:18:53 Jin^eLD: so, a datastructure? or a visual presentation? Nov 17 19:19:02 visual presentation of the content Nov 17 19:19:21 the js is only executed when data is imported Nov 17 19:19:35 Jin^eLD: I probably have to see it to fully understand it... Nov 17 19:20:05 I guess.. if you never had anything to do with UPnP it is better to download some server and play around with it Nov 17 19:20:20 if you have M$ you could try the Intel Tools for UPnP Nov 17 19:20:22 I think we have on in OE Nov 17 19:20:25 they are free Nov 17 19:20:29 koen: yes, you have ushare Nov 17 19:20:30 iirc effem added one Nov 17 19:23:08 Jin^eLD: to explain what we do: we simply write a endless MPEG program stream to disk, and in parallel fill a database with metadata (timecode, etc.) Nov 17 19:23:54 I see.. well, the UPnP server works with files Nov 17 19:24:10 so you would have to do some rewriting Nov 17 19:25:21 because you probably would want to offer some parts of that endless file, according to your metadata, right? Nov 17 19:25:43 rewriting - meaning - the code :) Nov 17 19:26:34 Jin^eLD: yes, currently the URL specifies begin and end UTC time Nov 17 19:26:40 * koen still doesn't really get the difference between TS and PS Nov 17 19:27:46 likewise: well, you would need to fill the server database accordingly and construct the resources correctly... uhm.. hard to explain Nov 17 19:27:47 well Nov 17 19:27:56 I know how you could do it so it would work with UPnP :) Nov 17 19:28:09 but I can't explain in just few words hehe Nov 17 19:28:29 koen: TS runs in small packets, optimized for transmission and synchronization. Nov 17 19:30:13 koen: PS assumes error-free transmission Nov 17 19:30:33 likewise: so if your disk start flipping bits, you loose? Nov 17 19:30:35 koen: and is optimized for storage (less overhead in packet headers) Nov 17 19:30:45 koen: yes, but disks all have ECC Nov 17 19:31:12 koen: CD/DVD has Reed Solomon error correcting and detecting codes. HD's have the same, a bit different Nov 17 19:31:56 yeah, murphy is right around the corner Nov 17 19:34:09 03zecke123 * r684 10bitbake-parser/ (ast.py evaluator_data.py nodecache.py parse-all.py): Nov 17 19:34:09 Implement inherit, include and require Nov 17 19:34:09 Parsing all .bb files takes 16 seconds on this macbook Nov 17 19:34:09 here. Nov 17 19:34:09 Handling of methods is still to do Nov 17 19:36:45 likewise: anyway... if you need information, just contact me Nov 17 19:38:44 Jin^eLD: I have little time. Nov 17 19:41:13 Jin^eLD: Can it be a URL instead of a file, in the media content? Nov 17 19:41:49 likewise: yes - it actually, when I said the server is working with files I meant - it is using files to organize the content / fill the database Nov 17 19:42:11 then the files are being served via HTTP, but can be RTSP or whatever Nov 17 19:42:18 ok, nice Nov 17 19:42:51 Does Windows (blugh) work with such a server? Nov 17 19:42:55 so you would only have to fill up the database correctly Nov 17 19:43:13 RP: ping Nov 17 19:43:16 likewise: not out of the box, you need a mediarenderer implementation Nov 17 19:43:49 or instead of filling a (second) database, convert our database content on-demand Nov 17 19:44:14 so - additoinal software; but usually people do not use MR software, usually you have a device at home like the D-Link DSM-320 or Philips Streamium or something like that Nov 17 19:45:00 Hi! Nov 17 19:45:28 likewise: you could just use your database; the only important thing is that you implement the UPnP ContentDirectoryService correctly Nov 17 19:45:39 koen|away, but would you update that corrupted gpe-icons package? Nov 17 19:45:40 ok Nov 17 19:55:19 likewise: btw, you are from .nl? where exactly? :) just curious Nov 17 19:55:54 I've been to .nl a couple of times, back in my Philips days ;) was fun, during the football championship, people really went crazy :) Nov 17 19:55:55 Eindhoven Nov 17 19:56:13 been there :) Nov 17 19:56:49 spent 2 weeks in Acht Nov 17 20:06:34 Jin^eLD: I've only been to Tannheim in .at, doing a parapente course in the mountains. Nov 17 20:07:58 likewise: do not even know where that is :> Nov 17 20:08:47 Jin^eLD: http://maps.google.nl/maps?f=q&hl=nl&q=tannheim,+austria&sll=52.372797,4.893749&sspn=3.635589,8.898926&ie=UTF8&om=1&z=12&iwloc=addr Nov 17 20:08:53 well, I am not a native .at guy, so I guess my low geographic knowlege of the country can be excused ;) Nov 17 20:09:59 aah, somewhere in Tirol Nov 17 20:12:18 psokolovsky: the issue isn't in the package Nov 17 20:12:30 psokolovsky: I regenerated the Packages file Nov 17 20:12:39 psokolovsky: the usual bugs in ipkg-make-index Nov 17 20:12:59 koen, I see, thanks! Nov 17 20:13:13 hi andersee Nov 17 20:14:22 woglinde: howdy Nov 17 20:19:11 andersee: I see some nice progress gettting made on uclibc Nov 17 20:20:18 koen: doing what I can... Nov 17 20:20:46 koen: I need to do a bit more to get the much delayed 0.9.29 release out the door though Nov 17 20:21:23 andersee: and go on a no-computer holiday after that? Nov 17 20:21:31 heh, i wish Nov 17 20:22:18 more likely, then get inundated with requests from lusers for help... Nov 17 20:22:53 which is exactly why a no-computer holiday is so needed after a release :) Nov 17 20:22:57 "I can't seem to get uClibc compiling on my PDP-11 and I demand that you fix it now" Nov 17 20:23:17 oh :( Nov 17 20:23:20 heh Nov 17 20:23:30 you broke pdp11 support!?!?! Nov 17 20:23:40 usually not quite that lame, though some do approach it. Nov 17 20:24:07 "I replaced libc6.so with the uclibc file and stuff doesn't work anymore" Nov 17 20:24:35 I've definitely seen that one Nov 17 20:24:46 sadly, I have seen it too Nov 17 20:29:26 zecke: parse meister! Nov 17 20:29:35 zecke: how did your test go? Nov 17 20:33:24 * koen stabs minimo Nov 17 20:36:26 mreimer: http://blogs.gnome.org/view/xan/2006/11/17/0 Nov 17 20:41:25 koen: which one? I should stufy for the 10 tests :) Nov 17 20:41:37 koen: eek I'm at 1:49 for parsing :( Nov 17 20:41:59 zecke: where did you get pyrex from? Nov 17 20:42:03 fink doesn't seem to have it Nov 17 20:42:13 but zero of the planned speedup is included... Nov 17 20:42:29 koen: as usual fixed and upgraded darwinports Nov 17 20:43:06 but I'm too stupid for darwinports! Nov 17 20:43:26 ??? Nov 17 20:43:27 koen: well, me too Nov 17 20:44:04 koen: those graphs aren't very impressive Nov 17 20:44:19 mreimer: that's what I though Nov 17 20:44:30 mreimer: so lets leave the xft patch in :) Nov 17 20:44:53 koen: yes. I am curious to see how GtkTextView performs comparatively Nov 17 20:45:00 yeah Nov 17 20:45:18 any volunteers for adding torturer to OE? Nov 17 20:45:38 torturer? Nov 17 20:45:39 could be neat for zecke's super QA plan Nov 17 20:45:55 hehe :) Nov 17 20:45:58 I love PAIN Nov 17 20:46:00 chouimat: gtk theme torturer, a synthetic benchmark Nov 17 20:46:10 bah gtk .... Nov 17 20:46:14 zecke where is my whip Nov 17 20:46:21 woglinde: whip? Nov 17 20:46:30 woglinde: I should punch you to add the msp stuff IIRC Nov 17 20:46:41 hm Nov 17 20:46:44 right Nov 17 20:46:52 woglinde: now you can even test it properly Nov 17 20:46:57 give me some time Nov 17 20:47:11 woglinde: Did you make a copy of this super secret C code? Nov 17 20:47:20 which? Nov 17 20:48:05 zecke I am to dull to get the timer working right on th esb Nov 17 20:48:13 03zecke123 * r685 10bitbake-parser/ (evaluator_data.py parse-all.py): Nov 17 20:48:13 Handle procedure, fakeroot, python and def... and parse time is at Nov 17 20:48:13 1:47 :( Nov 17 20:48:31 the timer? using this scatterweb stuff or plain on the register? Nov 17 20:48:40 plain registers Nov 17 20:48:50 woglinde: take a look at the scatter stuff :) Nov 17 20:49:28 no we will let mr. liers explain it to us Nov 17 20:49:29 who maintains the svn package? Nov 17 20:50:37 hi all Nov 17 21:10:09 what is the best channel to discuss haret? Nov 17 21:10:56 kergoth, if your around, let me know, i've some questions about tslib Nov 17 21:14:27 hey CoreDump|home Nov 17 21:14:33 hi Nov 17 21:15:29 03rwhitby 07org.oe.dev * rb86031f2... 10/ (1 conf/distro/openslug.conf packages/images/slugos-image.bb): slugos: Removed more kernel modules that are now compiled in Nov 17 21:18:18 any flex/lex expert around Nov 17 21:21:11 zecke: no, I've grown fat and rigid Nov 17 21:23:06 zecke hm Nov 17 21:23:18 woglinde: schocking that I have no clue about lex? Nov 17 21:23:18 I thought you were an expert Nov 17 21:23:40 woglinde: hehe, well I know a solution but... Nov 17 21:24:04 woglinde: the issue is lex is 'longest matching rule' based Nov 17 21:24:21 woglinde: and we have something like RVALUE which is everything behind a = and quoted Nov 17 21:24:37 woglinde: now we even have newlines and the '\' and even comments '#' Nov 17 21:24:48 woglinde: I would have to add an intermediate state to parse them Nov 17 21:25:49 hm Nov 17 21:29:09 i will need to do this... Nov 17 21:40:56 03zecke123 * r686 10bitbake-parser/bitbakec.pyx: -Work around lexer issue with multi line rvalues... Nov 17 21:41:16 03zecke123 * r687 10bitbake-parser/parse-all.py: -Reindent inside a method body to ease profiling Nov 17 21:46:35 re Nov 17 21:47:37 03zecke123 * r688 10bitbake/lib/bb/methodpool.py: Nov 17 21:47:37 Remember that we have compiled and added this file. Nov 17 21:47:37 This reduces the time of the new parser from 2:42 to Nov 17 21:47:37 1:6. The old one takes 1:42 on this macbook here Nov 17 21:58:06 zecke: pong Nov 17 21:59:01 zecke: Most of our time isn't in the parser then? :-( Nov 17 22:01:04 damn, wrong test :( Nov 17 22:01:41 RP: actually parsing of the files takes only a couple of seconds Nov 17 22:01:51 RP: building the data dicts takes most Nov 17 22:01:57 but we knew this Nov 17 22:02:42 zecke: an expand equivalent? Nov 17 22:04:15 zecke: It sounds like the runqueue which is taking an age on world builds. It does have 40,000 tasks to deal with though so I can't really blame it :-/ Nov 17 22:05:12 hmm up to 1:32 Nov 17 22:05:46 hmm i should go to bed Nov 17 22:06:26 * Spyro has a first cut at ac97 support for eseries... Nov 17 22:06:35 congrats Spyro Nov 17 22:06:53 :) Nov 17 22:07:33 Spyro: it's been a while since you worked on eseries, hasn't it? Nov 17 22:08:24 just a bit... Nov 17 22:17:56 03koen 07org.oe.dev * r6d9eca8a... 10/ (6 files in 3 dirs): Nov 17 22:17:56 dbus: add 1.0.1 Nov 17 22:17:56 RFC: remove dbus 1.0.10.92 as a cleanup? Nov 17 22:27:54 03zecke123 * r689 10bitbake-parser/ (ast.py evaluator_data.py parse-all.py): -Rolercoaster... From 1:8 up to 1:32 down to :30 by caching inherits again Nov 17 22:28:49 RP: I think we can stay below 1 minute on parsing :) Nov 17 22:28:52 zecke: good thing you didn't go to sleep :) Nov 17 22:29:01 zecke: cool :) Nov 17 22:29:21 zecke: btw, the greenphone is running 2.4.19-rmk7-pxa2 Nov 17 22:29:28 zecke: as we expected Nov 17 22:29:35 koen: no np patches? Nov 17 22:29:43 zecke: no idea Nov 17 22:29:50 koen: 2.6 kernels have no advantages, in fact they work worse Nov 17 22:30:03 zecke: twiun is disecting it right now in #openmoko Nov 17 22:30:08 zecke: -pxa2 means np patches Nov 17 22:30:10 RP: there is not even an rmk serie for 2.6 Nov 17 22:30:22 2.6 is bigger and user more ram Nov 17 22:30:47 2.6 can't be compiled by the only sane compiler ever releases: gcc 2.95.3 Nov 17 22:30:48 I'm just repeating what I was told about the Zaurus kernels. They were tested and found to be worse Nov 17 22:30:48 BTW: if someone wants to test svn co the code Nov 17 22:30:51 released* Nov 17 22:31:03 make CFLAGS="-I/usr/include/python2.X" CXXFLAGS=$CFLAGS Nov 17 22:31:23 RP: sure, but that is not due 2.6 but plainly because you suck ;) Nov 17 22:31:23 * RP notes this Nov 17 22:31:43 zecke: i'll try to test when I get darwinports working or networking in parallels Nov 17 22:31:53 zecke: but 2.6 sucks too, as do Zaurii ;-) Nov 17 22:32:18 koen: just compile pyrex Nov 17 22:32:21 koen: it just works Nov 17 22:32:32 and I need to add tasks and EXPORT_FUNCTIONS to it Nov 17 22:32:40 and someone needs to fix the recipes... Nov 17 22:32:43 I've never known anyone other than Lorn who has said that about a 2.6 zaurus kernel, ever... Nov 17 22:33:13 zecke: You should publish info on how to run the recipes tests ;-) Nov 17 22:33:30 cd bitbake-parser/standalone Nov 17 22:33:31 make Nov 17 22:33:45 ah :) Nov 17 22:34:01 ./bitbake-helper '/space/hacking/embedded/oe/org.openembedded.dev/packages/*/*.bb' Nov 17 22:34:09 if your OE is not in /space it is badly configured Nov 17 22:34:50 probabld yadd the same with '*.inc' Nov 17 22:35:05 and yes error messages suck Nov 17 22:36:16 moring Nov 17 22:36:34 hey XorA|gprs Nov 17 22:36:44 230 bb files and 16 inc files can't be parsed Nov 17 22:36:45 zecke: I can't find python.h in /usr/include/python2.X ;-) Nov 17 22:36:56 RP: for the standalone you don't need any python :) Nov 17 22:37:11 RP: and you violate my license if you don't send me a postcard Nov 17 22:37:32 RP: oh and that would be Python.h and i have a case insensitive fs (I do suck) Nov 17 22:37:33 XorA|gprs: how's london? Nov 17 22:38:05 gprs reception here sucks Nov 17 22:38:12 koen: I want Leopard badly :( Nov 17 22:38:41 koen: but erotica today wascool Nov 17 22:38:54 RP: does it compile? Nov 17 22:39:23 mhhh maybe i found (and correct) a bug on mesa_6.4.1+cvs20060101.bb ... seem sed -i don't like links ... bork only on real file :( ... strange ... Nov 17 22:40:03 zecke: me too Nov 17 22:40:14 zecke: actually I want a 12" mbp Nov 17 22:40:20 hehe Nov 17 22:40:26 zecke: this 15" version is way too big for me Nov 17 22:40:28 now you have an 15" one? Nov 17 22:40:32 koen: cut off 3"? Nov 17 22:40:41 thought about that Nov 17 22:40:45 koen: do it and post a video on youtube Nov 17 22:40:55 1.5" on each side Nov 17 22:41:28 yes, right along the speakers Nov 17 22:41:47 zecke: the backlit keyboard rocks, but apple put the controls under f8 and f9 Nov 17 22:41:53 zecke: hence disabling expose Nov 17 22:41:56 morons.... Nov 17 22:42:04 i do not F keys for expose Nov 17 22:42:55 zecke: CXXFLAGS=$CFLAGS doesn't work here ;-) Nov 17 22:42:58 I do when using D&D Nov 17 22:44:18 zecke: It does compile though and you did get the case in the header right, I just can't read/type ;-) Nov 17 22:44:56 If I escape the shell expansion, it even gives the errors :) Nov 17 22:46:49 RP: export them Nov 17 22:46:55 * XorA|gprs offskis Nov 17 22:47:16 RP: hehe, I sucked on the make cli but the quotes were copy and paste :) Nov 17 22:47:29 RP: it is a glob so exchange it with *.inc... Nov 17 22:48:02 zecke: \* works fine :) Nov 17 22:48:12 RP: yeah I fuzzed it already Nov 17 22:48:20 I ran it over /* on OSX Nov 17 22:49:09 RP: using the mix-in mickeyl proposed you can exchange the evaluation to a data dict Nov 17 22:49:23 with something using Dot and you get a complete syntax tree of OE Nov 17 22:49:34 so you can point people to "This is OE" Nov 17 22:50:12 zecke: Scary :) Nov 17 22:50:31 i have not tried to render the SVG... Nov 17 22:50:49 it would rock to do it on old metadata to see how it grew Nov 17 22:50:57 I gave up on converting the depgraph code due to being unable to read the SVGs :-/ Nov 17 22:51:18 zecke: You'd have to relax the syntax Nov 17 22:51:19 RP: oh and btw. I like hrw's idea of using overrides. Just prefix them with a namespace Nov 17 22:51:40 zecke: It makes everything machine specific Nov 17 22:51:48 zecke: That is a showstoper for me Nov 17 22:52:01 RP: I think I could relax it with two scanner lines Nov 17 22:52:34 RP: I simply don't do it, because I love to be a syntax... :} Nov 17 22:52:54 zecke: It would be handy to know how to enable if for the old metadata ;-) Nov 17 22:53:08 RP: I can do that, in the dark when no one looks Nov 17 22:53:17 zecke: I updated poky to bitbake trunk btw ;-) Nov 17 22:53:29 RP: including the sleep? Nov 17 22:53:37 ;) Nov 17 22:53:38 zecke: One of the few bits I changed ;-) Nov 17 22:53:57 The commit said it was 1.7.4ish :) Nov 17 22:54:21 why do we have 40.000 tasks? can't we share some somehow? Nov 17 22:54:38 I would say 80% of bbfiles don't have different set of tasks Nov 17 22:54:55 zecke: Every one of them has different depends Nov 17 22:54:58 decouble bb data from task data if we don't do that already Nov 17 22:55:08 ah okay Nov 17 22:55:11 The two are unconnected Nov 17 22:55:41 Its all the for x in y where y has 40,000 entries :-/ Nov 17 22:56:07 4,000 .bb's, 10 tasks each = nightmare :-/ Nov 17 22:56:30 RP: we will solve that... and I think they should give me a diploma for just this... Nov 17 22:56:31 Especially fetchall and its reverse rdeps - I think I know how to speed that up at least Nov 17 22:56:54 s/reverse/recursive/ Nov 17 22:57:34 Step 1) Make taskdata work properly. Step 2) Improve speed. We're still at step 1 ;-) Nov 17 22:58:10 Also, python threading is going to cause us big problems. We'll have to contiunue to use the C like fork methods :-/ Nov 17 22:58:27 RP: really? Nov 17 22:58:44 Question: Will we land the C Parser in BitBake 1.7 or in 1.9 Nov 17 22:58:50 zecke: Python threading isn't true threads - it only runs on one processor and timeshares :-/ Nov 17 22:58:58 lol Nov 17 22:59:07 well that is an implementation detail Nov 17 22:59:40 zecke: Its an implementation detail which means I'm sticking to C like forks ;-) Nov 17 23:00:08 that is the 1:n fork thread schema Nov 17 23:00:14 zecke: See how the C parser goes. I suspect it will be 1.7 Nov 17 23:00:16 where NTPL is m:n Nov 17 23:00:52 zecke: Before 1.8 we need a GUI which is going to take time. I suspect we'll have to split bitbake into a client and server Nov 17 23:01:12 (one using python threads, the other using C threads, talking via sockets) Nov 17 23:12:21 okay I have some bugs to fix... Nov 17 23:21:12 RP: hmm Nov 17 23:21:45 RP: maybe python2.5 already improved? One good optimisation we can do in the metadata is to shrink the anonymous handler Nov 17 23:22:02 python () {} or python __anonymous() {} Nov 17 23:22:13 these get compiled + evaled on __each__ bbfile Nov 17 23:22:17 _each_ Nov 17 23:24:53 wow I'm up at two minutes again :( Nov 17 23:25:37 time for bed Nov 17 23:26:13 koen: A good task is to almost kill the __anonymous methods Nov 17 23:26:22 koen: either by moving them into a method and calling them once Nov 17 23:26:28 right Nov 17 23:26:30 or by reqriting the baseclass Nov 17 23:26:50 although I hae no clue why these have to be anonymous Nov 17 23:26:59 probably some python thing Nov 17 23:27:05 they are called after the bb file was parsed Nov 17 23:31:41 fast parser + sane gui would rock Nov 17 23:32:33 fast parser - [sane||insane] gui would rock more Nov 17 23:33:07 zecke: is insane.bbclass still python 2.4 only? Nov 17 23:33:13 maybe Nov 17 23:33:45 anyway Nov 17 23:33:49 * koen gets back to bed Nov 17 23:45:09 03zecke123 * r690 10bitbake-parser/ (ast.py evaluator_data.py nodecache.py parse-all.py): Nov 17 23:45:09 Add handler, tasks, tmp up to 2:20 due copy.copy for the anonqueue Nov 17 23:45:09 Keep this information in the nodecache for now and we are currently Nov 17 23:45:09 at 52 seconds Nov 17 23:57:43 03freyther 07org.oe.dev * r7406f1b8... 10/ (18 files in 14 dirs): Fix parse error by adding one or two "'" or '"' removing whitespace behind \... Nov 17 23:57:51 zecke: I will talk to mickeyl about threads. There are some heavy optimisations I can make if I rewrite build.py so that is next on my "hitlist"... Nov 17 23:58:16 we need to get the parse errors sorted out so I can finally really compare... Nov 17 23:58:36 zecke: I will aim to look at solving a few :) Nov 17 23:58:54 RP: the matchbox ones would help, but don't waste your time with this stuff Nov 17 23:59:08 I agree on anonymous methods btw, I've been thinking that myself for a while Nov 17 23:59:22 Things like COMPATIBLE_MACHINE helped Nov 17 23:59:31 RP: I changed base.bbclass and that made only two seconds difference Nov 17 23:59:41 oh but that was cached... hmm Nov 18 00:01:49 When do we run evals in variables? Nov 18 00:02:04 depending on the type of assignment Nov 18 00:02:10 := == Immediate Assignment Nov 18 00:02:37 so = or += are deferred until the variable is accessed? Nov 18 00:02:43 yes Nov 18 00:02:52 they operate on the unextended version Nov 18 00:03:19 there might be something we can improve there then... Nov 18 00:03:37 (by moving anonymous methods to evals) Nov 18 00:04:02 hmm it only make a really small difference :( Nov 18 00:04:39 zecke: So expansion is the big time user? Nov 18 00:06:53 I think I won't work on the parser until next weekened Nov 18 00:08:22 I know the competing for time problem :-/ Nov 18 00:08:27 ~lart woglinde for disrupting my sleeping schedule Nov 18 00:08:27 * ibot says "boot to the head" and knocks woglinde over for disrupting my sleeping schedule Nov 18 00:09:10 I should really sleep now... Nov 18 00:09:23 wait a second please :) Nov 18 00:10:29 or just leave, have a good nite Nov 18 00:10:39 I'm still here ;-) Nov 18 00:10:47 i will need some more time Nov 18 00:10:55 I'm removing the anonfuncs from base.bbclass Nov 18 00:11:02 I can wait :) Nov 18 00:11:04 but need to reindent the classes Nov 18 00:14:13 removed some instances, base, gettext,update-rc.d,update-alternatives Nov 18 00:15:10 Keep in mind some of the anonymous methods call skippackage so that might skew the results :-/ Nov 18 00:16:06 waiting for the parser now Nov 18 00:17:32 ~7 secs difference Nov 18 00:18:42 http://rafb.net/paste/results/L6Cg9P97.html Nov 18 00:19:33 5 on the new parser Nov 18 00:19:40 6 even Nov 18 00:20:16 So whilst its a slowdown, its not the major problem... Nov 18 00:20:40 for the new parser this is 10% of the time Nov 18 00:21:01 That makes it sound better :) Nov 18 00:21:09 yes it does Nov 18 00:21:17 next optimisation Nov 18 00:24:29 okay... not now sleep well RP Nov 18 00:24:41 The next optimisation would be to execute all 'anon' funcs at the same time Nov 18 00:24:50 instead of simply compiling one line at a time Nov 18 00:24:59 but this requires the same indention which is not the case now Nov 18 00:25:08 That's a neat idea :) Nov 18 00:25:34 mickeyls policy was to have :set et sts=4 sw=4 for python stuff anyway Nov 18 00:26:05 yes, we need to reindent the classes Nov 18 00:27:22 oh I can't stop :( Nov 18 00:28:51 I can :) Nov 18 02:54:58 03rwhitby 07org.oe.dev * r9db97dcc... 10/ (1 packages/slugimage/slugimage.bb): slugimage: fixed syntax error introduced by parser fixes **** ENDING LOGGING AT Sat Nov 18 02:59:56 2006