**** BEGIN LOGGING AT Thu Jul 07 23:59:57 2005 Jul 08 00:40:00 morning Jul 08 00:41:56 morning Jul 08 00:47:28 monring all Jul 08 00:51:18 ade|desk: morning Jul 08 01:03:13 ibot: botmail for pb: see bug #137. I confirm that collection is working with the patch. Jul 08 01:04:00 mickeyl: hi Jul 08 01:04:34 good morning Jul 08 01:04:47 morning mickeyl Jul 08 01:04:48 mickeyl: Good morning. Jul 08 01:04:52 mickeyl: I put pb testcase and fix under #137. One of the most hackish piece of code I met. Jul 08 01:05:05 sirfred: thanks for the revised patch. It's in Jul 08 01:05:15 mickeyl: Thanks to you. Jul 08 01:05:22 mickeyl: Only a question about it. Jul 08 01:05:32 proti: ya, i'm utterly confused about all the different pkg_pn, pkg_fn, pkgdata, self.status, etc. Jul 08 01:05:45 mickeyl: I just send you the script for the apm_proxy, but I don't know how to integrate it. Jul 08 01:06:36 mickeyl: So was I. In original code, pkg_pn serves different purpose with different layout. I figured it yesterday. Jul 08 01:06:55 sirfred: I'll think about that and experiment. Jul 08 01:07:55 mickeyl: OK. Thanks. Jul 08 01:08:11 mickeyl: Take in mind that without that apm script, the patch is no useful. Jul 08 01:09:23 sirfred: shouldn't it work with suspend/resume from Opie without the script? ODevice sends exactly those two qcop messages before/after suspend Jul 08 01:09:44 mickeyl: It should if the messages are exactly those. Jul 08 01:10:21 mickeyl: I tested it using 'apm --suspend' by hand, anyway. Jul 08 01:10:35 mickeyl: Tested three or four times, and worked fine. Jul 08 01:10:58 http://handhelds.org/cgi-bin/cvsweb.cgi/~checkout~/opie/libopie2/opiecore/device/odevice_zaurus.cpp?rev=1.40&content-type=text/plain Jul 08 01:11:02 Zaurus::suspend() Jul 08 01:12:56 mickeyl: Yes, that is the same that I was invoking by hand. Jul 08 01:13:09 excellent. now the #1 question... Jul 08 01:13:21 how will the ATIcore driver behave when it gets the messages twice ? Jul 08 01:13:39 which it will in case the script is installed and we are resuming from opie Jul 08 01:13:47 mickeyl: I think it won't be a problem. Jul 08 01:14:13 mickeyl: I'm now storing the state of the driver. Jul 08 01:14:35 mickeyl: So, I've changed a lot the driver, and we have now a core class that knows if the driver is attached or not. Jul 08 01:14:50 sirfred: ah, excellent. can't wait to test it :) Jul 08 01:14:52 mickeyl: So, when a second prepareToSuspend() arrives, nothing should happens. Jul 08 01:15:09 mickeyl: The same with resuming, the attaching only happens once. Jul 08 01:15:26 mickeyl: To avoid race conditions trying to paint when the driver is not attached. Jul 08 01:15:57 mickeyl: I'm also honoring the state of the driver, so, drawing operations before complete resuming won't be accelerated. Jul 08 01:16:12 mickeyl: I'm just redirecting them to the software gfxraster. Jul 08 01:16:46 mickeyl: That seemed to be the big problem, and making the driver segfaults on resume. Jul 08 01:17:11 sirfred: oh wow, that was quite an undertaking. good work Jul 08 01:17:46 ibot: sirfred is Qt/Embedded Überhacker Manuel Teira Jul 08 01:17:47 okay, mickeyl Jul 08 01:17:51 ~sirfred Jul 08 01:17:51 extra, extra, read all about it, sirfred is Qt/Embedded Überhacker Manuel Teira Jul 08 01:18:00 mickeyl: :) Jul 08 01:18:09 mickeyl: What überhacker means' Jul 08 01:18:12 mickeyl: What überhacker means? Jul 08 01:18:27 sirfred: you know Hacker, right? Jul 08 01:18:42 mickeyl: Yes, I know. Jul 08 01:19:09 sirfred: Über is a word play derived from Nietsche's (german philosopher) Übermensch, which is a supernatural human being. So, Überhacker means supernatural hacker. Jul 08 01:19:30 mickeyl: I'm not worthy. Jul 08 01:20:03 heh, you are :) I've only look briefly at the driver and that low level code scared me away Jul 08 01:20:30 mickeyl: It has a lot of problems. Jul 08 01:20:42 mickeyl: Sometimes, some pixmaps get corrupted, it's gonna be hard to find the remaining problems. Jul 08 01:20:54 mickeyl: Also the problem you found, with some pixmaps not being drawed. Jul 08 01:21:30 mickeyl: The problem with this is that debugging problems is difficult, because every render to screen implies hundreds of operations, so, it's difficult to know what operation matches with what rendering. Jul 08 01:24:12 i see. sounds unpleasant. Jul 08 01:24:34 mickeyl: Restarting monotone build after disabling malloc_stats and memcpy stats. finger crossed. Jul 08 01:24:48 Results in a couple of hours. Jul 08 01:25:07 Did I say that I love how TTF fonts look in opie? Jul 08 01:25:50 I should go back to 3.5.3 to test properly the rotation, but I don't want! Jul 08 01:26:14 proti: cool. keep me posted. Jul 08 01:26:34 sirfred: yeah, I'm glad hrw convinced me to do the TTF convertion Jul 08 01:27:26 mickeyl: AtiCore has some accelerated functions to draw chars, but only monochrome ones, as far as I know. Jul 08 01:27:59 I suppose opie ones are using subpixel rendering, perhaps. Jul 08 01:31:20 ya, i think so as well. the glyphs are rendered by freetype Jul 08 01:33:22 mickeyl: Anyway, I think that it could be useful. We only need to check on blt if the sourcedepth == 1 Jul 08 01:33:33 mickeyl: In that way, we know we are rendering a monochrome pixmap. Jul 08 01:43:04 * mickeyl slowly unpacks his shiny new NSLU2 Jul 08 01:45:52 morning Jul 08 01:46:01 * hrw just for few minutes - vacations ;) Jul 08 01:46:32 mickeyl: If you remove R83 in less than 10 minutes, you'll have the record on shortest time from box to turboslug ;-) Jul 08 01:46:59 sirfred: with 3.5.3 you just need to install opie-freetype/3.5.3 and get opie-ttf-support from current OE to have ttf fonts (and ttf-dejavu ofcourse) Jul 08 01:47:49 hrw: Thanks. I cannot leave without this ttf fonts. Jul 08 01:48:39 sirfred: I use ttf in opie since 3.5.1 or maybe even 3.5.0 Jul 08 01:49:08 I was leaving in the dark all this time. Jul 08 01:49:13 I was living in the dark all this time. Jul 08 01:49:22 ~google opie and ttf Jul 08 01:49:44 1st two links should me mine Jul 08 01:50:35 morning hrw Jul 08 01:51:04 NAiL: hehe, untalented as I am regarding hardware I'll probably not attempt to break that record :) Jul 08 01:51:23 ~praise hrw for inspiring us re. TTF Jul 08 01:51:23 All hail hrw for inspiring us re. TTF! Jul 08 01:51:47 mickeyl: People have removed the resistor using nail clippers (!) ;) Jul 08 01:52:08 NAiL: heh, so I'm going to ask my wife to do it :D Jul 08 01:52:21 haha Jul 08 01:52:44 the nslu2 is smaller than i thought. very nice Jul 08 01:53:37 Yeah. And there's lots of space inside for doing cool mods. Like adding a USB hub inside, adding a couple of flash sticks, usb display, and stuff like that ;) Jul 08 01:53:53 a display would rock Jul 08 01:56:30 mickeyl: http://slug.repvik.info/slugpics/lcd.png Jul 08 02:00:13 * NAiL notices the image takes a whopping 2.5mb Jul 08 02:02:02 nslu2 is ~3.5" hdd size? Jul 08 02:03:15 slightly less Jul 08 02:03:22 s/less/smaller Jul 08 02:03:57 nice.. Jul 08 02:04:20 about 1cm shorter, and 1cm less wide Jul 08 02:04:54 about the same thickness, I'd say Jul 08 02:06:32 What's the opposite of wide anyways? "less wide" sounds kinda awkward ;) Jul 08 02:06:48 narrow? Jul 08 02:06:57 thin? Jul 08 02:08:07 heh, can't make it make sense Jul 08 02:08:18 mickeyl: Did you look at the pic? :) Jul 08 02:10:41 ok - I go shopping and tomorrow I go to my parents or week of nothing-to-do vacations.. lakes, bikes etc.. Jul 08 02:11:06 hrw: You're desperatly healthy. Jul 08 02:11:39 What's that NLSU supposed to have in? Jul 08 02:12:28 Every real Hacker would use the week to bury itself in the basement and hack 20H a day. :) Jul 08 02:13:17 I don't want to be a real hacker. Jul 08 02:13:48 proti: I'm tired of hacking Jul 08 02:14:23 proti: and before starting such hackparty I would need to learn qte2, qt4 to can hack opie to opie/x11 Jul 08 02:14:38 hrw: That's why I do so little hacking. My wife keeps me out of the den. Jul 08 02:14:54 She says I need to get a life. Jul 08 02:15:42 NAiL: hehe, that's pretty cool! Jul 08 02:15:55 mickeyl: Yeah :D Jul 08 02:16:39 sirfred: nslu is basically a small linux machine with 2 usb ports and an ethernet ports. it is marketed as "usb -> nfs/smb" bridge but of much broader use Jul 08 02:16:53 NAiL: opposite of wide is narrow. Jul 08 02:17:07 thanks Jul 08 02:17:24 you're welcome Jul 08 02:17:38 mickeyl: No HD? Jul 08 02:17:55 motorola sb5100 (cable modem) + linksys wrt54gs (router & AP) + nslu + harddisk == small home server Jul 08 02:17:59 sirfred: you attach external USB-disks Jul 08 02:18:34 but it need too many power supplies.. Jul 08 02:18:45 hrw: how much mem has the nslu ? Jul 08 02:18:52 Oh, I understand. Jul 08 02:18:57 hrw: not if you use 2.5" disks :) Jul 08 02:19:03 proti: 32mb Jul 08 02:19:44 and 8mb flash Jul 08 02:19:52 Not very much if you want to run decent services. Jul 08 02:20:15 NAiL: now under desk I have ~15 cables, 3 power supplies, subwoofer, wrt, sb5100 - too much Jul 08 02:20:24 How much does it cost? Jul 08 02:20:27 I keep my sun. Jul 08 02:20:30 proti: thttpd + imap/ssl is enought for me Jul 08 02:20:34 hrw: ~$80 at newegg Jul 08 02:20:36 duh Jul 08 02:20:42 sirfred: $80 ;) Jul 08 02:21:06 hrw: One power brick = nslu2+2x2.5" disks. Jul 08 02:22:38 NAiL: sure - but it will be 4th one Jul 08 02:23:03 hrw: I have imap apache postfix ssh named racoon not so much. Jul 08 02:23:04 NAiL: I'm out of power sockets Jul 08 02:23:14 hrw: that bad, huh? Jul 08 02:23:58 NAiL: too many cables Jul 08 02:24:03 ok. I have to go really Jul 08 02:24:24 hrw: Enjoy your holidays. Jul 08 02:24:39 have a nice weekend, week, weekend all - I will be back 20050718 (before maybe will be from time to time here) Jul 08 02:24:42 sirfred: I will Jul 08 02:24:58 hrw: have fun Jul 08 02:25:22 and my mother promised me that she will make 'kartacze' :) (regional food which idea came from Lithuania) Jul 08 02:25:38 My holidays are near too. This Sunday, I'm leaving to my country. Jul 08 02:25:41 sounds good, have fun hrw Jul 08 02:26:16 next vacations will be in August: 5-22 (and there will be only 15 free days left after) Jul 08 02:26:51 will you be away during the vacation in august ? Jul 08 02:27:08 mickeyl: yep - dont know where yet Jul 08 02:27:18 ok. i'm asking because of preparing the 3.5.4 release which is supposed to happen somewhere in august. we can wait until you're back Jul 08 02:27:26 ok Jul 08 02:27:33 now go and enjoy the weekend :) Jul 08 02:27:41 and vacations;) Jul 08 02:27:44 yeah Jul 08 02:28:41 hrw|vacations: Have nice holidays! Jul 08 02:28:44 mickeyl: good morning Jul 08 02:29:22 RP: Morning. Did you have time to play with the wince dll ? Jul 08 02:29:49 sirfred: No, I'm afraid not Jul 08 02:30:18 sirfred: Its unlikely I'll be able to do much for the next week now - some holiday time for me as well Jul 08 02:30:20 good morning pb_ Jul 08 02:30:42 RP: Yes, it's time to rest a little. Jul 08 02:31:13 hi all. i'm trying to get kbdd work on my c750 OpenZaurus GPE, but it looks like the uinput module isn't availiable. I whish i could provide a bb file, but i don't know anything in coding. Does anyone have some free time to build this kernel module ? (unless we can do without...) Jul 08 02:31:19 RP: I found a coredll.dll that's one of the dependencies of the ati dll. Jul 08 02:32:14 RP: Some calls instead of unknown references appeared, but, anyway, we still have a barrier in something like a zeroed jump table. Jul 08 02:32:48 alan|gotnewjob: It needs adding to the kernel's defconfig. I'll do that but I'm not in a position to rebuild it atm Jul 08 02:33:09 sirfred: So its likely some other dependencies are missing? Jul 08 02:33:33 RP: I think that the problem is that some jump table is loaded when the library is setup. Jul 08 02:33:35 RP : thanks a lot. it's not urgent, anyway. Jul 08 02:33:49 RP: But I'm not an expert, as you know. Jul 08 02:33:54 alan|gotnewjob: If not before, it wil appear in ox 3.5.4 Jul 08 02:34:07 in oz even... Jul 08 02:34:44 RP: that would be great ! Jul 08 02:34:50 sirfred: I'll have to look at it... Jul 08 02:34:55 RP: Perhaps some kind of native debugging is needed. Jul 08 02:35:14 RP: Yes, It would be nice to have another source to compare, and I suppose, another microcode. Jul 08 02:37:18 sirfred: I will look at it. I just don't have any time for a while :-( Jul 08 02:37:33 hmm that forced ext3 is disputable Jul 08 02:37:47 RP: Take it easy Jul 08 02:38:34 mickeyl: A think I asked you yesterday. Jul 08 02:38:57 mickeyl: About no script in /etc/apm/scripts.d. Jul 08 02:39:05 mickeyl: Perhaps they're not needed anymore? Jul 08 02:39:21 mickeyl: There used to be a pair of scripts to suspend/resume ifaces and usb. Jul 08 02:40:00 mickeyl: Yesterday, testing the driver, my ifaces seemed to work properly without those scripts that stopped and resumed them. So, are they no longer needed? Jul 08 02:40:50 What I want to know if the lack of those scripts is a fault of me (building opie-image), a fault of opie-image itself or is something deliberate Jul 08 02:42:03 iirc we still ship scripts in default images Jul 08 02:42:18 i have no pda here to verify though Jul 08 02:42:34 what's mplayer-atty? Jul 08 02:43:02 giel: atty.jp Jul 08 02:43:31 heh, typically that you answer, i was just browsing xora.org.uk :) Jul 08 02:44:00 * XorA wonders what is actually there these days Jul 08 02:44:11 nothing muchj Jul 08 02:44:19 but uhm, i can't really read japanese Jul 08 02:44:27 so, what's the difference with the normal mplayer? Jul 08 02:44:53 sirfred: anyway many of the scripts origin in 2.4-embedix kernels which need pretty much help to work at all :) Jul 08 02:44:56 giel: I think is the one with the w100_vidix driver. Jul 08 02:45:00 hm ok Jul 08 02:45:04 giel: But not sure. Jul 08 02:45:30 giel: as sirfred says, w100/bvdd drivers and some optimised arm pxa assembly Jul 08 02:45:37 'cause the normal mplayer hangs for me when it is "Checking audio filter chain" Jul 08 02:45:55 now i'm trying to run it on some weird device, but it should be a fairly standard device Jul 08 02:46:03 mickeyl: Well, those scripts were there on 3.5.3 Jul 08 02:46:29 giel: try with -oa null to make sure it is working without audfio Jul 08 02:46:38 giel: your device is simultaneously weird and standard? crumbs. Jul 08 02:46:39 XorA: yep, running it with -ao null -vo null Jul 08 02:47:05 pb_: it's weird as in, i'm about the only one that has it, but it should be fairly standard, according to the design docs Jul 08 02:47:36 hey Jul 08 02:47:42 zecke: hail Jul 08 02:47:51 morning Jul 08 02:49:53 moin zecke Jul 08 02:53:01 * mickeyl ponders whether to use his slug at least once with the default firmware or to flash openslug right away... Jul 08 02:53:21 mickeyl: heh, might as well flash openslug immediately. Jul 08 02:53:27 mickeyl: Go OpenSlug :D Jul 08 02:53:29 mickeyl: can I help in migrating your (I hope lua scripts) to monotone 0.20 Jul 08 02:53:39 unless you plan to use it as a fileserver, which isn't really one of its core competencies anyway, the default firmware is not much use. Jul 08 02:54:09 zecke: it's not lua scripts, it's python. viewmtn and my hacked version of ciabot_monotone.py Jul 08 02:54:29 mickeyl: oh I can't help with python ;) Jul 08 02:54:30 zecke: yeah, you could help. they need to be adapted to the new CLI Jul 08 02:54:51 zecke: hehe too late, we already know that :) I'll put my hacked versions up in a minute Jul 08 02:55:05 pb_: yeah, you're probably right. Jul 08 02:55:28 i reckon openslug will remove the need of ext3 reformatting? Jul 08 02:55:57 mickeyl: correct Jul 08 02:56:05 pb_: I'll now mess with your testcase while waiting for my ABI bugs to magically disappear Jul 08 02:56:17 ö Jul 08 02:56:41 s/ö//zecke: did you see proti's fix-collections patch? Jul 08 02:57:03 mickeyl: didn't get that mail ;) Jul 08 02:57:42 zecke: http://www.frankengul.org/~seb/fix_collection.patch Jul 08 02:58:02 * mickeyl wanders through the nslu2 wiki Jul 08 02:58:52 mickeyl: http://bugs.openembedded.org/attachment.cgi?id=64 is the same? Jul 08 02:58:55 mickeyl: feel free to "/j openslug" ;) Jul 08 02:59:09 should I risk comitting this patch and getting the next bugreport by proti that selection is wrong? Jul 08 02:59:40 zecke: does that patch interfere with 'build PN-PV' ? Jul 08 03:00:16 mickeyl: dunno Jul 08 03:00:41 zecke: yes, those patches seem to be identical. Jul 08 03:00:47 mickeyl: isn't there a prepend method for lists? Jul 08 03:00:49 zecke: ok, this should be tested before commiting. proti? Jul 08 03:00:58 mickeyl: it shouldn't interfere at all Jul 08 03:01:11 zecke: sure. there's list.insert( index, element ) Jul 08 03:01:41 and I figure it is faster than [something] + tmp_pn? Jul 08 03:02:38 is gcc-Version 3.4.4 (release) (CodeSourcery ARM Q1cvs 2005) known to be miscompiling? Jul 08 03:02:38 a thousand times Jul 08 03:02:53 mickeyl: doesn't sound worth changing it Jul 08 03:03:02 heh Jul 08 03:03:14 both are O(n) then with a slighlty different constant Jul 08 03:03:43 okay, mplayer works now, but slow as hell until about halfway the short clip, and then it plays normally Jul 08 03:03:43 GNU assembler version 2.15.96 (arm-linux) using BFD version 2.15.96 20050323 Jul 08 03:03:46 wierd Jul 08 03:03:46 ei* Jul 08 03:04:52 giel: Perhaps it's a problem of IO? Jul 08 03:05:58 sirfred: could be... -nocache or -cache 8192 don't seem to differ Jul 08 03:06:09 i'll try compiling now with xscale optimization Jul 08 03:06:25 giel: I was going to sugest you using -cache. :) Jul 08 03:06:32 s/using/to use/ Jul 08 03:08:50 zecke: excellent Jul 08 03:10:24 mickeyl: http://svn.berlios.de/viewcvs/bitbake/trunk/bitbake/bin/bitbake?rev=238&r1=236&r2=238 Jul 08 03:10:28 sirfred: and i don't think it's IO, cause mplayer is taking 100% cpu, not io... later on, mplayer only takes 30% Jul 08 03:10:59 mickeyl: now compare that to protis patch Jul 08 03:12:43 zecke: at first glance, it looks like they're reciprocal Jul 08 03:13:11 mickeyl: at least two hunks Jul 08 03:15:24 I'm compiling with "-march=armv5te -mtune=xscale", is that a stupid idea in the first place and second did anyone try that optimisation with the CSL gcc? Jul 08 03:17:24 zecke: no, that's a reasonable idea. Jul 08 03:17:30 I don't know if anyone has tried it with the csl gcc. Jul 08 03:18:25 g++ seems broken 60 pages of missing vtables when trying to link Jul 08 03:19:58 that does sound unfortunate Jul 08 03:20:14 could try gcc 4.0 and see what happens there. Jul 08 03:21:06 haha gcc4 and Qt2.3.10 :} Jul 08 03:21:18 I'm trying first with stock 3.4.4 and then later with gcc4 Jul 08 03:21:28 righto Jul 08 03:21:53 OT: that idiot of bitbake maintainer should finally fix the slow parser Jul 08 03:25:58 yeah, I heard he was investigating ways to use bison and lex from python Jul 08 03:26:14 better for him Jul 08 03:26:24 let's hope Jul 08 03:27:09 monotoneExploringThread().join(); do_real_work(); Jul 08 03:34:21 http://vanille.de/temp/monotone.vanille.de-20070708.tar.bz2 Jul 08 03:34:30 contains a copy of the site Jul 08 03:58:00 proti: your patch is wrong Jul 08 04:03:12 pb_: found the issue Jul 08 04:03:47 re Jul 08 04:06:45 zecke: Why ? Jul 08 04:09:52 proti: one part is a revert of a patch from you Jul 08 04:10:40 proti: it is likely PREFERRED_PROVIDERS are broken again Jul 08 04:12:12 ah I see gcc 4.0.1 is out Jul 08 04:13:06 zecke: excellent Jul 08 04:13:14 jacques: ah, cool Jul 08 04:14:29 pb_: I've a fix for your issue as well but it breaks other stuff :} Jul 08 04:14:45 proti: PREFERRED_VERSION is indeed broken Jul 08 04:14:59 zecke: :-} Jul 08 04:15:41 bitbake -BUG 1 or -BUG 2 -BUG 3 to decide in which way the feature should be broken ;) Jul 08 04:15:49 heh Jul 08 04:16:18 proti: got some time to talk? Jul 08 04:16:46 proti: the issue is within the for file_name in files selection Jul 08 04:16:58 proti: if no preferred provider is set the latest version will be used Jul 08 04:17:20 proti: but with BB collections the highest prior*ized is in front of the list Jul 08 04:19:02 so Jul 08 04:19:04 if preferred_file is None: Jul 08 04:19:04 preferred_file = pkg_pn[pn][0] Jul 08 04:19:04 preferred_ver = self.status.pkg_pvpr[preferred_file] Jul 08 04:19:34 would fix PREFERRED_VERSION and bbcollection but break 'select the latest by default' Jul 08 04:34:11 Would this PREFERRED_VERSION problem be why bitbake tried to build both glibc 2.3.2 and then glibc 2.3.5 in a single build? Jul 08 04:35:36 zecke: You are correct about PREFERED_VERSION broken. I realize now. Jul 08 04:36:46 I realize why the function was done like this. r236 should be almost reverted. Jul 08 04:37:10 zecke: Whay I disliked was the pkg_pn[pn] = tmp_pn Jul 08 04:37:48 zecke: This will certainly break the shell if you invoke build 2 times. Jul 08 04:40:39 zecke: line 276, if we iterate on the untouched pkg_pn, we don't obey the priority order. Jul 08 04:40:58 s/pkg_pn/pkg_pn[pn]/ Jul 08 04:42:27 zecke: The actual code will elect the first element that match the PREFERRED_VERSION if no revision is given Jul 08 04:43:34 proti: If I remember, the pkg_pn are ordered in a way that make the latest package is at the end, so it will elect always the lowest packages. Jul 08 04:44:14 zecke: And it is selected because the glob (*.bb) will sort in alpha order. Jul 08 04:45:07 zecke: Two solution,1- restore the inital walk of tmp_pn (which is pkg_pn[pn] ordered by chunk of same priority). Jul 08 04:47:17 zecke: 2- rebuild the pkg_pn in an ordered way, and at the same time build the tmp_pn for respecting the priorities given. Jul 08 04:49:18 zecke: And keep both under hand since we need them afterward. Jul 08 04:52:50 zecke: In the old oe code, and the # get highest priority file set, was under the condition if preferred_file == None: Still true ? Why has this cond been removed ? Jul 08 04:53:32 zecke: It meant that preferred_version was of much higher importance than the priority level. Jul 08 05:10:32 RP: no ;) that problem comes from uncomitted 'fixes' Jul 08 05:14:57 zecke: Ah ;-) Jul 08 05:16:45 proti: hmm let me think a bit more Jul 08 05:26:52 proti: 1) sound more easy? Jul 08 05:33:04 proti: do you want to change it? Jul 08 05:33:21 zecke: Yes, I will. Jul 08 05:34:04 however you do it... please keep the shell working ;) Jul 08 05:34:29 yeah prio 1) do not destroy global data structure Jul 08 05:34:41 Indeed Jul 08 05:35:49 zecke: Can you improve pb testcase to show the bad behaviour of PREFERRED_VERSION, or give me a little testcase ? Jul 08 05:36:06 wait a sec Jul 08 05:43:34 proti: testcase added Jul 08 05:49:41 Ok thanks Jul 08 05:56:10 Holidays, at last! Jul 08 06:21:39 zecke: ping. build nano with your testcase and my patch try to build nano-1.1.0. Where is the problem ? Jul 08 06:22:23 'your patch' != what you published Jul 08 06:23:33 zecke: My bitbake : http://pastebin.com/309458 Jul 08 06:25:05 lol whatever Jul 08 06:25:45 zecke: What result do you have ? Jul 08 06:26:01 13:35 < proti> zecke: You are correct about PREFERED_VERSION broken. I realize now. Jul 08 06:26:18 you contradict yourself, luckily I've now work to do so I stay calm Jul 08 06:27:10 with 0% probability things continue to work if you change from list to list of lists Jul 08 06:27:18 zecke: I says that there can be a problem with it by reading the code. Jul 08 06:27:39 for the same reason that when going to [[]] to [] there is a reason (where you sent a patch) Jul 08 06:28:36 do you've a PREFERRED_VERSION in your bitbake.conf? Jul 08 06:29:15 I do but I don't run the bitbake so often on the machine. It takes ages to parse bb Jul 08 06:30:15 in the test case are three files Jul 08 06:30:29 sorry 6 by now Jul 08 06:34:05 zecke: Ok. I tried to build nano and it tried to build the nano 1.1.0 as specified by the local.conf. Didn't looked any further. Jul 08 06:34:58 zecke: I don't see why should be angry. I asked you for a testcase and I see it works. I just asked you what I did wrong. Jul 08 06:35:34 nano 1.1.0 is right but in that case something worked that can not work with your code Jul 08 06:36:19 the time I need to spend to 'evelate' you is more than I would need to spend on fixing it and this part gets from time to time annoying Jul 08 06:37:21 looking at conf/bitbake.conf would reveal a PREFERRED_VERSION_nano = "1.1.0" Jul 08 06:38:06 zecke: I was asking you the result you have on this testcase, nothing more. Jul 08 06:38:18 File "/home/freyther/programming/bitbake/bitbake/bin/bitbake", line 326, in findBestProvider Jul 08 06:38:21 preferred_ver = self.status.pkg_pvpr[preferred_file] Jul 08 06:38:23 TypeError: list objects are unhashable Jul 08 06:38:25 bitbake -s with your patch Jul 08 06:39:13 zecke: which revision on bitbake ? Jul 08 06:39:15 hmm my fault Jul 08 06:39:37 nano 1.2.3-r0 1.1.0-r0 Jul 08 06:39:37 nano2 1.4.0-r0 1.1.3-r0 Jul 08 06:39:37 test 100-r0 200-r0 Jul 08 06:39:49 is the result using your patch Jul 08 06:40:52 wait a sec Jul 08 06:40:58 zecke: I tried bitbake -D -D -D nano Jul 08 06:41:06 and it builds nano-1.1.0 Jul 08 06:41:21 bitbake -s gives the following error : Jul 08 06:41:22 ERROR: anonymous function: access() argument 1 must be string, not None Jul 08 06:42:09 hmm now works nicely.. Jul 08 06:42:16 I had other changes in my bin/bitbake Jul 08 06:44:09 proti: hmm your patch seem to work alright Jul 08 06:45:15 zecke: apologies accepted. Jul 08 06:45:25 proti: yeah I had old cruft in it Jul 08 06:45:38 proti: but I'm still confused http://svn.berlios.de/viewcvs/bitbake/trunk/bitbake/bin/bitbake?rev=238&r1=236&r2=238 with this Jul 08 06:45:44 proti: I do not want to play ping pong Jul 08 06:46:10 your patch works because we do not do pkg_pn[pn] = tmp_pn Jul 08 06:46:10 zecke: I acknowledge there can be a problem. I will try to build a testcase that will show it. Jul 08 06:47:01 no your patch is allright Jul 08 06:47:17 let us split two methods out of findBestProvider Jul 08 06:47:32 to clarify the code Jul 08 06:48:27 pb_: how should PREFERRED_VERSION work? should it pick the bbfile from the higher prioritized collection? Jul 08 06:49:43 proti: the whole PREFERRED_VERSION 'loop' does not operate on tmp_pn at all Jul 08 06:52:37 zecke: It was the potential problem I saw. Fixing it, will reintroduce some deleted code. Jul 08 06:53:14 I Just need to figure out in what case we get bad behaviour. Jul 08 06:55:23 I presume that it will be some edge case involving priorities and collections. Jul 08 06:58:45 zecke: it should do what that comment in bitbake says Jul 08 06:59:24 If there is a PREFERRED_VERSION, find the highest-priority bbfile Jul 08 06:59:24 providing that version. If not, find the latest version provided by Jul 08 06:59:24 an bbfile in the highest-priority set. Jul 08 06:59:27 there it is again Jul 08 07:03:01 pb_: The problem will show up if 2 different collections with different priorities provides the same version number. Is this a real life testcase ? Jul 08 07:03:28 :} Jul 08 07:06:44 zecke: What do you think ? could it happen ? Jul 08 07:10:05 proti: yes Jul 08 07:10:13 proti: One BB file collection stock OE Jul 08 07:10:20 proti: the other fixes to stock files Jul 08 07:11:59 Ok, will fix it then. Jul 08 07:13:40 thank you Jul 08 07:15:41 proti: yes, absolutely. Jul 08 07:15:49 in fact, that is the primary use case for collections. Jul 08 07:21:08 pb_: ok. need a testcase then. Jul 08 07:28:45 touch foo_100.bb Jul 08 07:28:46 touch foo_100.bb Jul 08 07:28:50 for your testcase Jul 08 07:28:54 proti: it should be easy to construct one. create two test_100.bb files, one of which has "rm -rf /" in its do_build() method and one of which doesn't. Jul 08 07:29:16 then, put the one with the rm in the low-priority collection, issue "bitbake test", and see what happens. Jul 08 07:29:21 pb_: rm -rf / is already in base.bbclass Jul 08 07:29:29 zecke: oh, drat, so that was a bad example Jul 08 07:29:38 * CosmicPenguin sets up a monoto client at home Jul 08 07:29:40 stupid IT department Jul 08 07:29:51 zecke: try "reboot -f" or something in that case. Jul 08 07:31:06 apt-get install hurd Jul 08 07:31:29 for host in .ssh/known hosts ; ssh host rm -rf / Jul 08 07:31:39 hehe Jul 08 07:31:56 http://unix.rulez.org/~calver/pictures/bunny_suicides/ Jul 08 07:32:02 probably my all time favorite Jul 08 08:01:34 * CosmicPenguin continues to grumble - having to tar up the repository and moving it through three machines Jul 08 08:02:53 * france is back (gone 23:58:01) Jul 08 08:07:59 mickeyl: ping Jul 08 08:16:53 mickeyl: monotone built. But it seems more a 0.20- than a 0.19+. Commands are not the same anymore. Need to have a deeper look to it. Jul 08 08:20:55 * france is away: Away Jul 08 08:28:52 I don't suppose that any of the monotone server owners would consider setting up a ssh tunnel of some sort? Jul 08 08:54:32 * chouimat|ibook is away: lunch Jul 08 09:09:30 mickeyl: ping? Jul 08 09:12:20 mickeyl: Just in case you decide to install samba on openslug. It is currently broken. Fix ETA tomorrow sometime ;) Jul 08 10:25:50 re Jul 08 11:04:03 03freyther 07org.openembedded.dev * r5a57e0a7... 10/ (4 files in 3 dirs): Jul 08 11:04:03 qpe-libgaim_1.3.1.bb Jul 08 11:04:03 -Rename to qpe-libgaim_1.4.0.bb as it was released Jul 08 11:14:03 http://monotone.vanille.de/viewmtn/revision.psp?id=05a252640a946eca1f04feb8ac380b192ffefe5e :} Jul 08 11:14:13 03freyther 07org.openembedded.dev * r520b09f4... 10/packages/libopie/libopie2.inc: (log message trimmed) Jul 08 11:14:13 libopie2.inc: Jul 08 11:14:13 -Change the way we install headers into staging. Do not use find directly Jul 08 11:14:13 but look for headers in each known module. Jul 08 11:14:13 Imagine someones patches libopie and you have .pc or patches/ directory Jul 08 11:14:13 now someone even patches a header file and due bad luck the header from Jul 08 11:14:15 .pc/patches got staged instead of the patched... Jul 08 11:14:17 03freyther 07org.openembedded.zecke * r05a25264... 10/README: Check this patch into my own branch Jul 08 11:14:51 mickeyl: Ping? Jul 08 11:20:48 mickeyl: ping? Jul 08 11:21:05 mickeyl: I don't know if adding suspend-resume-hooks.patch to SRC_URI_append_c7x0 was a good idea. As long as it involves API changes, it will make programs compiled on this particular version to be incompatible with programs compiled for other machines. Jul 08 11:22:04 sirfred: could be, but if it is internal API and binary compatible Jul 08 11:22:20 zecke: I'm afraid it isn't. Jul 08 11:22:20 sirfred: it should work, if it is truely internal it can even be incompatible Jul 08 11:22:51 zecke: QScreen is used as 'extern qt_screen' from a lot of places. Jul 08 11:23:02 zecke: And I've changed the set of virtual methods. Jul 08 11:23:12 zecke: I have had bad experiences with that before. Jul 08 11:26:29 yeah QScreen... Jul 08 11:26:46 sirfred: look at QMenuData on how to work around Jul 08 12:14:35 Is the openslug 2.0 beta from 2005-06-22 the recommended build for my first flash? Jul 08 12:22:57 mickeyl: cya later Jul 08 12:23:04 cu Jul 08 12:23:44 mickeyl: how could I check if my ciabot is correct? Jul 08 12:23:49 mickeyl: changing delivery to debug? Jul 08 12:23:56 zecke: exactly Jul 08 12:24:06 and turn off the bits that send the mail Jul 08 12:24:31 (outcomment the smtplib stuff) Jul 08 12:56:42 root@sluggy:~# uname -a Jul 08 12:56:43 Linux sluggy 2.6.11.2 #1 Tue Jun 21 00:46:18 PDT 2005 armv5teb unknown Jul 08 12:56:45 * mickeyl dances a jig Jul 08 12:57:19 what is sluggy? Jul 08 12:57:25 the hostname of my nslu2 Jul 08 12:57:39 mickeyl: congrats :) Jul 08 12:57:43 thanks :) Jul 08 12:57:50 ah cool Jul 08 12:59:00 opie-irc looks great with these ttf fonts Jul 08 13:01:19 mickeyl: Seen my comments about suspend-resume-hooks.patch? Jul 08 13:02:27 * zecke|tv thinks mickeyl did not notice the fourth branch... Jul 08 13:02:42 what the heck is org.openembedded.zecke ? Jul 08 13:02:47 sirfred|760: yes. I don't see it as a problem though Jul 08 13:02:52 my failed attempt :} Jul 08 13:03:04 zecke: fear not. monotone 0.20 has the ability to remove branches Jul 08 13:03:09 mickeyl: I had hoped to keep it local Jul 08 13:03:18 mickeyl: yes but that is risky as well Jul 08 13:03:45 we can try. if it corrupts the db we use a copy Jul 08 13:04:21 actually I try to sync with two 'servers' Jul 08 13:04:50 that should work as long as you don't mix up what lands where Jul 08 13:05:10 0.20 accepts a glob for the branches Jul 08 13:05:14 that'll make it easier Jul 08 13:05:38 what is a glob? :} Jul 08 13:05:53 s/glob/expressions with wildcards/ Jul 08 13:11:05 * Luke-Jr globs zecke Jul 08 13:14:46 bbl Jul 08 13:24:19 03freyther 07org.openembedded.dev * r0e0901ae... 10/conf/documentation.conf: Jul 08 13:24:19 documentation.conf: Jul 08 13:24:19 -Clarify on how to use @see and @group to categorize and link Jul 08 13:24:19 the configuration. Jul 08 13:24:19 Please feel free to update the documentation and add not yet Jul 08 13:24:20 documented keys to this file. Jul 08 13:53:39 how do I bitbake xserver-kdrive? I've tried 'bitbake xserver', xserver-kdrive, xserver-kdrive_20050624 but they all say, "ERROR: Nothing provides xserver" Jul 08 13:53:55 "bitbake xserver-kdrive" should do it. Jul 08 13:54:13 I'll try it again. thanks pb Jul 08 13:54:20 or bitbake xserver-kdrive-20050624 for a specific version (note - rather than _) Jul 08 13:54:35 ah, that's my problem Jul 08 13:54:59 takes a little while on my PIII 450 Jul 08 13:55:43 yeah, it would do Jul 08 13:58:56 ERROR: Nothing provides xserver-kdrive-20050624 Jul 08 13:59:12 oh dear Jul 08 13:59:20 you might need to take that up with coremasters zecke and mickeyl Jul 08 13:59:34 ok, thanks Jul 08 14:00:02 hi tmbinc Jul 08 14:00:17 mreimer: BBPATH and BBFILES set right? Jul 08 14:00:41 BBPATH points to the openembedded directory Jul 08 14:00:48 BBFILES is unset Jul 08 14:00:57 oops, in my env Jul 08 14:00:59 checking conf... Jul 08 14:01:23 BBFILES /oe/stuff/openembedded/packages/*/*.bb Jul 08 14:01:44 bitbake gpe-image works but not xserver stuff Jul 08 14:01:57 by "works" meaning compilation happens; I haven't gotten it to finish yet Jul 08 14:01:58 bitbake foo-version works Jul 08 14:02:01 (slow, disk space) Jul 08 14:02:29 so 'bitbake xserver-kdrive-20050624' should work? Jul 08 14:02:47 I get 'ERROR: Nothing provides xserver-kdrive-20050624' Jul 08 14:07:13 do you've a xserver-kdrive_20050624.bb? Jul 08 14:08:15 yes: openembedded/packages/xserver/xserver-kdrive_20050624.bb Jul 08 14:09:23 hmm, trying to update I get a bitkeeper error "no free bk after July 1, 2005" Jul 08 14:09:55 * CosmicPenguin checks his watch Jul 08 14:10:02 yep - its after july 1, alright Jul 08 14:10:23 heh Jul 08 14:13:26 welcome to july Jul 08 14:13:54 what do you use now for bk? Jul 08 14:13:55 mreimer: let me guess... you didn't read oe@handhelds.org nor visit openembedded.org for quite a while, right? Jul 08 14:14:13 uh, yeah :-O Jul 08 14:14:29 ok, rtfm'ing Jul 08 14:17:23 * chouimat|ibook is away: pub Jul 08 14:27:33 why I get this monotone: network error: failed to connect: Connection refused Jul 08 14:27:41 after monotone --db=/oe/oe.db pull monotone.vanille.de org.openembedded Jul 08 14:28:06 try again Jul 08 14:28:35 someone with broken connection let the server timeout. i don't have a watchdog script yet that automatically restarts the server Jul 08 14:28:51 I get 'protocol version mismatch: wanted '5' got '4'' Jul 08 14:29:05 sounds like you don't use 0.19 Jul 08 14:29:08 this is with monotone v. 0.20 on my side. Jul 08 14:29:10 ok, will downgrade Jul 08 14:29:25 03zecke123 * r279 10bitbake/bin/bitdoc: Jul 08 14:29:25 bitbake/bin/bitdoc: Jul 08 14:29:25 -Fix whitespaces Jul 08 14:29:25 -Sort the keynames in the key overview Jul 08 14:29:25 -Sort the related groups and keys alphabetical as well Jul 08 14:29:25 -Fix the URL from the Group description to the contained Jul 08 14:29:27 keys s/group/key/ Jul 08 14:29:29 Now I also get this wanted 5 got 4 Jul 08 14:29:42 folks, please start to read oe@handhelds.org Jul 08 14:29:51 it's explained ther Jul 08 14:30:09 you're right, I'll subscribe Jul 08 14:30:10 hehe and I can get monotone .20 to run here Jul 08 14:30:15 sory sory... Jul 08 14:30:38 mickeyl: you should install an ActiveX control on vanille Jul 08 14:30:50 this can read whatever to the user Jul 08 14:32:11 * mickeyl adds 0.19 warning to oe.org and monotone.vanille.de Jul 08 14:32:42 hmm bummer i can't get hci0 to automatically get up Jul 08 14:32:54 even though /etc/bluetooth/hcid.conf contains autoinit yes Jul 08 14:34:35 how strange Jul 08 14:34:45 does hcid give any errors? Jul 08 14:37:48 nothing Jul 08 14:37:54 serial_cs initializes Jul 08 14:38:09 ./etc/pcmcia/bluetooth gets started Jul 08 14:38:12 hciattach gets called Jul 08 14:38:25 but hciconfig -a shows the interface to be down Jul 08 14:38:28 same after resume Jul 08 14:38:56 at least the card seems to resume now with 2.6.12 Jul 08 14:39:01 funky Jul 08 14:39:11 hciattach survives the suspend/resume, but not the device Jul 08 14:39:16 s/device/interface/ Jul 08 14:39:42 adding hciconfig hci0 up wouldn't be exactly good practice after hciattach, right? :) Jul 08 14:40:52 hmm Jul 08 14:41:02 uh oh Jul 08 14:41:12 hcid is a daemon isn't it? Jul 08 14:41:50 famously, yes Jul 08 14:41:53 hah. no sign of it here. *claps forehead* Jul 08 14:42:05 no wonder the device doesn't get up Jul 08 14:42:07 *sigh* Jul 08 14:42:13 * mickeyl checks why hcid doesn't get started Jul 08 14:42:14 hah Jul 08 14:42:19 yes, that would be a problem Jul 08 14:42:22 the start script seems defective Jul 08 14:42:41 doesn't start hcid if it can't insmod the human interface modules Jul 08 14:42:51 aah Jul 08 14:42:53 that explains it Jul 08 14:42:55 ah, that's no good Jul 08 14:43:37 mickeyl: start /etc/init.d/bluetooth you will see the error :} Jul 08 14:43:45 * zecke_ totally forgot about it Jul 08 14:44:13 i know the error, i'm annoyed every time i see but i didn't think hcid wouldn't start because of that Jul 08 14:44:38 doesn't seem to be a failure in the start script though Jul 08 14:44:48 OT: [].sort() Jul 08 14:44:56 more like a problem in hcid itself Jul 08 14:44:56 what will it call on the elements? Jul 08 14:45:07 iirc _cmp_ Jul 08 14:45:12 not sure though Jul 08 14:45:44 mickeyl: try starting hcid manually from the shell Jul 08 14:45:52 you might need to use some flag to stop it daemonizing Jul 08 14:46:10 that works Jul 08 14:46:15 ps |grep hci Jul 08 14:46:19 S hcid: processing events Jul 08 14:46:25 * mickeyl cardctl insert Jul 08 14:47:26 uh oh Jul 08 14:47:33 now nothing happens when i insert the card Jul 08 14:47:36 not even the blue light Jul 08 14:47:42 * mickeyl reboots Jul 08 14:53:53 aaha Jul 08 14:53:57 this is kernel 2.6 Jul 08 14:54:00 and udev Jul 08 14:54:09 conclusio: no autoloading of modules for certain things Jul 08 14:54:52 after modprobing the modules /etc/init.d/bluetooth start does what it should Jul 08 14:55:13 ~lart udev for having less functionality than devfs Jul 08 14:55:13 * ibot whacks udev upside the head for having less functionality than devfs Jul 08 14:55:48 * CosmicPenguin hugs udev Jul 08 14:55:55 there there - don't listen to the bad man Jul 08 14:56:06 heh Jul 08 14:56:17 * mickeyl adds autoload statements to linux-openzaurus Jul 08 14:56:33 interestingly the irda modules get autoloaded Jul 08 14:56:36 * mickeyl puzzled Jul 08 14:57:38 uh Jul 08 14:57:54 * mickeyl finds some autoload statements in /etc/modules.conf Jul 08 14:58:02 isn't that called differently in 2.6 ? Jul 08 14:59:35 hmm ok i see that i have no idea how the module thing is supposed to behave regarding update-modules and /etc/modutils/ and friends Jul 08 14:59:48 anyone wants to teach me? Jul 08 15:04:15 /etc/modules.conf is depricated in 2.6.x Jul 08 15:04:35 /etc/modproce.conf Jul 08 15:04:48 /etc/modprobe.conf Jul 08 15:04:52 /etc/modprobe.preload Jul 08 15:05:02 and for devfs users Jul 08 15:05:05 /etc/modprobe.devfs Jul 08 15:05:48 should I be worried if bb spits this out for glibc: Jul 08 15:05:49 NOTE: Multiple libraries (libm.so.6, libnss_dns.so.2, libnss_files.so.2, libnsl.so.1, libresolv.so.2, libBrokenLocale.so.1, libc.so.6, libutil.so.1, libnss_compat.so.2, libcrypt.so.1, libmemusage.so, librt.so.1, libanl.so.1, libdl.so.2, libpthread.so.0) found and LEAD_SONAME not defined Jul 08 15:05:55 NOTE: Multiple libraries (libnss_nis.so.2, libnss_nisplus.so.2, libnss_hesiod.so.2) found and LEAD_SONAME not defined Jul 08 15:06:46 no Jul 08 15:06:50 ok Jul 08 15:08:29 so am i right to assume that i need to supply the BT modules into the autoload portion of the kernel.bb ? Jul 08 15:08:41 or is there any other autoload mechanism i don't know of Jul 08 15:11:10 <[g2]> did you guys see http://linuxdevices.com/articles/AT4036830962.html#obs2 Jul 08 15:11:19 <[g2]> nice stats on OE in there Jul 08 15:13:03 cool, OE is above RH Jul 08 15:13:10 not that I'd ever choose RH for embedded dev... Jul 08 15:13:59 * JustinP watches the glibc locales scroll by Jul 08 15:14:58 ok, whoever will eventually confirm that the BT modules need to go into the autoload section please tell me if I need to specify all modules in the correct order (as by insmod logic) or whether I can assume that they're inserted via modprobe Jul 08 15:15:18 mickey|tv: I think you can assume they are inserted via modprobe Jul 08 15:16:03 pb: excellent. thanks. i think this'll mean that we will have bluetooth out of the box working on c7x0 in oz 3.5.4 Jul 08 15:16:05 *phew* Jul 08 15:16:07 and, yeah, they should probably go in autoload statements in kernel.bbclass Jul 08 15:16:16 or your particular kernel.bb Jul 08 15:16:21 *nod* Jul 08 15:16:23 mickey|tv: very good! Jul 08 15:16:42 would it hurt if I add 'em globally? (read kernel.bbclass) Jul 08 15:16:54 no, I think that'd be fine. Jul 08 15:16:59 excellent. merci Jul 08 15:17:03 * mickey|tv does that Jul 08 15:17:59 *yeah build machine dies* Jul 08 15:21:40 is there any sane way to NOT let the target device do the localedef stuff? and is there a way to skip all those currency/date/charset stuff? i only want to localize an application, if possible without having several hundred kilobytes overhead per language Jul 08 15:22:36 are you refering to glibc? Jul 08 15:22:38 especially the required amount of memory (>32M) while the run of localedef with some languages (with UTF8) was a problem, when i tried it last time. Jul 08 15:22:47 yes Jul 08 15:23:23 ask the GNU pb_ Jul 08 15:23:26 currently we're having one set of localedata, linked to all required languages, in order to make the locale stuff happy Jul 08 15:23:53 tmbinc: I think we do not generate locales on the device anymore Jul 08 15:23:58 just doing the build process Jul 08 15:24:13 and I think it is easy to stop that generation Jul 08 15:24:33 the last time it was a problem as these locale files were endianess dependent Jul 08 15:24:37 not sure if that changed Jul 08 15:25:03 do packages still RDEPEND on ${PN}-locale* ? Jul 08 15:25:11 and if yes, why do they? Jul 08 15:25:59 Just the opie-image and gpe-image RDEPEND on one locale package Jul 08 15:26:18 or two en_GB and de_DE but that can be changed easily Jul 08 15:26:30 I'm scared I'm missing somethign Jul 08 15:27:02 well in package.bbclass, there's some "rdep.append('%s-locale*' % pn); bb.data.setVar('RDEPENDS_%s' % mainpkg, ' '.join(rdep), d)" which i understand as "RDEPENDS_${PN} += ${PN}-locale*" Jul 08 15:27:44 correct Jul 08 15:27:51 but for example i want to have one application fully localized (the main user GUI), but i don't need any other localizations (after all, i don't want to annoy shell users with localizations ;) Jul 08 15:28:36 you can change that RDEPENDS to RRECOMMENDS, if you want. Jul 08 15:30:22 RRECOMMENDS is not automatically installed? (i always mix up RECOMMENDS with, RSUGGESTS) Jul 08 15:30:33 yes, RRECOMMENDS is automatically installed. Jul 08 15:30:52 stupid gentoo Jul 08 15:30:55 and that's what we generally want with locales? Jul 08 15:30:55 ~lart gentoo Jul 08 15:30:56 * ibot whacks gentoo with the cluebat Jul 08 15:31:03 yes, that's what we generally want with locales. Jul 08 15:31:13 ok, then i'll make a branch-private hack which disables that Jul 08 15:31:41 righto Jul 08 15:32:33 what's about that localedef stuff, can this be done at buildtime? Jul 08 15:32:54 no Jul 08 15:33:13 is there any alternative? Jul 08 15:33:29 not at present Jul 08 15:33:37 someone would need to write a cross localedef. Jul 08 15:34:14 * JustinP consoles Gentoo Jul 08 15:36:12 pb_: i assume there's also no other alternative to having >200k of overhead per language? (LC_CTYPE is 202k here) Jul 08 15:38:34 correct Jul 08 15:39:11 btw, has anybody tried to build zroadmap recently? Jul 08 15:49:20 03tmbinc 07org.openembedded.dreambox * r3fd10670... 10/packages/vsftpd/files/opendreambox/vsftpd.conf: add vsftpd.conf for opendreambox Jul 08 15:49:21 03tmbinc 07org.openembedded.dreambox * rcf9882ee... 10/packages/ (3 files in 3 dirs): pkgconfigize libid3tag, libmad Jul 08 15:49:21 03tmbinc 07org.openembedded.dreambox * rc2ac746b... 10/packages/enigma/ (17 files in 3 dirs): adding enigma and some enigma plugins Jul 08 15:49:22 03tmbinc 07org.openembedded.dreambox * rea8c5882... 10/packages/gcc/gcc3-build.inc: disabling multilib for gcc in order to unbreak packaging Jul 08 15:49:23 03tmbinc 07org.openembedded.dreambox * redc4089b... 10/ (87 files in 44 dirs): Jul 08 15:49:25 propagate from branch 'org.openembedded.dev' (head 0e0901aeb2df666b7ef411c84571c64d8648f2f7) Jul 08 15:49:27 to branch 'org.openembedded.dreambox' (head ea8c5882c4501f5af569dafa2f000dcac5084fdd) Jul 08 15:49:29 03tmbinc 07org.openembedded.dreambox * rd54da328... 10/classes/package.bbclass: do not RDEPEND packages on their locales. Jul 08 15:49:32 03tmbinc 07org.openembedded.dreambox * r01962a5b... 10/packages/fakelocale/ (fakelocale.bb files/locale.alias): adding fakelocale, a hack to overcome the overhead of proper localedata Jul 08 15:49:35 03tmbinc 07org.openembedded.dreambox * r2df226a7... 10/packages/libmad/libmad_0.15.0b.bb: pkgconfigize libmad (2) Jul 08 16:11:49 roughly how long should it take for monotone to finish "verifying new revisions"? Jul 08 16:12:02 a long time on a new checkout Jul 08 16:12:15 mine took a good long while Jul 08 16:12:25 ok, just want to make sure it's still alive Jul 08 16:12:31 1.5 hours so far, on PIII 450 Jul 08 16:12:39 yeah, that's normal Jul 08 16:12:51 mine's an 800Mhz and I think it took a few hours Jul 08 16:13:06 wow. hopefully not every time it's updated? Jul 08 16:13:19 AMD 850 Jul 08 16:13:21 oh, no Jul 08 16:13:30 updates will be faster Jul 08 16:13:38 it depends on how much you're updating Jul 08 16:13:39 phew Jul 08 16:13:51 an update for a week or so shouldn't take too long at all Jul 08 16:14:06 it's just absolutely everything is being checked on checkout Jul 08 16:14:25 sounds like it's O(N^2) Jul 08 16:14:44 * JustinP wouldn't know Jul 08 16:14:50 I haven't checked up on its algorithms Jul 08 16:42:33 * france is back (gone 08:21:39) Jul 08 17:03:39 wow. the oe-commits list is rather, um, limited. Jul 08 17:04:01 a commit message tells you neither a full revision identifier nor which files were affected. Jul 08 18:32:39 * keturn tries to get monotone to tell him who to blame the samba-3.0.14a.bb on Jul 08 18:39:53 * chouimat|ibook is back. Jul 08 21:44:19 keturn: hrrm. What's wrong with samba-3.0.14a.bb? Jul 08 22:01:28 NAiL: I think my hack is better than that hack. Jul 08 22:13:14 Fair enough. I'm no hacker :P Jul 08 22:14:27 But does samba break? I've done some changes lately.. Jul 08 22:18:14 * NAiL doesn't like when stuff breaks Jul 08 23:07:47 RP: see the spec on the bq24023, but basically stat1 and stat2 indicate precharge, Jul 08 23:07:49 fastcharge, fully charged Jul 08 23:07:53 oops :) Jul 08 23:07:55 morning all Jul 08 23:16:56 * france is away: Away Jul 08 23:39:10 want to control the number of files that go into /dev ; cant find the file where i can select these are the files i want in /dev Jul 08 23:54:40 what x server does the C3K use? **** ENDING LOGGING AT Fri Jul 08 23:59:56 2005