**** BEGIN LOGGING AT Fri May 27 23:59:56 2005 May 28 01:50:52 03mickeyl 07 * r1.3435.1.1 10openembedded/ (5 files in 2 dirs): use hardcoded SRC_URI for flex 2.5.31 - this source seems rare May 28 01:58:16 *yawn* May 28 01:58:16 morning May 28 01:58:52 morning. May 28 01:59:11 I just got fastcooked out of bed... pffff May 28 02:00:07 heh May 28 02:00:21 couldn't sleep any longer, t'was to warm May 28 02:00:29 too, even May 28 02:01:59 cheers a-atwood May 28 02:02:03 ~lart tab completion May 28 02:02:03 * ibot whips tab completion with a wet and grimy noodle just because May 28 02:02:05 hi Arjan May 28 02:02:08 ~lart tab completion again May 28 02:02:08 * ibot declares tab completion again a moron May 28 02:02:09 hir ar_ May 28 02:02:22 *g* too early for irc May 28 02:02:36 need to eat breakfast before May 28 02:02:38 hi mickeyl May 28 02:03:22 Hertog: did i miss your mail to oe@handhelds.org or didn't you send it yet ? May 28 02:07:46 mickeyl: stil working on it :) it wasn't kernel after all (according to RP) so no toying with gdb on Z. :) May 28 02:07:58 ah :) May 28 02:09:32 hmm, no toying = now toying :) May 28 02:14:02 yeah, i've corrected that online ;) May 28 02:14:27 bbl, breakfast May 28 02:21:58 good morning all May 28 02:22:03 morning koen May 28 02:22:53 morning May 28 02:22:57 hey XorA & Arjan May 28 02:46:28 * Hertog hates supporting crappy old hardware with this kind of weather *sigh* May 28 02:57:27 morning May 28 04:05:17 g'morning May 28 04:07:06 hey Twiun May 28 04:15:35 anyone played with the shell enough to comment on a policy question May 28 04:15:36 ? May 28 04:17:34 In the shell right now, the build command accepts both a provider or a direct bbfile. For the tab completion this is suboptimal, since it doesn't know if a provider or a bbfile is being looked after. I can ignore this and just return both providers and bbfiles until the user chose one or I could say let's use a different command for building a bbfile instead of a provider, i.e. bbuild. Opinions on that? May 28 04:18:11 having a dedicated command may be better anyway May 28 04:18:17 dunno May 28 04:34:54 mickeyl: I'd say having separate command is less confusing for users May 28 04:35:51 s/command/commands/ May 28 04:39:48 yeah, i agree. otherwise the subtle distinction between provider and package may tend to get mixed May 28 04:45:55 i may call it build for a provider and buildpkg for a package May 28 04:54:26 sounds like a good idea to me May 28 04:55:56 pb_: you remember our recent problem with _append being swallowed when in a local.conf? May 28 04:57:25 zecke told me to insert 'bb.data.update_data( make.cfg )' which i did in bitbake line 640 and this fixed the _append being unrecognized. however this statement is responsible for removing all the standard deps (libc, gcc, etc.) for a package then. Do you have any idea why and whether there would be a more appropriate place to call this? May 28 05:04:00 mickeyl: I'm not sure that applying overrides to the config data in isolation will work. I don't know offhand what would be causing your DEPENDS to disappear, though: those are added by some python code in base.bbclass. May 28 05:04:40 You were trying to fix overrides in PREFERRED_PROVIDER, right? May 28 05:04:52 yes May 28 05:05:14 You might need to take a copy of the data, run the overrides on the copy, and then pull the PREFERRED_PROVIDERS out of there. May 28 05:06:48 i see. so bb.data.update_data( make.cfg ) essentially runs through overrides? May 28 05:06:54 Right. May 28 05:07:18 It's a destructive process, though, so doing that when you don't yet have all the data on hand can cause bad results. May 28 05:10:10 bbiab, going to do some work in the garden May 28 05:10:15 have fun May 28 05:13:00 time to hit the town and enjoy the weather. May 28 07:52:49 ping May 28 07:53:08 bleh May 28 07:58:10 i have a small pb :) May 28 07:58:21 i try to bitbake opie-oytazee May 28 07:58:35 but OE told : ERROR: Nothing provides opie-oyatzee_1.2.0.bb May 28 08:00:03 you're not using correct syntax. May 28 08:00:04 rtfm. May 28 08:00:20 you want 'bitbake opie-oyatzee', not 'bitbake opie-oyatzee_1.2.0.bb' May 28 08:00:57 i know but the first gave me same error message May 28 08:01:11 so i try the second one May 28 08:01:18 and copy/paste last message May 28 08:02:59 ERROR: Nothing provides opie-oytazee May 28 08:17:10 Snoogie: fix your typo. May 28 08:17:25 oh well. if zecke shows up, I fell asleep. May 28 08:46:00 does anybody have a use for 2 amd 1200MP cpus? May 28 09:56:36 moin zecke May 28 09:56:49 zecke: 17:17 < ggilbert> oh well. if zecke shows up, I fell asleep. May 28 09:57:00 damn... May 28 09:57:15 it was too hot here to turn the computer on May 28 09:58:14 it was even too hot to sit outside and read a book May 28 10:38:47 mickey|away: ok got the botmail May 28 10:39:18 zecke: tinderbox is basically a task executor, right? May 28 10:40:17 koen: no. It summarizes and relates the works of others May 28 10:40:29 ah, ok May 28 10:40:29 koen: groups builds to commits to fixed bug numbers May 28 10:40:56 koen: so build was broken on client Solaris10, after jörg did the check in to fix bug #2500 May 28 10:41:15 koen: after bug #2501 the memleak in bitbake got reduced by 10kb on Debian Woody May 28 10:41:47 feature request: irc insult bot May 28 10:45:32 koen: we could do that, but let me get the basic features done May 28 10:45:47 my gal just left so I've a slight delay :) May 28 10:45:50 it *is* a basic feature ;) May 28 10:57:10 zecke: have you decided where to run the tinderbox server? May 28 10:57:21 03pb 07 * r1.3439 10openembedded/packages/dhcpcd/ (dhcpcd-1.3.22-pl4/paths.patch dhcpcd_1.3.22-pl4.bb): put dhcpcd in /sbin (per report from Jason Waterman) and correct packaging of manpage May 28 10:57:44 koen: no but the server itself should not produce 'much' load May 28 10:58:45 it would be nice to have a 'tinderbox.openembedded.org' May 28 10:58:57 but that would also mean it can't really run on hh.org May 28 10:59:13 koen: I've to get the tinder-'client' going, sandbox tinderbox (apache+suexec), fix viewcvs/svn, fix a signal problem in the VC_svn, implement mail receiving May 28 10:59:29 koen: I would be for tinderbox.handhelds.org May 28 11:00:04 koen: we could run tests for handhelds.org apps as well May 28 11:00:18 virtuall anything May 28 11:00:23 virtually* May 28 11:00:37 koen: it would even rock if we could autoflash an ipaq, boot it May 28 11:00:41 zecke: what's the issue with viewcvs? May 28 11:00:58 koen: and after 10 minutes get a screenshot ;) May 28 11:01:05 zecke: heh, that would be cute May 28 11:01:23 nfs/tftp boot would do as wel May 28 11:03:12 it would probably require a minor change to bootldr, but it shouldn't be too hard May 28 11:04:29 morning May 28 11:04:32 sorry zecke I fell asleep May 28 11:04:34 hey treke May 28 11:04:55 ggilbert even May 28 11:07:43 treke|home: hey May 28 11:07:55 this heat makes me sick :} May 28 11:08:03 hehe May 28 11:08:18 no kidding it upsets my stomach/coltis.. May 28 11:08:27 this weather is strange May 28 11:08:33 it's in the lows 70s here May 28 11:08:35 in may May 28 11:08:46 oh May 28 11:08:57 that sounds too low May 28 11:09:23 OT: did mickeyl fix the DEPENDS problem? May 28 11:09:24 oh sorry. mid 60s May 28 11:09:26 I was wrong May 28 11:09:40 update_data should be moved then May 28 11:09:53 koen: viewcvs is low on my list though May 28 11:10:09 koen: but pretty much is now 'known' and just a matter of being done... May 28 11:10:25 zecke: viewcvs for what? May 28 11:11:04 koen: you see 'zecke123' committed something, you click on 'zecke123' -> and then on view check in May 28 11:11:15 koen: and you go to viewcvs at svn.berlios.de May 28 11:11:26 ah, I understand May 28 11:12:21 koen: the issue is you get thrown to the wrong page ;) May 28 11:13:17 mickey|away: http://svn.berlios.de/viewcvs/bitbake/trunk/bitbake/bin/bitbake?rev=216&r1=213&r2=216 May 28 11:13:29 mickey|away: that seems too early, we need to call it after we've collected the bbfiles May 28 11:13:45 * zecke pets the X protocol... May 28 11:14:29 zecke: do you have a sample page online? May 28 11:15:23 koen: I could tar you the directory May 28 11:15:38 koen: it really runs on 127.0.0.1 only ;) May 28 11:15:57 or file -> save as in your browser :) May 28 11:16:09 or scap May 28 11:16:22 koen: but it looks quite empty ;) May 28 11:16:31 just time and commits for now as no client data arrived May 28 11:16:37 treke|home: do we want to postpone it? May 28 11:19:43 I think so. I need to get some stuff done and then take off for LA May 28 11:21:37 treke|home: tomorrow? What setup do you've KDE3.4 + kdepim cvs? May 28 11:21:41 treke|home: or even kdelibs cvs? May 28 11:22:07 tomorrow would work. I have kde3.4 and kdepim cvs May 28 11:22:17 installed in vmware so I can fairly easilly roll back May 28 11:22:25 that and not torch my desktop May 28 11:22:36 treke|home: ok I'll build kdepim cvs then and install it in /usr May 28 11:22:49 but I can build whatever it is you want me to build May 28 11:24:40 want me to build kdelibs svn? May 28 11:28:10 morning treke May 28 11:28:21 treke|home: you could start building May 28 11:44:54 someone here have experience with building a .bb file ? May 28 11:45:54 Sno_Away: I've not done by hand yet ;) May 28 11:47:09 i want to learn how build a .bb file May 28 11:47:10 but May 28 11:47:25 the guide in starting guide is a bit ... not enough :) May 28 11:47:25 Snoogie: build as in bitbake -b the.bb May 28 11:47:35 ok May 28 11:47:41 Snoogie: yeah docu needs to improve May 28 11:47:48 Snoogie: or do you want to create your own bb file? May 28 11:47:57 i want to creat my own :) May 28 11:47:59 [/opt/openembedded/packages/sbrsh]$ cat sbrsh_6.8.bb May 28 11:47:59 SRC_URI = "http://scratchbox.org/download/files/sbox-files/sbrsh-6.8.tar.gz" May 28 11:47:59 inherit autotools May 28 11:48:03 i create this one May 28 11:48:10 trying to compile sbrsh May 28 11:48:18 but i have a compile error May 28 11:48:32 i fix it by patching / moving some headers May 28 11:48:42 but i don't know how fix do_package May 28 11:48:46 i have binaries May 28 11:48:54 but i don't know how fix do_package procedure May 28 11:49:19 u don't know if there is a page who someone explain how he can port a tar.gz to .ipk succefully ? May 28 11:50:04 what' the problem with do_package? May 28 11:51:48 second ... relaunching it May 28 11:51:55 i have tried to modify the Makefile May 28 11:52:00 but seems to be a bad idea :) May 28 11:52:45 if it uses autotools you want to patch the makefile.in May 28 11:53:59 hey Hertog, how was de oude markt? May 28 11:54:29 koen: nice ;) altho the lack of eyecandy was a bit disapointing :) May 28 11:54:55 :) May 28 11:55:01 it's always the wrong people running around in summerdresses May 28 11:56:06 how clean all step i did for a package ? May 28 11:56:11 in order to start from zero ? May 28 11:56:29 bitbake -b file.bb -c clean May 28 11:56:34 thanks May 28 11:58:17 i ln -s manually a folder to my package folder in order to compile it succesfully (and modify a #include <....> to #include "...") May 28 11:58:23 is it good or a bad method ? May 28 11:59:26 it is a bad method, but if it works to get it started it should be fine May 28 11:59:31 ok May 28 11:59:55 if you want to submit the file to us you should see if there's another way May 28 12:00:06 it's entirely possible that's the only way, though May 28 12:00:15 Snoogie: the #include change can be right May 28 12:00:37 in fact my package need file May 28 12:00:49 so i just ln -s fakeroot folder to my package folder May 28 12:00:57 and change < & > to " May 28 12:01:18 humm maybe i can install fakeroot ... May 28 12:01:20 aha May 28 12:01:35 fakeroot doesn't stage it's headers May 28 12:01:48 stage ? May 28 12:01:57 (i am not english so sometime i don't understand ...) May 28 12:02:09 do_stage() install headers so other packages can use them May 28 12:02:34 in fakeroot package , right ? May 28 12:02:37 yes May 28 12:02:45 but this .bb file is from u May 28 12:02:55 yes, we have bugs :) May 28 12:02:58 ok May 28 12:03:11 so i open fakeroot_...bb May 28 12:03:12 ok May 28 12:03:38 for the moment i have this : May 28 12:03:40 SECTION = "base" May 28 12:03:40 DESCRIPTION = "Gives a fake root environment" May 28 12:03:40 HOMEPAGE = "http://joostje.op.het.net/fakeroot/index.html" May 28 12:03:40 LICENSE = "GPL" May 28 12:03:40 SRC_URI = "${DEBIAN_MIRROR}/main/f/fakeroot/fakeroot_${PV}.tar.gz" May 28 12:03:41 inherit autotools May 28 12:04:15 i add a do_stage section May 28 12:05:24 this is good ? : May 28 12:05:26 SRC_URI = "${DEBIAN_MIRROR}/main/f/fakeroot/fakeroot_${PV}.tar.gz" May 28 12:05:26 do_stage(){ May 28 12:05:26 install headers May 28 12:05:26 } May 28 12:05:26 inherit autotools May 28 12:05:58 see http://oe-devel.bkbits.net:8080/openembedded/patch@1.3434?nav=index.html|ChangeSet@-7d|cset@1.3434 May 28 12:06:01 for an example May 28 12:06:30 pb_: autotools_stage_includes didn't work work me, will look into why when I have some time May 28 12:07:37 righto May 28 12:11:04 ok my patch works :) May 28 12:11:09 i patch fakeroot May 28 12:11:12 i rebuild it May 28 12:11:18 so my package compile without pb :) May 28 12:11:26 phew May 28 12:11:27 eh, now I finaly got cardmgr to build with debugging info... Now to figure out what all the cruft in the structs mean :) May 28 12:12:02 compile crashed sorry May 28 12:12:05 another windows :) May 28 12:14:59 * zecke finds it amusing how the name crashed gets abused... May 28 12:15:10 sorry May 28 12:15:10 :) May 28 12:15:37 for my mother konqueror crashes as soon as one pixel is off by one when rendering... May 28 12:16:02 ERROR: function do_install failed May 28 12:16:02 ERROR: log data follows (/opt/build/tmp/work/sbrsh-6.8-r0/temp/log.do_install.9282) May 28 12:16:02 | NOTE: make DESTDIR=/opt/build/tmp/work/sbrsh-6.8-r0/image install May 28 12:16:02 | make: **no rule install ... May 28 12:16:19 no rule to build "install" target May 28 12:16:47 and i don't know how fix this May 28 12:18:03 Snoogie: so sbrsh does not have a install target? May 28 12:18:24 nope but in makefile there is 2 target : ipk and deb May 28 12:18:54 Snoogie: so either patch sbrsh to have a install target (and send it to the maintainer) May 28 12:18:58 Snoogie: or do install by hand May 28 12:19:09 i ll modify Makefile May 28 12:19:12 and try again May 28 12:23:36 NOTE: package sbrsh-6.8: completed May 28 12:23:44 how be sure my ipk file is OK ? May 28 12:24:08 Snoogie: tmp/deploy/ipk May 28 12:24:12 Snoogie: and check the content? May 28 12:24:14 not inside May 28 12:24:41 i got this package sbrshd_6.8_arm.ipk in /opt/build/tmp/work/sbrsh-6.8-r0/sbrsh-6.8 May 28 12:24:55 there is a command to see what is inside a package ? May 28 12:25:04 man ipkg May 28 12:25:05 :) May 28 12:25:11 ar x file.ipk ; tar zxvf data.tar.gz May 28 12:26:26 I use midnight commander May 28 12:27:35 ok i understand May 28 12:32:42 Any chance I can mix and match kernels? I've got modules for both... May 28 12:32:47 (in the image) May 28 12:33:12 In particular, I want to use the old 2.4 kernel w/ my new OE builds May 28 12:35:14 luke-jr_, you been using linux long? May 28 12:39:39 emte: ... May 28 12:39:44 at least 5 years May 28 12:40:24 and binary mixing hasn't always been fun May 28 12:40:37 then i think you can probably answer that question for yourself May 28 12:40:54 you should know the common barriers May 28 12:41:16 I'm hoping the biggest problem will be the init scripts whining May 28 12:42:04 no, depending on what your mixing there will be far more than that May 28 12:42:25 such as? May 28 12:42:53 I'm going after the older 2.4 kernels where the microphone still worked May 28 12:43:20 is the new version using 2.4 or 2.6? May 28 12:43:25 2.6 May 28 12:43:31 I can't compile 2.4 May 28 12:43:47 then your going to have a huge anmount of issues May 28 12:43:50 I have the 2.4 modules, tho May 28 12:43:59 the 2.6 modules wont work with 2.4 May 28 12:44:27 "I've got modules for both..." (in original question) May 28 12:44:30 if you cant compile the 2.4 there may be a good reason May 28 12:45:07 you might be farther ahead to figure out why 2.4 doesnt compile May 28 12:45:15 GCC 2.95 doesn't support my arch May 28 12:45:34 not a whole lot I can do about that ;) May 28 12:45:52 short of setting up OE on another box May 28 12:46:23 your target or host?: May 28 12:46:32 host May 28 12:47:15 wasnt aware that gcc actually cared May 28 12:47:44 luke-jr_: yes, mixing kernels should work fine if you have all the modules on hand. as you said, you will need to defeat the WARNING in the initscripts, but other than that it should be ok. May 28 12:48:10 pb_: and IIRC, init scripts can easily be changed on the Z after I wait out the delay May 28 12:48:27 emte: 2.95 certainly seems to... May 28 12:49:26 yeah i know quite a few of the 2.95 versions were broken May 28 12:49:49 but not that they were arch depenant May 28 12:50:59 if they're not arch-dependent, I wouldn't expect configure to complain... so I'm assuming they are May 28 12:51:08 pb_: thx, btw May 28 12:51:25 well RH shipped a few distros with the broken gcc May 28 12:51:37 couple others did as well to May 28 12:51:39 too* May 28 12:52:10 hard to say May 28 12:53:26 When is bitbake.conf loaded? May 28 12:53:40 and/or where is the results cached? May 28 12:57:41 the cache dir? May 28 12:58:31 luke-jr_: bitbake.conf is not cached May 28 12:59:07 have u try to build tslib ? May 28 12:59:35 I'm currently at the cross compiler, I will get to tslib soon May 28 12:59:44 i have found a bug May 28 12:59:47 and a way to fix May 28 12:59:52 if u have a pb tell me May 28 13:00:19 Snoogie: what bug? May 28 13:00:53 *yawn* May 28 13:01:06 zecke: well, it seems to be for me... May 28 13:01:08 morning ljp May 28 13:01:15 evening! :) May 28 13:01:23 in particular, IMAGE_NAME May 28 13:01:27 man, getting cold here May 28 13:01:37 or DATETIME May 28 13:01:40 9c! May 28 13:01:56 or IMAGE_CMD_jffs2 May 28 13:02:03 etc May 28 13:02:36 all that stuff is stored in the CACHE. if you've edited bitbake.conf, touch local.conf to invalidate the CACHE and rerun bitbake. May 28 13:02:50 ljp: crumbs, better wrap up warm. May 28 13:02:51 luke-jr_: persistent data (the bbfiles) have a copy of bitbake.conf in them (inherited) May 28 13:03:19 pb_: I'm not editing it, but DATETIME changes on its own May 28 13:03:34 zcke : rpl_malloc May 28 13:03:40 zecke: so what would be the best way to kill the caches for the images? May 28 13:03:44 zecke : rpl_malloc in main.c May 28 13:04:12 preferably w/o deleting the entire cache May 28 13:04:12 u just need to comment the line in config.h who #define malloc rpl_malloc May 28 13:04:15 luke-jr_: build everyting in one run? May 28 13:04:40 zecke: I do more than one build in the time that would take =p May 28 13:05:35 luke-jr_: modify bitbake.conf to use "=" rather than ":=" when assigning to DATE and TIME. May 28 13:06:05 bitbake will run a little bit more slowly, but that should stop DATETIME from being cached. May 28 13:06:38 pb_: any reason not to have that change in BK? May 28 13:06:56 no strong reason, but the caching doesn't bother most people. May 28 13:07:14 well, every time I build an image, it overwrites the old one :| May 28 13:07:28 Snoogie: what setup do you've? May 28 13:07:41 would seem to be a bug, IMO May 28 13:07:41 i just modify the config.h :) May 28 13:08:03 maybe a stupid question, but where i can found all last .bb ? May 28 13:08:18 Snoogie: why is it needed, and why is it not set for you? May 28 13:08:26 Snoogie: what does config.log say? May 28 13:08:47 before i patch config.h May 28 13:08:49 | ts_test.o(.text+0x418): In function `main': May 28 13:08:49 | ts_test.c: undefined reference to `rpl_malloc' May 28 13:08:49 | fbutils.o(.text+0x278): In function `open_framebuffer': May 28 13:08:50 | fbutils.c: undefined reference to `rpl_malloc' May 28 13:08:53 pb_: I get errors from bitbake w/t hat change :( May 28 13:09:01 What does it say? May 28 13:09:09 NOTE: exceptions.NameError:name 'time' is not defined while evaluating: May 28 13:09:09 ${DATE} May 28 13:09:15 NOTE: exceptions.NameError:name 'time' is not defined while evaluating: May 28 13:09:15 ${@time.strftime('%Y%m%d',time.gmtime())} May 28 13:09:41 wow, that's pretty freaky. May 28 13:10:02 those two May 28 13:10:04 you might need to refer to ubercoremeister zecke to find out why "time" isn't in scope there. May 28 13:11:16 zecke: any ideas? May 28 13:11:52 Snoogie: not the answer I'm looking for May 28 13:12:00 luke-jr_: because import time is not done May 28 13:12:05 zecke : sorry my english is broken May 28 13:12:28 Snoogie: I want to know MACHINE, DISTRO, CVSDATE... May 28 13:12:41 zecke: any idea what module of bitbake to add it to, though? May 28 13:12:57 pb_: the problem is we've not defined which names should be known May 28 13:13:34 luke-jr_: go to lib/bb/make.py May 28 13:13:46 luke-jr_: /Except (assuming you use vim) May 28 13:14:02 luke-jr_: uncomment except Exception: and the line below, rerun May 28 13:14:05 zecke: yah. I'm a bit bemused that it would work in the immediate-assignment case, and fail in the deferred one, though. May 28 13:14:15 if anything, I would have expected it to be the other way round May 28 13:14:32 pb_: we evaluate that stuff at multiple points May 28 13:14:41 some have time, some not May 28 13:15:08 oh, right. so I guess the parser has time included. May 28 13:15:15 yeah, that would make sense. May 28 13:15:24 where should I send the patch when I'm done? May 28 13:15:37 luke-jr_: just paste me the bt... May 28 13:15:50 or traceback as python calls it May 28 13:16:09 Snoogie: tslin just built here May 28 13:16:20 tslib :) May 28 13:16:35 ;) yeah, it is hot and my eyes are burning :} May 28 13:16:39 MACHINE=PIV,DISTRO=UBUNTU(DEBIAN),CVS=20050518 May 28 13:16:45 http://pastebin.ca/12860 May 28 13:17:15 Snoogie: still the wrong answer (I'm refering to the content of conf/local.conf) May 28 13:17:23 sorry :( May 28 13:17:38 Snoogie: (the target you want to create) May 28 13:17:43 SH3 May 28 13:17:45 jornada 680 May 28 13:18:01 # object files. Also, don't build glibc itself with these May 28 13:18:01 # flags, or it'll fail to build. May 28 13:18:01 # May 28 13:18:01 # PROFILE_OPTIMIZATION = "-pg" May 28 13:18:02 # SELECTED_OPTIMIZATION = "${PROFILE_OPTIMIZATION}" May 28 13:18:04 #LDFLAGS_append = " -static" May 28 13:18:08 # Uncomment this to enable parallel make. May 28 13:18:08 Snoogie: ok, tslib autotools magic might be broken May 28 13:18:10 # This allows make to spawn mutliple processes to take advantage of multiple May 28 13:18:12 # processors. Useful on SMP machines. This may break some packages - we're May 28 13:18:14 # in the process of marking these so let us know if you find any. May 28 13:18:15 ... May 28 13:18:16 Snoogie: can you stop that? May 28 13:18:16 # PARALLEL_MAKE = "-j 4" May 28 13:18:18 # Uncomment this if you want BitBake to emit the log if a build fails. May 28 13:18:20 BBINCLUDELOGS = "yes" May 28 13:18:22 please tell me he's not pasting the entire local.conf here May 28 13:18:22 # Specifies a location to search for pre-generated tarballs when fetching May 28 13:18:24 # a cvs:// URI. Outcomment this, if you always want to pull directly from CVS. May 28 13:18:26 CVS_TARBALL_STASH = "http://www.treke.net/oe/source/" May 28 13:18:28 SOURCEFORGE_MIRROR="http://heanet.dl.sourceforge.net/sourceforge" May 28 13:18:30 #include 'conf/machine/tune-sh3.conf' May 28 13:18:32 #EDIT THIS FILE and then remove the line below before using! May 28 13:18:33 someone kick him please May 28 13:18:34 sorry May 28 13:20:12 zecke: http://pastebin.ca/12860 May 28 13:21:05 Maybe stupid question, but would/could optimization of (c-)code shuffle the initialisation of local vars in a function? May 28 13:21:35 Hertog: shuffle independant vars May 28 13:22:44 they get initialized as soon as the compiler thinks they might be needed? May 28 13:24:15 Hertog: could be May 28 13:25:10 luke-jr_: try with a new bitbake May 28 13:26:40 zecke: thanks May 28 13:28:52 OT: can one create user-'aliases' with exim? May 28 13:29:02 or do I need to create multiple user accounts? May 28 13:29:07 and place .foward May 28 13:31:18 you can surely do aliases. It might even be as easy as editing /etc/aliases, although there might be a trick to it. May 28 13:33:16 keturn: thank you May 28 13:33:30 keturn: is that part of POSIX.3? May 28 13:33:48 I don't know. May 28 13:38:00 usually edit the aliases file and run newaliases, IRIC May 28 13:38:01 IIRC* May 28 13:40:42 Hertog: yes, the compiler will rearrange local variables any way it likes (including deleting them) as long as it thinks the results will not be visible. May 28 13:42:16 zecke: yeah, either /etc/aliases or multiple accounts and .forward will work. May 28 13:43:38 pb_: thank you, I assume aliases is better May 28 13:43:54 probably, yes May 28 13:44:12 if you make accounts, there's the possibility that the accounts will get logged in to, etc. May 28 13:44:50 this kdepim is a build nightmair May 28 13:45:41 keturn: no risk of that, I think zecke sets the shell for all accounts to /opt/QtPalmtop/bin/qpe May 28 13:46:04 treke|home: hell May 28 13:46:55 it's like it's half written in qt2 May 28 13:46:58 and half in qt3 May 28 13:47:22 that can't be good May 28 13:48:54 pb_: actually it is /bin/zsh, as nobody can't cope with that ;) May 28 13:51:13 the strange bit is it uses qt2 headers May 28 13:51:21 but compiles fine when you change the includes to qt3 May 28 13:52:01 treke|home: kdepim or kdepimpi? May 28 13:52:04 pi May 28 13:52:10 O.o May 28 13:52:14 pb_: hey is it 8 yet there? May 28 13:52:31 what's that kde fixation with 3.14? May 28 13:52:37 treke|home: the time, you mean? it's nearly 22:00 May 28 13:52:43 koen: ?? May 28 13:52:45 pb_: cool May 28 13:52:47 * zecke currently compiles kdepim svn May 28 13:52:52 pim/pi May 28 13:52:54 ko/pi May 28 13:52:59 bla/pi May 28 13:53:06 I don't think it's pi the # May 28 13:53:15 KDE PIMP I (polish notation) May 28 13:53:17 Nor is that stuff officially KDE, IIRC May 28 13:54:02 Hmm May 28 13:54:10 DATE/TIME = instead of := didn't fix it May 28 13:54:22 so my weekly dr who fix should be coming soon May 28 13:54:52 PI = Platform Independent May 28 13:54:58 == doesn't require kdelibs =p May 28 13:55:03 it sure is May 28 13:55:13 it's 3.14 + something everywhere May 28 13:55:19 treke|home: heh, right May 28 13:55:26 koen: it's based off of 3.1.4 May 28 13:55:32 it isn't stock though as far as I can tell May 28 13:55:34 1592653589793238462643383279502884197169399 May 28 13:55:35 it's hacked up a bit May 28 13:56:35 evalf(Pi,999999); May 28 13:56:42 if you want to keep maple busy for a while May 28 13:56:43 so... OE supports Qt/X11 for kde*/pi stuff now? May 28 13:56:55 or some other reason for the Qt 3 involvement May 28 13:56:58 ? May 28 13:57:14 no idea May 28 13:57:27 I stopped using kde when I stopped using rpm based distros May 28 13:57:43 I use KDE on my desktop boxen May 28 13:57:49 last I checked we compiled kdepim/pi against qt2 in oe May 28 13:57:50 somewhere around mandrake 6 May 28 13:57:53 and Kphone is the only VoIP app that works on the Z May 28 13:58:03 minisip works May 28 13:58:07 treke: Hence why I was curious with your using Qt3 May 28 13:58:15 koen: Not with c760 kernels May 28 13:58:43 luke-jr_: I didn't say anything about building it with oe... May 28 13:58:48 c760 will only do both audio in and out if you open /dev/dsp once for both May 28 13:59:02 treke: I know. That's why I was asking and not stating it as a fact May 28 13:59:09 c700 here May 28 13:59:23 koen: what kernel? May 28 13:59:32 2.6.something May 28 13:59:38 Interesting May 28 13:59:43 I can't get mic to work under 2.6 at all May 28 13:59:59 koen: you built the new nokia stuff yet? May 28 14:00:00 c700 doesn't have a mic May 28 14:00:09 O.o May 28 14:00:12 How are you using minisip? May 28 14:00:12 treke|home: built yes, tested no May 28 14:00:23 receiving only :) May 28 14:00:27 ah May 28 14:00:32 if it compiles it's done May 28 14:00:36 if it runs, it's perfect May 28 14:00:55 I need to put the navigator in OE to make it functional May 28 14:01:30 koen: so they've released apps as well? May 28 14:01:55 the framework apps May 28 14:02:02 not the real candy May 28 14:02:29 maemo.org is a bit of a mess atm May 28 14:02:47 it seems they needed another month or so but didn't get it May 28 14:04:10 well, I'd say the deadline was effectively set by guadec May 28 14:04:26 * zecke moves downstairs May 28 14:04:33 reenoo_: Florian is poking people at guadec for an update on the source issue May 28 14:05:33 koen: "the source issue"? source code to applications? May 28 14:05:51 and the kernel May 28 14:06:14 ah, cool. fingers crossed then. May 28 14:06:24 and of course the dsp code May 28 14:06:29 It doesn't use Linux? May 28 14:07:07 it does May 28 14:07:12 2.6-omap May 28 14:07:19 so how are they getting away w/o releasing src? May 28 14:07:34 the item isn't on sale yet May 28 14:07:39 ah May 28 14:07:56 I wonder if they will actually comply May 28 14:07:59 and include the notice May 28 14:08:16 I think they will May 28 14:08:33 they did so far May 28 14:08:59 I wonder if Sharp included notices May 28 14:09:05 I don't recall seeing one w/ my c760 May 28 14:09:15 but then again, it could have been in Japanese-- and I wouldn't have noticed that May 28 14:10:00 personally, I wouldn't buy one even if they were on sale. no phone capabilities and I've got an iPAQ already... May 28 14:10:25 heh May 28 14:10:48 but you wouldn't send it back if you got it for free either :) May 28 14:11:27 heh. no, certainly not. May 28 14:12:37 If I get the discount I'll probably sell some stuff and buy a 770 May 28 14:13:00 developer discount? May 28 14:13:05 yep May 28 14:13:23 E 100 is pretty cheap May 28 14:14:21 It's a very small chance, we'll see May 28 14:15:46 i just hate the fact their gtk and gnome-vfs stuff is patched to hell May 28 14:15:58 you need their version to run stuff May 28 14:28:10 how badly did they mangle gtk? May 28 14:28:59 stylus fixes, policy changes + 70 other patches May 28 14:29:37 one patch/file instead of groups..... May 28 14:37:59 * reenoo_ rotfl May 28 14:38:14 "Of course, its only missing Mono for extended productivity." - Miguel on maemo May 28 14:38:48 where did you read that? May 28 14:38:51 pgo? May 28 14:39:03 http://tirania.org/archive/2005/May-25.html May 28 14:39:30 not reading pgo any more May 28 14:40:05 (after someone had the brilliant idea of using a CSS that breaks with anything but current gecko) May 28 14:40:14 Hmm. What's the quickest way to build a native dev environment for my target? Or at least one I can run 'configure' in? May 28 14:42:04 reenoo_: tigrux said novell is getting arm boxes to test mono on May 28 14:43:20 I wonder if they improve their buildsystem May 28 14:43:33 it's not just the build system May 28 14:43:34 2 years back it did work May 28 14:44:12 for mono to become an option they'd have to come up with a fairly efficient JIT May 28 14:44:32 indeed May 28 14:44:41 that's where the armboxes come in May 28 14:50:34 * keturn has a pile of messages like 'kern.debug kernel: Skipping dirent: "swapon", ino #288, type 10, because curofs 50 < offset 56'. Are those worrisome? May 28 14:53:51 koen: well, maybe. I don't expect it to become a success anytime soon though. (with the performance and resource hungryness of mono on x86 and the fact that ARM JITs appear to be far from trivial in mind) May 28 14:54:25 yeah May 28 14:54:31 but novell has deeeeeep pockets May 28 14:55:00 they should hire zecke to cut down on memory usage :) May 28 14:55:15 heh May 28 14:55:23 For money I would even force my mother to program in C ;) May 28 14:55:25 just kidding May 28 14:55:53 maemo must be a slap into Trolltechs face May 28 14:56:00 heh. they should hire zecke to write an arm JIT. May 28 14:56:31 pb_: Some Linux Vendor needs to get a Mobile Linux strategy first May 28 14:56:34 * koen considers writing comments in bf May 28 14:56:49 pb_: I always wonder what happens if Novell switches all Desktops to Linux May 28 14:57:04 pb_: will they use Magneto on their handhelds? May 28 14:57:39 reenoo_: koen : I would prefer to get money to work on sandboxing/secure execution of applications May 28 14:57:41 zecke: what's Magneto? May 28 14:57:44 * pb_ ignorant May 28 14:57:46 we should call gpe 2.8 'Xavier' May 28 14:57:53 pb_: M$ Mobile 5 May 28 14:58:05 aha May 28 14:58:12 *happy to have set up suexec* May 28 14:58:29 zecke: cheers May 28 15:02:22 Sorry if this is not OE related (directly that is) I see in gdb that a function is in need of a certain variable, and only then the program executes the assignment of said variable. Is this the late-binding I have been reading about? May 28 15:06:02 Hertog: could be May 28 15:07:47 zecke: it sure doesn't look like anything like the procedural programming I learned, at some point in my life :) It's jumping all over the place :) maybe I should disable optimization grin May 28 15:08:39 AFAIK Hertog: actually the stuff in the dwarf section should hide such things May 28 15:11:00 zecke: Could also be my (not to be underestimated) ignorance May 28 15:37:16 hey ar_ May 28 15:37:32 hi zecke May 28 15:38:07 zecke: the DWARF data can't maintain a complete debug illusion if you are single-stepping through optimized code. May 28 15:38:43 it does patch up some holes, like variables flipping from one register to another, but it can't do anything about values that just haven't been computed yet at the point where you try to examine them. May 28 15:40:09 thanks May 28 15:41:37 pb_: hey, you packaged howl 1.0. Thanks. May 28 15:41:57 keturn: yah, I did. I haven't actually managed to get it to work on arm, though. May 28 15:42:05 I'd be interested to know if you have any success with it. May 28 15:42:08 "work", or "compile"? May 28 15:42:15 work May 28 15:42:26 I'm failing to build 0.9.7 on mispel at the moment. May 28 15:42:31 it compiles fine, but the resulting binaries don't seem to have any useful effect. May 28 15:43:08 as in, if I run mDNSResponder and mDNSPublish on my arm box, my i386 machine running mDNSBrowse doesn't seem to be able to find those services. May 28 15:45:06 Have you seen a good summary of the license issues with it? I'm still a little unclear under which circumstances it will get me in trouble. May 28 15:49:57 No, I haven't really looked into that stuff. The project that I was planning to use Howl in got canned, so I didn't do much with it lately. May 28 16:02:24 'night all May 28 16:09:28 * CoreDump|home chuckles May 28 16:09:30 http://www.bluephod.net/coolpic/C64SXEnglish.jpg May 28 16:10:28 hehe May 28 16:11:10 'night all May 28 16:11:32 reenoo_: good night May 28 16:12:57 gee, launching opie and gpe at the same time isn't a good idea May 28 16:16:07 obviously not, since launching opie is never a good idea May 28 16:16:36 some people say the same about GPE :) May 28 16:16:42 * Hertog starts looking for his asbestos underwear May 28 16:18:22 I'm looking at netbase, trying to figure out how netbase/netbase/$MACHINE/interfaces gets installed. Is $P/$MACHINE isn't usually in FILESPATH, is it? May 28 16:19:14 what's pgo? May 28 16:19:33 keturn: yes, it is May 28 16:19:48 $P/{anything in OVERRIDES} is in FILESPATH. May 28 16:21:46 (see base_set_filepath in base.bbclass) May 28 16:27:53 I see. May 28 16:28:22 I am starting to understand OVERRIDES, after, um, about five months on-and-off of this. May 28 16:29:31 It would work better if I actually trusted bitkeeper instead of using a seperate BBFILE_COLLECTION. It's harder to override things outside the tree. May 28 16:30:28 ah. yeah, that's a shortcoming in our current scheme. if you want to have your own interfaces file, you must create a local copy of the whole netbase package in your higher-priority collection. May 28 16:30:54 * luke-jr_ just appends -ljr to the package name or version May 28 16:31:05 (pkg name if meta-pkg, version otherwise) May 28 16:35:24 yay. howl 1.0 compiles. May 28 18:28:28 ~husky May 28 18:28:29 [husky] Sharp Zaurus SL-C760 or a dog May 28 19:18:25 Um. Heh. May 28 19:19:20 $TMPDIR/staging/i686-linux/bin/bison is a script. May 28 19:19:48 it is one line: exec $TMPDIR/staging/i686-linux/bin/bison -y "$@" May 28 19:20:05 This is not very productive. May 28 19:20:35 kergoth: rm the "yacc" symlink from the same folder May 28 19:20:42 and rebuild bison-native May 28 19:20:59 s/kergoth/keturn May 28 19:25:01 * keturn tries that. May 28 19:33:40 that looks better. Is that leftover from old versions of things, and not likely to happen on new installations? May 28 19:37:42 I believe so, yes May 28 20:27:29 crap (finally trying to mix & match kernel & initrd) May 28 20:27:36 udev doesn't support 2.4 XD May 28 20:33:18 seems to work well enough, tho =D May 28 21:36:53 Huh. Samba, um, kinda works. Sorta. If you don't need to know what your files are named. https://bugzilla.samba.org/show_bug.cgi?id=2287 May 28 22:00:57 * france is away: Away **** ENDING LOGGING AT Sat May 28 23:59:56 2005