**** BEGIN LOGGING AT Sat Jul 08 02:59:59 2006 Jul 08 04:29:51 CoreDump|afk: .....I figured it out.... Jul 08 04:30:07 CoreDump|afk: rxvt is broken....rxvt-unicode, however, is not.... Jul 08 05:27:46 03justinp 07org.oe.oz354x * r03c4b710... 10/packages/meta/task-e-x11.bb: task-e-x11: rxvt doesn\'t work, but rxvt-unicode does Jul 08 06:25:29 how do I set the foreground text color for my virtual terminal, in opie? Jul 08 07:19:40 03coredump 07org.oe.oz354x * r924987a9... 10/packages/wmctrl/wmctrl_1.07.bb: wmctrl: Add wmctrl, a cmd line tool to interact with an EWMH/NetWM compatible X Window Manager Jul 08 07:19:45 03coredump 07org.oe.dev * ra1b31cd3... 10/packages/wmctrl/wmctrl_1.07.bb: wmctrl: Add wmctrl, a cmd line tool to interact with an EWMH/NetWM compatible X Window Manager Jul 08 07:31:21 hrw|gone: are your current build scripts in OE? The one in the oz354x branch seems....old to me Jul 08 07:35:43 mickey|thesis: ping Jul 08 07:42:29 njs: thanks for 'pluck', I hope to try it soon Jul 08 07:51:01 * JustinP :-\ Jul 08 07:51:20 hehe Jul 08 07:53:11 hmm I can't pull from venge.net Jul 08 07:58:52 ~lart me Jul 08 07:58:52 * ibot declares zecke a moron Jul 08 08:05:58 morning all Jul 08 08:06:27 morning RP Jul 08 08:06:36 hey RP Jul 08 08:07:44 zecke: As you know, I started to look at rewriting the dependency code. I needed to start by trying to refactor the code in bin/bitbake. Would you agree a lot of that should become modules? Jul 08 08:08:06 * RP should really make some patches to discuss Jul 08 08:08:25 RP: well, mickeyl killed the modules and put it into bin/bitbake Jul 08 08:08:41 RP: I'm fine with creating modules, and I don't expect mickeyl to resist Jul 08 08:08:58 zecke: Which modules did mickeyl kill? Jul 08 08:09:13 I think it was only 'make.py' but I'm not sure Jul 08 08:10:07 I doubt mickeyl would (or could) make it worse :) Jul 08 08:10:27 oh this wasn't what I tried to say :) Jul 08 08:10:40 RP: no I'm fine with more modules Jul 08 08:10:46 mickeyl will be as well Jul 08 08:11:05 zecke: He said as much when I mentioned it earlier this week Jul 08 08:11:26 this damn time occupation... Jul 08 08:11:54 I'll generate some patches when I find some of that :) Jul 08 08:12:07 *time* :} Jul 08 08:16:11 03eFfeM 07org.oe.dev * rba9294b5... 10/packages/meta/slugos-packages.bb: slugos-packages.bb: added hdparm and lirc Jul 08 08:16:33 zecke: Is this too much of a hack? http://www.rpsys.net/openzaurus/temp/bb_fetcher_git_improve-r0.patch Jul 08 08:18:36 it is portable Jul 08 08:18:49 but I would better make sure we are in the right directory :} Jul 08 08:19:18 zecke: There is a chdir at just off the top of the diff Jul 08 08:20:00 this is to save storage size? Jul 08 08:20:05 zecke: git is a bit untidy with its files and it does save a lot of space/time to remove those ones... Jul 08 08:20:12 okay Jul 08 08:20:23 well then this is a justified hack Jul 08 08:21:06 We could probably use git more effectively but I can't seem to figure out a better checkout command (which will checkout files but not touch git's index file) Jul 08 08:21:10 hey zecke, RP Jul 08 08:21:15 hi kergoth Jul 08 08:25:27 * kergoth just got back from seeing the pirates of the carribean 2 Jul 08 08:25:55 kergoth: poor guy Jul 08 08:26:19 I saw the first one and it wasn't as bad as I was expecting :) Jul 08 08:26:20 hi koen Jul 08 08:26:41 RP: feel free to commit this speed improvement Jul 08 08:26:43 koen: hey Jul 08 08:26:52 koen: how is battling glibc going? Jul 08 08:26:55 i quite liked it. neither it nor the first one are movies you go to expecting something of incredible quality, you see them to be entertained, and both succeed admirably Jul 08 08:29:33 okay Jul 08 08:29:49 but what is this "entertaining" you talk about? Isn't it all about suffering? Jul 08 08:34:58 kergoth: The first one wasn't something I'd have watched given a choice but I will probably watch the second one voluntarily. I guess I did like it at some level :) Jul 08 08:40:00 hehe Jul 08 08:40:10 i didnt particularly want to see the first either, didnt seem like itd be worthwhile Jul 08 08:42:16 good morning all Jul 08 08:42:22 hey koen Jul 08 08:42:23 zecke: I bested it yesterday Jul 08 08:43:45 morning koen Jul 08 08:45:40 hey kergoth & RP Jul 08 08:48:40 koen: and did you score? Jul 08 08:48:49 koen: or is that ulrich too strong? Jul 08 08:49:50 zecke: Why can't the dot generation code use filterProviders? Jul 08 08:50:09 there was a reason Jul 08 08:50:23 I think I wanted some information that was thrown away in filterProviders Jul 08 08:50:57 You wanted all providers including those which couldn't be built? Jul 08 08:51:22 No matter, we'll perhaps have a better way to do things soon :) Jul 08 08:51:43 thank you soo much Jul 08 08:51:52 can it get me thai food yet? Jul 08 08:52:12 as I'm forbidden to eat pizza, I'm more interested in ordering thai/chinese food Jul 08 08:52:31 zecke: I was just about to ponder the status of the pizza fetcher :-/ Jul 08 08:52:53 I've just realised my bitbake svn is way out of date and have merge conflicts everywhere :-/ Jul 08 08:53:02 zecke: I did not score yet Jul 08 08:56:08 hey master Jul 08 08:56:24 pb_: how to get a ATgnu.org address? Jul 08 08:56:29 zecke: I was about to disect the tarball and the ipk when I tripped the 1/0 button on the extension cord Jul 08 08:56:56 hail zecke Jul 08 08:57:05 I guess you have to write to the gnu folks and ask Jul 08 08:58:02 morning pb_ Jul 08 08:58:10 hi rp Jul 08 09:04:55 kergoth: how's the new job? Jul 08 09:06:17 dont start work for ohand until monday Jul 08 09:09:59 you're staying in AZ or moving back to minnesota? Jul 08 09:13:07 staying in AZ for another 6 months to a year, then will see. not back to minnesota, somewhere else warm, so that limits to the south half of the US, basically Jul 08 09:13:37 morning Jul 08 09:19:10 g'day kergoth Jul 08 09:22:19 hmmm Jul 08 09:22:19 monotone: 0 | 0 | 192*** glibc detected *** monotone: double free or corruption (fasttop): 0x008e80c8 *** Jul 08 09:22:23 hmmm Jul 08 09:22:42 what did I do to deserve this :-) Jul 08 09:22:43 koen: my fridge is full of cassis :) Jul 08 09:22:52 ~lart monotone Jul 08 09:22:52 * ibot crushes monotone with a full height scsi disk Jul 08 09:25:42 koen: even buildroot segfaults .. which is good in a way. at least OE and buildroot have a common problem Jul 08 09:25:48 zecke: ah, good Jul 08 09:26:07 ade|desk: that's 'good' news Jul 08 09:26:19 eFfeM: rebuild your system Jul 08 09:28:05 zecke, I wanted to work on a package that only has a native bb file (and I figured out that creating a cross one was beyond my capabilities) Jul 08 09:28:21 so I just did ipkg install monotone-5 boost-dev etc Jul 08 09:28:25 on my target system Jul 08 09:29:19 was boost build with the same gcc as the rest of the system? Jul 08 09:29:55 koen, I hope so, because my system is on the head and the packages are also from the head Jul 08 09:30:46 the only thing I've found is: Jul 08 09:30:49 monotone: /usr/lib/libstdc++.so.6: no version information available (required by /usr/lib/libboost_regex.so.1.33.0) Jul 08 09:31:00 now trying to reinstall libstdc++ Jul 08 09:31:42 i've seen this version thing before a while ago with another lib and seem to recall that explicitly removing and reinstalling helped Jul 08 09:32:39 koen: my plan of action is to try using buildroot with different binutils and once i have a working busybox under buildroot, align OE with that and then try OE again Jul 08 09:32:53 this could take a while Jul 08 09:32:59 sounds like a good plan Jul 08 09:42:56 hmm 2.17 doesn't exist Jul 08 09:43:32 its not on kernel.org/pub/linux/devel/binutils/ Jul 08 09:43:45 have to grab from debian Jul 08 09:47:59 right thats two builds on their way Jul 08 09:48:39 * ade|desk hugs 4 way boxes Jul 08 10:01:07 www.idpz.net/tomytom/Linux.htm Linux news now ! :) Jul 08 10:02:24 www.idpz.net/tomytom/Linux.htm Linux news now ! :) Jul 08 10:07:09 ade|desk: http://www.gnu.org/software/binutils/ Jul 08 10:08:33 ehm Jul 08 10:08:35 ade|desk: http://ftp.kernel.org/pub/linux/devel/binutils/binutils-2.17.50.0.2.tar.bz2 Jul 08 10:09:01 thats 2.17.50 not 2.17 Jul 08 10:13:22 aren't the 2.17.50.x post 2.17 snapshots? Jul 08 10:13:29 i've grabbed the debian one and converted .gz to .bz2 to make buildroot happy Jul 08 10:13:59 koen: nope, they are hj lu's linux only versions of binutils Jul 08 10:14:14 aha Jul 08 10:14:23 his is the beta release of binutils 2.17.50.0.2 for Linux, which is based on binutils 2006 0526 in CVS on sources.redhat.com plus various changes. It is purely for Linux. Jul 08 10:14:25 * koen <- binutils n00b Jul 08 10:14:54 koen: and you think i know what i am talking about ?? haha Jul 08 10:16:13 help... my Z (860) locks when trying to load opie... Jul 08 10:16:38 I'm running oz3.5.4.1 I flashed yesterday Jul 08 10:16:50 YoG: try #opie or #openzaurus Jul 08 10:16:55 ok Jul 08 10:18:48 hmm something makes me suspect its not just binutils thats the problem, i think uclibc has ldso problems at the moment in svn Jul 08 10:19:02 grr Jul 08 10:45:52 koen: latest uClibc snapshot works :) Jul 08 10:47:03 still waiting for the 2.17 binutils build to finish Jul 08 10:50:59 hehe 2.17 and 2.16.91.0.7 work with the snaphost from last night. Jul 08 10:51:28 koen: which binutils do you use for angstrom Jul 08 10:51:51 2.17 has some fixes for arm so the changelog says Jul 08 10:57:22 right lets try OE with that snapshot Jul 08 10:57:41 * ade|desk kicks of an other fresh build Jul 08 10:58:36 zecke: How do you envisage debug output being controlled in bitbake in the future? Jul 08 11:00:07 oh Jul 08 11:00:38 Levels and Regions and then continue to use bb.debug(Level_Three,BuildCore,"Msg") Jul 08 11:00:47 zecke: I was wondering about some kind of debug domains Jul 08 11:01:02 right, regions = domains :) Jul 08 11:01:51 zecke: How do we convert to three options without breaking everything? :) Jul 08 11:02:03 Perhaps some kind of msgs module? Jul 08 11:02:55 RP: there is a msg module in python, we might see if that is okay Jul 08 11:03:03 RP: and we can either use default parameters Jul 08 11:03:20 RP: or create bb.debug2/debug_new/debug_moo Jul 08 11:03:46 zecke: We'd still need a bb module to wrap around it so we can change the implementation at some later date without having to change every reference in the source Jul 08 11:04:09 zecke: We're going to need better message control when we do multithreading anyway Jul 08 11:04:24 yes Jul 08 11:04:32 zecke: Another scary thought I had was how was tinderclient going to handle multithreading? Jul 08 11:05:01 ouch :) Jul 08 11:05:08 thanks for the pain Jul 08 11:05:22 zecke: sorry :} Jul 08 11:05:39 koen: have you read the interview with Havard? Jul 08 11:05:54 koen: it is funny how even Qt2 gets relabeled QtopiaCore Jul 08 11:10:00 woglinde: ^hey Jul 08 11:10:12 he zecke Jul 08 11:43:44 03eFfeM 07org.oe.dev * r80526ab0... 10/packages/meta/slugos-native.bb: slugos-native: previous fix did not work, now added DEPENDS "libc6-dev" Jul 08 11:51:07 There are so many different things I want to change bitbake, I can't see a sensible way to start patching as they're all interdependent :-/ Jul 08 11:51:47 hehe Jul 08 11:53:57 zecke: Take http://www.rpsys.net/openzaurus/temp/bb/split_providers.patch - the only way to fix the FIXMEs for now is to add two variables to the call which ultimately we don't want/need :-/ Jul 08 11:54:23 zecke: Would you agree we can remove these: http://www.rpsys.net/openzaurus/temp/bb/task_cleanup.patch ? Jul 08 11:55:05 RP: I have never used these (also the complete graph code is a stranger to me) Jul 08 11:55:15 so I think we can remove them Jul 08 11:55:25 zecke: They appear to do nothing and it still works when I remove them :) Jul 08 11:56:06 I just spent a while trying to work out what they did as couldn't quite believe they did nothing... Jul 08 11:56:45 hehe Jul 08 12:00:11 RP: are you working on better reporting of wrong PREFERRED_* as well? Jul 08 12:04:22 anyone an idea what is wrong if a lib does not contain version definitions? Jul 08 12:04:55 i don't get these with libstdc++ when I compile gcc 4.1.1. cross r5 Jul 08 12:05:03 the older r4 work dir has a proper lib Jul 08 12:05:06 anyone an idea? Jul 08 12:06:22 zecke: Better reporting would come later. My plan is to rework the structure so its sane, then look at our message structure Jul 08 12:12:11 In some svk repo I have some code for that already (error reporting) Jul 08 12:12:34 RP: I will leave bitbake up to you, and try to get some perl hacking on the tinderbox done Jul 08 12:13:37 zecke: ok. I'm not quite sure where these changes will lead atm. I want to sort the multithreading but there's a lot to fix before it'll be possible :-/ Jul 08 12:13:47 zecke: Did you consider buildbot btw? Jul 08 12:14:39 RP: yes, it could be a lack of understanding but it doesn't cut it for what I have in mind :} Jul 08 12:14:46 Hi, not sure if off topic, but can anyone point me to documentation (if there is any) about familiar-gpe virtual keyboards? Jul 08 12:14:54 RP: e.g. I do not want to have the server push commands to the clients Jul 08 12:15:06 RP: we have too many different configurations Jul 08 12:15:11 zecke: right, I can understand that Jul 08 12:15:47 RP: buildbot works great if you have one target and build it on many different systems Jul 08 12:16:14 damn kspread, gnumeric and ooffice broken at the same time :( Jul 08 12:16:58 zecke: They all seem to have good and bad sides as far as OE is concerned. I agree we need to push from client to server rather than the other way around Jul 08 12:51:48 * ade|desk does the happy dance: a real live working uclibc X window system image in 3.05 mb compressed Jul 08 12:52:57 ade|desk: neat :) Jul 08 12:53:52 i lied, its 3125 K Jul 08 12:54:23 i dont need locale but its in there anyway Jul 08 12:54:36 as well as rdesktop Jul 08 13:07:00 hehe 2.3G to make 3.1M Jul 08 13:35:50 ade|desk: neat Jul 08 13:36:28 koen: have you tried 2.17 for angstrom ? Jul 08 13:38:00 RP: I'm touching build.py now, is that okay? Jul 08 13:41:46 zecke: Yes, I'll handle any merge Jul 08 13:41:48 hi .. i started a new build today with today`s oe.db .. now qemu-native fails to patch mouse_fix-r0 Jul 08 13:42:37 rob_w: use an older SRCDATE for qemu-native or update the patch? Jul 08 13:42:47 RP, ok Jul 08 13:43:01 ade|desk: not yet Jul 08 13:44:16 RP: okay, I'm on 1017 (which is a bitbake bug) Jul 08 13:44:26 ade|desk: probably because 2.17 isn't in OE yet :) Jul 08 13:44:54 koen: now that i've started drinking .. you might like to take the patches from buildroot for 2.17 and add them to a 2.17 bb Jul 08 13:45:15 else i'll do it monday Jul 08 13:45:47 I'm adding 2.17 right now Jul 08 13:46:01 do you have a direct url for the br patches? Jul 08 13:46:34 not direct only svn checkout Jul 08 13:47:51 doh i've had too much to drink to drive now Jul 08 13:47:54 bugger Jul 08 13:48:52 ade|desk: psst: http://svn.uclibc.org/cgi-bin/viewcvs.cgi/trunk/buildroot/toolchain/binutils/2.17/ Jul 08 13:49:32 that would do it ;) Jul 08 13:50:45 we should have a closer working relationship with the uclibc team , we nick so much stuff from buildroot Jul 08 13:52:46 ade|desk: convince vapier to use OE :) Jul 08 13:53:21 unlikely, he's a embedded gentoo boy i think Jul 08 13:53:26 s/a/an Jul 08 13:53:29 he is Jul 08 13:53:53 'embedded gentoo - for the extremely patient' Jul 08 13:54:49 koen: oh well, patience is not the issue. swap over nbd - deadlocking is the issue :) Jul 08 13:55:04 there was an embedded gentoo user in #gpe a while back Jul 08 13:55:24 he said OZ was too big (~20MB) so he installed a uclibc gentoo image Jul 08 13:55:29 which was ~200MB Jul 08 13:55:57 that was when I lost respect for gentoo/e Jul 08 13:56:55 each project has slightly differing aims, i think it is good to share knowledge and patches.. after all we all want the same in the end Jul 08 13:57:14 group hug anybody ? Jul 08 13:58:18 sharing patches and knowledge it good Jul 08 14:00:41 i think it is the disease of the open source community, distro wars and slaping down other projects. seems like wasted effort to me Jul 08 14:03:04 lol bbc2 has the first ever star trek episode on Jul 08 14:03:14 captain pike :)# Jul 08 14:07:23 ade|desk: binutils 2.17 + br patches is on its way Jul 08 14:08:04 ade|desk: congrats Jul 08 14:08:07 morning folks Jul 08 14:08:36 hey mickey|thesis Jul 08 14:08:56 howdy mickey|thesis Jul 08 14:10:10 mickey|thesis: that h4000 patches you mentioned on the mailing list is funny Jul 08 14:10:21 koen: why? Jul 08 14:10:28 mickey|thesis: about 60% of it only works around bugs in the familiar tree Jul 08 14:10:34 urghs Jul 08 14:11:07 it munges the kernel PV inside machine.conf Jul 08 14:11:18 because they have a fixed SRCDATE for the release Jul 08 14:11:27 it is just wrong in too many ways Jul 08 14:12:44 hmm.... is at least the h4000 overrides in a couple of packages useful? Jul 08 14:12:52 mickey|thesis: did you push some csets just now? Jul 08 14:13:00 yes, 3 Jul 08 14:13:02 mickey|thesis: only on tslib.bb Jul 08 14:13:13 mickey|thesis: I thought so -> /me merges Jul 08 14:13:39 03mickeyl 07org.oe.dev * r1a49e7e2... 10/packages/python/ (python-2.4.3-manifest.inc python_2.4.3.bb): python 2.4.3 add PROVIDE line for subpackages to satisfy the BitBake RDEPENDS changes Jul 08 14:13:44 03koen 07org.oe.dev * r43a56f99... 10/packages/binutils/ (9 files in 2 dirs): binutils 2.17: add normal and cross version Jul 08 14:13:48 03mickeyl 07org.oe.dev * r8ef22559... 10/packages/ (4 files in 3 dirs): add python-webpy, a lightweight framework for web applications Jul 08 14:13:55 03koen 07org.oe.dev * r2e3b25af... 10/conf/distro/angstrom-2006.9.conf: angstrom: use a released binutils instead of a snapshot Jul 08 14:14:42 ade|desk: there you go Jul 08 14:14:52 nice one Jul 08 14:15:57 zecke, mickey|thesis: How does http://www.rpsys.net/openzaurus/temp/bb/msg_module.patch seem to try and improve message handling? Jul 08 14:16:16 (obviously I've not done a complete conversion but we don't have to all at once) Jul 08 14:17:21 RP: having a bitfield for the domains would rock more Jul 08 14:17:38 RP: then one could enable/disable the domains on the fly Jul 08 14:18:42 zecke: I had a dict of strings with enable/disable set against them. I'd like to enforce a clear set of domains on that field but my python isn't up to it. help :) Jul 08 14:18:53 s/them/them in mind/ Jul 08 14:19:02 ah that will work :) Jul 08 14:19:27 pythons doesn't do enums? :) Jul 08 14:19:50 it does - the weird way Jul 08 14:20:32 using classes or other static stuff Jul 08 14:23:07 RP: trunk is free again Jul 08 14:24:52 zecke: ok, thanks Jul 08 14:27:59 mickey|thesis: http://venge.net/monotone/wiki/DaggyFixes might interest you Jul 08 14:29:20 03koen 07org.oe.dev * r8c08038b... 10/packages/tslib/tslib_20060703.bb: tslib: add h4000 support Jul 08 14:29:25 03koen 07org.oe.dev * r71c0819b... 10/conf/machine/h4000.conf: h4000.conf: add cleaned up version of h4000.conf Jul 08 14:29:29 03koen 07org.oe.dev * r01a9e0cf... 10/classes/rootfs_ipk.bbclass: rootfs_ipk.bbclass: add make_zimage_symlink_relative Jul 08 14:29:33 03koen 07org.oe.dev * rc1685b2c... 10/conf/machine/h2200.conf: h2200.conf: make symlink relative to fix booting with LAB Jul 08 14:31:48 mickey|thesis: h4000 support added Jul 08 14:31:56 thanks Jul 08 14:32:10 mickey|thesis: only a small portion of the patch survived Jul 08 14:32:21 a good read Jul 08 14:32:24 (daggyfixes) Jul 08 14:32:58 I don't think we can educate most of our developers and users to make use of daggyfixes Jul 08 14:33:09 it's a long way from the cvs paradigma Jul 08 14:33:51 also note that the bitbake/OE interaction makes true daggyfixes hard to use Jul 08 14:33:55 mickey|thesis: hey Jul 08 14:34:29 hail zecke Jul 08 14:34:37 hm Jul 08 14:34:38 zecke: a very good mail, btw. Jul 08 14:34:40 yo woglinde_ Jul 08 14:34:41 ~hail zecke Jul 08 14:34:45 * ibot bows down to zecke and chants, "I'M NOT WORTHY!!" Jul 08 14:34:57 stop that, my colleagues make fun Jul 08 14:35:12 mickey|thesis: shortly afterwards I have read the interview with harvard Jul 08 14:35:26 dont chat as root Jul 08 14:35:27 Good morning everybody Jul 08 14:35:28 which one? Jul 08 14:35:28 its evil Jul 08 14:35:29 URL? Jul 08 14:35:43 mickey|thesis: on linuxdevices.com Jul 08 14:35:56 good mail from zecke, where? Jul 08 14:36:14 Hey, is this Mickey from the openezx list? Jul 08 14:36:27 danielsantos: yes, he is Jul 08 14:36:30 mickey|thesis: http://linuxdevices.com/articles/AT3980891782.html Jul 08 14:36:33 oh, hey there :) Jul 08 14:36:45 and http://linuxdevices.com/news/NS6437147703.html Jul 08 14:36:57 ah yeah the oz articel Jul 08 14:37:01 woglinde: may I use your SUSE buy out idea, to start the buy out of Trolltech? Jul 08 14:37:20 zecke I claimed a patent on that Jul 08 14:37:30 koen: it is sooo funny, but I knew it would come to this Jul 08 14:37:39 Hey, anybody got a second to help me with my OpenEmbedded configuration? I'm having trouble with where my directories are. Jul 08 14:37:51 koen: Moto didn't license Qtopia, only QtE Jul 08 14:38:10 koen: so they relabeled Qt to Qtopia Core and suddenly Motorola is a Qtopia user as well Jul 08 14:38:26 danielsantos: did you read gettingstarted on the wiki? Jul 08 14:38:28 #tell woglinde hehe, and then I promise I'll go setup a user account :) Just got this machine going Jul 08 14:38:31 doh! Jul 08 14:38:37 Oh yea Jul 08 14:38:55 daniel try /msg woglinde Jul 08 14:39:07 I wanted my tmp files to happen in tmpfs, but then I found that my cross-compiler was going away when I re-booted Jul 08 14:39:08 ok, create one bild dir, and checkout OE to some dir Jul 08 14:39:16 woglinde: try /msg NickServ setunfiltered on Jul 08 14:39:26 heh Jul 08 14:39:30 okies, thanks... been a while since was in IRC Jul 08 14:39:37 zecke hm Jul 08 14:39:44 danielsantos: hehe, you can set a different dir for the cross compiler and staging Jul 08 14:39:48 zecke right bloody nickserv Jul 08 14:40:47 Ok, so which directories should I keep in a persistient place then? I got my cross-compiler where it should be now (/usr/arm-a780-linux-gnu) Jul 08 14:41:09 danielsantos: OE has its own crosscompiler Jul 08 14:41:35 Oh yea, well that is where I had OE put the cross-compiler Jul 08 14:42:00 Actually, I think I moved it there from my /var/tmp/oetemp/cross directory... Jul 08 14:42:21 but what other directories should I keep around? Just staging and deploy? Jul 08 14:42:23 danielsantos: see bitbake.conf for possible directories Jul 08 14:42:36 danielsantos: staging/ cross/ and deploy/ Jul 08 14:42:46 danielsantos: and stamps :) Jul 08 14:42:53 I promise I couldn't find docs on this :) Jul 08 14:42:58 sweet! thanks zecke!! :) Jul 08 14:43:19 danielsantos: that is too easy with OE (lack of docs, but we are working on it) Jul 08 14:43:56 danielsantos: and if you look at bitbake.conf you could just move 'work' to a tmpfs (without symlinks) Jul 08 14:43:59 Aye, maybe I can update that part of the Wiki. Hehe, I followed the Getting Started to a T except that I wanted my builds to happen in tmpfs. I didn't know I was putting everytyhing there! :) Jul 08 14:44:43 RP: you might want to update poky's bitbake Jul 08 14:44:55 RP: or give it a try at least :) Jul 08 14:45:04 zecke: For which bugs? :) Jul 08 14:45:11 what's poky's bitbake? Jul 08 14:45:12 1017 Jul 08 14:45:24 zecke: Is this better http://www.rpsys.net/openzaurus/temp/bb/msg_module.patch ? Jul 08 14:45:43 danielsantos: poky is a subset of OpenEmbedded, created by o-hand.com to testdrive some of their technology Jul 08 14:45:52 ~poky Jul 08 14:45:55 well, poky is http://projects.o-hand.com/poky Jul 08 14:46:06 ahhh Jul 08 14:46:31 interesting, thanks Jul 08 14:47:08 RP: wtf, where did you get your python skills from? Jul 08 14:47:27 assert names, "E..." <- I didn't know this even exists Jul 08 14:47:45 lmao! ;) Jul 08 14:47:48 zecke: I found a good reference about enums ;-) Jul 08 14:47:56 hehe Jul 08 14:48:11 : don't know jack about python yet :) Jul 08 14:48:27 RP: http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/413486 ahhhh Jul 08 14:48:39 zecke: ;-) Jul 08 14:48:51 ACTION Jul 08 14:49:01 * danielsantos don't know jack about python yet :) Jul 08 14:49:12 Ok, now I remember the command, hehe Jul 08 14:49:25 RP: wow, looks nice :) Jul 08 14:49:53 zecke: Is utils the right place for the Enum class? Jul 08 14:50:14 sounds good Jul 08 14:50:28 ibot: poky also is a subset of OpenEmbedded, created by o-hand.com to testdrive some of their technology Jul 08 14:50:29 koen: what are you talking about? Jul 08 14:50:31 zecke: We should move digraph to there as well Jul 08 14:50:41 lol Jul 08 14:50:45 do we need it at other places? Jul 08 14:50:47 ~poky Jul 08 14:50:49 rumour has it, poky is http://projects.o-hand.com/poky Jul 08 14:50:52 ibot: poky is also a subset of OpenEmbedded, created by o-hand.com to testdrive some of their technology Jul 08 14:50:53 okay, koen Jul 08 14:50:56 zecke: Its in __init__ atm :-/ Jul 08 14:50:58 ~poky Jul 08 14:51:01 from memory, poky is http://projects.o-hand.com/poky, or a subset of OpenEmbedded, created by o-hand.com to testdrive some of their technology Jul 08 14:51:19 ibot: s/or/and/ :) Jul 08 14:51:28 yeah, stupid ibot Jul 08 14:51:41 RP: will you make clutter fast on the Sharp zaurus? Jul 08 14:51:49 okay I need to go to the gym Jul 08 14:51:50 later Jul 08 14:51:54 I've learned a long time ago not to click random links in an irc channel Jul 08 14:51:59 hey XorA Jul 08 14:52:02 zecke: Using openGL? :) Jul 08 14:52:04 hey koen Jul 08 14:52:18 bye zecke, greet al Jul 08 14:52:23 from me and tobi Jul 08 14:52:24 koen: havng real trouble with my image :-) Jul 08 14:52:32 zecke: cya Jul 08 14:53:02 woglinde: have fun with tobi :) Jul 08 14:53:07 XorA: if you spot lrg or mikearthur, I think I solved their binutils troubles with some help from ade|desk Jul 08 14:53:10 woglinde: treat him good, otherwise you get trouble Jul 08 14:53:22 koen: cool Jul 08 14:53:55 koen: not sure if its gtk+ 2.10 or dejavu fonts 2.7 but I get thousands of alignment errors now, my Z crawls Jul 08 14:56:12 koen: help ? Jul 08 15:02:25 ok, thanks for the help all Jul 08 15:02:49 msg: woglinde going to setup my user account now, LOL! :) Jul 08 15:06:59 ade|desk: pointing me to uclibc patches :) Jul 08 15:10:32 koen: was it not you who pointed me ? Jul 08 15:10:38 mickey|thesis: could you send a mail to the openezx list detailing ASoC stuff? Jul 08 15:11:04 ade|desk: you said that there were patches, and I found the http url Jul 08 15:11:16 * koen -> food Jul 08 15:11:48 * mickey|thesis notes to TODO Jul 08 15:12:39 * XorA wonders why dejavu fonts suddenly cause loads of alignment errors Jul 08 15:15:48 lol Jul 08 15:15:50 http://thedailywtf.com/forums/thread/80084.aspx Jul 08 15:24:11 mickey|thesis, one question. Im building for 2 platforms (using multimachine), how should the feed be put up? Can I have same feed dir for both machines or do I need to build one at a time and upload to different? It should check machine before downloading any .ipk methink, but want to make sure Jul 08 15:24:31 hi mickey|thesis Jul 08 15:24:35 hey pb_ Jul 08 15:25:43 Kristoffer: you can use a common directory for both feeds, since the architecture is encoded in the ipk filename and if you set the machine ipk priorities correctly, then the machine specific package gets preferred when building the feed. Jul 08 15:26:13 l8er Jul 08 15:26:29 mickey|sports, danke Jul 08 15:29:03 Kristoffer: if ipkg gets too slow on your target you can consider splitting the feeds Jul 08 15:29:13 Kristoffer: if not you can leave it as one big feed Jul 08 15:29:43 koen, Ive already experienced the feed getting abit too slow for my taste. How can I do that? Jul 08 15:29:57 Would need to make dirs inside the deploy/ipk folder? Jul 08 15:30:53 you can do it on your webserver Jul 08 15:31:23 But If I lets say want to split up all my 6xx packages into subgroups, how can it keep track of where to find a file? Jul 08 15:31:26 mkdir ../sh3 ../arm ; mv *arm.ipk ../arm ; mv *sh3.ipkg ../sh3 Jul 08 15:31:27 etc Jul 08 15:31:51 and add those directories to /etc/ipkg/.conf Jul 08 15:32:30 I understand that for arch, but is it also possible then to set up subfolders to each arch, for lets say sh3/opie.. Jul 08 15:32:31 Kristoffer: have a look at http://www.openembedded.org/repo/org.openembedded.oz354x/contrib/buildscripts/openzaurus/feed-split/ Jul 08 15:32:42 koen, thx Jul 08 16:00:23 koen, one quick question. If I move around ipk's (to example /a /b /c /d..subfolders) and then run package index, would it create the correct package index then? Jul 08 16:18:07 Task-opie/Opie-image seems to remove my CVS snapshots of libopiecore2 when building image Jul 08 16:18:36 no idea why, but It deletes them and tries to use 1.2.2 and then failing cause ive defined that i want the SRCDATE one Jul 08 16:19:13 Its like it thinks that 1.2.2 > 1.2.2+CVS-20060709 or 1.2.2 = 1.2.2+CVS-20060709 Jul 08 16:20:22 posted a bug about it Jul 08 16:22:10 Kristoffer: no, you'd need to make a Packages file by hand Jul 08 16:22:22 koen, roger Jul 08 16:22:31 and ipkg-make-index removes those files Jul 08 16:22:44 actually it doesn't remove them, but moves them to morgue/ Jul 08 16:23:51 sweet Jul 08 16:23:55 koen, btw working on getting a jornada7xx image, but it complains about not being able to set/get suffix Jul 08 16:23:56 40 days uptime on my h2200 Jul 08 16:24:01 koen, nice :D Jul 08 16:26:43 hi Jul 08 16:26:48 CoreDump|home, yo Jul 08 16:53:06 account off Jul 08 16:53:09 oops Jul 08 17:26:49 hey Jul 08 17:28:32 RP: I fear, this is what we do already! Jul 08 17:30:04 adding PN to overrides Jul 08 17:32:05 zecke: locally Jul 08 17:32:15 zecke: RP wants to add it globally Jul 08 17:32:26 (at least that was what I was told earlier) Jul 08 17:32:30 hmmm Jul 08 17:32:38 monotone + photoshop = slow Jul 08 17:36:24 I guess I can put that on my "excuses to buy a macbook pro" list Jul 08 17:36:36 anything with phoyoshop is slow Jul 08 17:36:51 koen: hehe, macbook pro *poser* Jul 08 17:36:57 koen: I hate the macbook keyboard Jul 08 17:36:59 it will happily eat all your ram unless you limit it's defults Jul 08 17:37:15 koen: I had an offer to exchange my ibook against a macbook, I refused to get an intel :} Jul 08 17:37:16 zecke: I'd be happy with a macbook if it wouldn't have that &*%(@#&$%*)@#& glossy finish on the screen Jul 08 17:37:38 but it makes it look pretty Jul 08 17:37:45 and thats what sells it Jul 08 17:37:58 hmm zaurus updater and rm_work do not like each other Jul 08 17:40:54 koen: ah that glossy look is not too bad Jul 08 17:41:21 zecke: my 12" powerbook is already to glossy for my taste Jul 08 17:41:39 I'm too disabled to see any difference :) Jul 08 17:43:32 zecke: We only do it locally atm, as koen says Jul 08 17:44:19 RP: hmm, what would be the benefit of doing it globally? Jul 08 17:45:15 hmm buildroot is GNU only as well :( Jul 08 17:51:45 hmmm, my printer only handles 8bit colour Jul 08 17:52:00 lol Jul 08 17:56:23 mine does two bit color Jul 08 17:56:30 black and orange Jul 08 17:56:43 i think i need to get the colors refilled Jul 08 17:57:27 NAbyss: progressing? Jul 08 17:57:36 I think I now why the printer was so cheap :( Jul 08 17:58:01 it doesn't try to compress values >8bit Jul 08 17:58:08 if you didnt live so far i'd send you one of mine for free Jul 08 18:00:30 koen: I said it was a blocker because It's stopping us progress with our image Jul 08 18:00:50 sorry if I was mistaken about the usage of blocker Jul 08 18:01:08 tkp_, you mean koen or me? Jul 08 18:01:09 if it stops your image, it's a blocker Jul 08 18:01:38 ok, well it stops out image, so I guess it is a blocker Jul 08 18:02:19 emte: if your the one they call nigel Kostiuck ;) Jul 08 18:03:17 this means that policy has changed to allow any minor package to become a blocker? Jul 08 18:03:33 as opposed to ones blocking base images or hardware damage? Jul 08 18:03:53 tkp_: what package? Jul 08 18:03:59 xerces-c Jul 08 18:04:01 zecke: xerces-c Jul 08 18:04:06 and xml parser Jul 08 18:04:09 tkp_: I will hack on it tonight Jul 08 18:04:22 zecke: that would be appreciated Jul 08 18:04:57 tkp_: why do you need it? I needed/need it for MCL Jul 08 18:05:18 it's for one of our own binaries Jul 08 18:06:38 zecke|food: MCL? Jul 08 18:09:03 * emte wonders what you would use xerces-x on an AS/400 for ... or even how Jul 08 18:16:58 Not downgrading package fluxbox-gpe on root from 0.99+1.0rc-r4 to 0.99+1.0rc2-r0 Jul 08 18:17:10 so much for debianizing the version grrr Jul 08 18:20:39 hi all Jul 08 18:23:54 emte: so, I have another bug to report... it's a blocker for me Jul 08 18:24:12 should I put it in as a blocker? Jul 08 18:46:54 03hrw 07org.oe.oz354x * r90f4d2b8... 10/packages/keymaps/ (files/tosa/keymap-2.6.map keymaps_1.0.bb): Jul 08 18:46:54 keymaps: updated keymap for tosa/2.6 Jul 08 18:46:54 - Record, Home, Rotate changed to the one expected by OPIE Jul 08 18:46:54 - LeftOK is now Control (was Escape) Jul 08 18:46:55 tkp_, probably, as i asked earlier i guess policy has changed Jul 08 18:46:59 03hrw 07org.oe.dev * racafceaa... 10/packages/keymaps/ (files/tosa/keymap-2.6.map keymaps_1.0.bb): Jul 08 18:46:59 keymaps: updated keymap for tosa/2.6 Jul 08 18:46:59 - Record, Home, Rotate changed to the one expected by OPIE Jul 08 18:47:00 - LeftOK is now Control (was Escape) Jul 08 18:47:03 synced with .oz354x Jul 08 18:47:22 emte: ah well... went in as normal Jul 08 18:47:49 tkp_, the policy used to be for things that broke the build system, possible hardware damage, and minimal image build impairment Jul 08 18:48:22 emte: tbh, that probably makes more sence Jul 08 18:48:30 it's only a blocker for me Jul 08 18:49:06 in the greater picture of OE, it's probably a smaller issue Jul 08 18:49:23 it does break your full image so i would see it as a major bug, but as far as a blocker... i dont know Jul 08 18:50:03 this is sort of where the debaucle with the OE vs hh.o lists is at Jul 08 18:50:44 since oe is the build system and it's meta where as hh.o covers familiar specific, gpe, etc Jul 08 18:51:06 emte: sorry, hh.o? Jul 08 18:51:08 but parts of familiar do overlap into oe Jul 08 18:51:13 handhelds.org Jul 08 18:51:20 ah, k Jul 08 18:51:59 i belive oz uses oe only tho Jul 08 18:52:24 and a few others have thier own Jul 08 18:53:09 emte: please keep still if you don't know what you are talking about Jul 08 18:53:25 so in all that mix where things belong is rather confusing for everyone excpet the core oe devs Jul 08 18:53:41 and koen loves me too Jul 08 18:53:58 mm. I'm certainly finding it a little confusing to put things in their place Jul 08 18:54:58 OE is the build system, but it seems very closly tied with specific distros Jul 08 18:55:43 yes, it seems to me to be going farther towards that direction Jul 08 18:56:16 03cbrake 07org.oe.dev * r3efc3491... 10/packages/python/ (python-pyro_3.4.bb python-pyro_3.5.bb): python-pyro: upgrade from 3.4 to 3.5 Jul 08 18:56:19 ofcourse everything i say is my opinion and impression Jul 08 18:56:28 just to qualify for koen Jul 08 18:57:39 now lets see if i can fry my ipaq Jul 08 19:04:00 I seem to get this quite often: Jul 08 19:04:04 NOTE: Couldn't find shared library provider for libgcc_s.so.1 Jul 08 19:04:32 not sure if i'm supposed to Jul 08 19:06:01 not sure myself, there are some things that are okay to ignore Jul 08 19:06:07 dunno if that is one Jul 08 19:06:23 tkp_: that is some kind of bug, I have never looked at :} Jul 08 19:06:51 zecke|food: could it be the cause of the other problems I'm having? Jul 08 19:06:57 * tkp_ is just speculating Jul 08 19:07:24 tkp_: the only error could be: libgcc_s.so.1 missing from your rootfs Jul 08 19:09:46 * tkp_ wonders why that might be Jul 08 19:11:22 tkp_: rootfs? Jul 08 19:12:22 zecke: rootfs... what about it? Jul 08 19:12:31 I was installing alsa-utils when I got the error Jul 08 19:12:43 * tkp_ has to go eat... bbiab Jul 08 19:13:54 tkp_: I have missed your error but it is unlikely you are suffering from libgcc... Jul 08 19:25:05 zecke: The global benefit is that you can influence packages from your local.conf file Jul 08 19:25:47 but SRCDATE_PN = "foo" is already working Jul 08 19:25:58 for each task PN is prepended to OVERRIDES Jul 08 19:26:09 sorry, I don't see a difference :} Jul 08 19:31:11 woglinde: hey Jul 08 19:31:16 he zecke Jul 08 19:31:18 0:0 Jul 08 19:31:32 woglinde: is this his first soccer match? Jul 08 19:31:50 oh no Jul 08 19:32:03 in the hospital he watched the halfinal Jul 08 19:32:04 zecke: It is not in overrides. SRCDATE only works due to magic in the fetcher modules Jul 08 19:32:38 ah well :) Jul 08 19:32:47 zecke: If it were, I'd agree :) Jul 08 19:32:50 fetchers should use update_data Jul 08 19:32:56 zecke but he is sleeping right now Jul 08 19:33:19 RP: well, update_data should be called before asking the fetchers to fetch Jul 08 19:33:36 zecke nachos Jul 08 19:33:36 RP: and most of the tasks put PN into the OVERRIDES Jul 08 19:34:07 and yes I agree we need some 'INSERT GOOD WORD HERE' to have a consistent usage of OVERRIDES Jul 08 19:35:54 zecke: The fetchers do use update_data IIRC Jul 08 19:36:01 some :) Jul 08 19:36:13 but most of the time to get the right FETCHCOMMAND Jul 08 19:36:33 anyway update_data is so incredible fast you can always call it :) Jul 08 19:36:44 ;-) Jul 08 19:37:17 Parsing time is still our bottleneck Jul 08 19:37:45 bitbake-1.5 should be a lot faster already Jul 08 19:37:53 it has mithros COW code Jul 08 19:38:14 I have to admit I've not noticed much difference :-/ Jul 08 19:38:34 it is Jul 08 19:39:02 admitingly it makes no difference to "Start Bitbake; Get Coffee and order a pizza; beat wife; get drunk; return to console" Jul 08 19:39:10 -Routine Jul 08 19:39:50 haha Jul 08 19:39:56 I'm hoping to make depends resolution faster by only touching the fast cache and not requiring further file parsing Jul 08 19:40:12 RP: I have relaunched the poky builds Jul 08 19:40:38 zecke: I hope they're working :} Jul 08 19:41:59 RP: I had to hack tinderclient.bbclass Jul 08 19:42:01 so what do you use for the bb LICENSE field for something like this: http://pastebin.ca/82643 Jul 08 19:42:16 RP: somehow the firewall fascist blocked the longer HTTP posts... Jul 08 19:42:24 zecke: Ah :-/ Jul 08 19:42:45 cbrake: looks like BSD? Jul 08 19:42:56 zecke: k, thx Jul 08 19:43:00 or even MIT :} Jul 08 19:43:03 cbrake: check osi :) Jul 08 19:43:11 zecke: will do Jul 08 19:45:21 we want binutils-2.17 to be the default binutils? Jul 08 19:45:33 zecke if works yes Jul 08 19:46:02 woglinde: I have no proof that it does Jul 08 19:46:29 mickey|fifa2006: we should try michael meeks -Bdirect option for binutils Jul 08 19:47:18 zecke: does not match BSD exactly Jul 08 19:47:44 MIT? X? Jul 08 19:49:33 03eFfeM 07org.oe.dev * rc0ec3b82... 10/packages/meta/slugos-native-packages.bb: slugos-native-packages: added php-native Jul 08 19:49:52 lo? Jul 08 19:52:12 zecke: google: "THE AUTHORS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE" site:opensource.org Jul 08 19:52:17 zecke: does not return anything :-( Jul 08 19:52:37 well that is the normal warranty disclaimer Jul 08 19:52:43 re koen Jul 08 19:53:05 and don't use google, google is evil Jul 08 19:53:49 jkp__: I'm not on AMD64, it fails to build here as well :) Jul 08 19:53:49 zecke: but if I google: "THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY" site:opensource.org Jul 08 19:53:58 zecke: it returns the MIT license. Jul 08 19:54:05 zecke: which? Jul 08 19:54:34 zecke: cheetah must not use a standard license Jul 08 19:55:01 hehe Jul 08 19:55:10 cbrake: it sounds like MIT then :) Jul 08 19:55:17 MIT-like :) Jul 08 19:56:14 zecke: dosfstools-native? Jul 08 19:56:21 zecke: the second google string is part of the MIT license -- just a google sanity check. Jul 08 19:56:38 zecke: I'd say MIT-like is close enough :-) Jul 08 19:56:55 tkp_: no xerces-c Jul 08 19:57:01 ah Jul 08 19:57:14 tkp_: for dosfstools-native check if a newer release is available Jul 08 19:57:25 tkp_: or if someone has some gcc4 patches Jul 08 19:58:25 zecke: wre on gcc 3.4.4 Jul 08 19:58:38 tkp_: on the 'host'? Jul 08 19:58:47 ah, no Jul 08 19:58:57 tkp_: we are taking about a -native package Jul 08 19:59:10 on the host were on 4.1.1 Jul 08 19:59:56 patches would be in the oe reposirory? Jul 08 20:00:03 guess I'll need to pull and update Jul 08 20:04:42 tkp_: no patches won't be in the oe repository Jul 08 20:04:51 tkp_: otherwise they would have been applied :) Jul 08 20:05:06 zecke: good point :p Jul 08 20:05:17 :) Jul 08 20:05:18 anyone able to tell me how to clean the kernel build so i can force a rebuild? Jul 08 20:05:44 -b kernel-bb-file -cclean Jul 08 20:05:51 Nothing provides dpendency kernel Jul 08 20:06:32 didnt like that zecke Jul 08 20:07:07 jkp__: well, I don't know which kernel you built, so you must think enough to fill the blanks Jul 08 20:07:46 i dont understand the whole virtual mapping thing Jul 08 20:07:47 if i say Jul 08 20:07:51 bitbake kernel Jul 08 20:08:04 it says "Nothing provides dependency kernel" Jul 08 20:08:12 that is right Jul 08 20:08:18 ok - so clue me in here :) Jul 08 20:08:42 ive got a custom kernel bb and config called stv-kernel, i want to build it Jul 08 20:08:43 jkp__: as you have said. it is "virtual/kernel" and not kernel Jul 08 20:09:08 jkp__: as you might notice I have used "-b A-BB-FILE" where A was kernel Jul 08 20:09:11 ahhh - so if i do that it will clean whatever i have set with the mapping? Jul 08 20:09:15 jkp__: so now you need to create a turing machine Jul 08 20:09:21 hehehehe Jul 08 20:09:31 that replaces the string "A-BB-FILES" with a path to the bb file you use for your kernel Jul 08 20:09:46 and within finite steps you have replaced the string and can ask bitbake to clean it :) Jul 08 20:09:50 heh Jul 08 20:10:15 so, so why when you have RDEPENDS or DEPENDS for the kernel, do you not specify virtual/kernel? Jul 08 20:10:18 you specify kernel Jul 08 20:10:22 jkp__: bitbake -b will not do any dependency handling, it will just open the file after the -b Jul 08 20:10:52 jkp__: I don't get that, but I think the answer is: We do Jul 08 20:11:08 jkp__: but most of the userspace is not depending on the kernel directly (at built) Jul 08 20:12:30 ive seen packages where the RDEPENDS are just kernel, not virtual/kernel Jul 08 20:12:39 you say i should use virtual/kernel though right? Jul 08 20:13:55 jkp__: then this is likely to be a bug Jul 08 20:14:06 ok - if i see it again ill let you know Jul 08 20:14:40 the whole virtual thing is just to let you map a package name to something specific for your distro / machine? Jul 08 20:17:18 zecke: i need one more small bit of help Jul 08 20:17:53 im trying to use bootimage.bbclas Jul 08 20:18:06 I have not used it myself Jul 08 20:18:17 it looks for the kernel in $STAGING_KERNEL_DIR Jul 08 20:18:22 for some reason this is wrong on mys system Jul 08 20:18:28 any ideas where it is set? Jul 08 20:19:01 3:0 Jul 08 20:19:03 jesus Jul 08 20:19:05 jkp__: your kernel.bb files should put the bzImage there Jul 08 20:19:30 well, they do, but where they put it is diffrerent to where bootimage.bbclass thinks it is Jul 08 20:20:07 its looking in tmp/work/i586-linux/kernel Jul 08 20:20:20 no that is wrong Jul 08 20:20:32 STAGING_KERNEL_DIR is tmp/work/i586-linux/kernel ? Jul 08 20:20:56 oops Jul 08 20:21:18 its looking in tmp/staging/i586-linux/kernel/ Jul 08 20:21:38 but its actually in tmp/staging/mini_itx-linux/kernel/ Jul 08 20:21:53 we created a custom kernel config and .bb Jul 08 20:22:58 jkp__: you have two options: Add a new variable to bootimg.class Jul 08 20:23:08 jkp__: declaring path and filename of the kernel Jul 08 20:23:19 jkp__: or install your kernel the way bootimg assumes it Jul 08 20:23:23 heh - seems hacky but yes :) Jul 08 20:23:57 well. Looking at the head of bootimg.bbclass, we can just add another variable Jul 08 20:23:59 is there no nice way to make it recognise what kernel im building and therefore where it should be? Jul 08 20:24:16 jkp__: no, we lack that sort of thing Jul 08 20:24:32 jkp__: and it is tricky as you could decide to build different kernels etc... Jul 08 20:24:35 heh - ok...as long as i know im not hacking when there is a better way Jul 08 20:24:47 so the best way in your opinion? Jul 08 20:25:13 $STGAING_KERNEL_DIR7/bzImage is not a bad assumption Jul 08 20:25:23 the only assumption is that this will be a linux/x86 kernel Jul 08 20:25:23 * Harvy is away: playing pool Jul 08 20:25:32 which is true as it is syslinux Jul 08 20:25:41 why does STGAING_KERNEL_DIR map to x86? Jul 08 20:25:44 so I would just install the kernel into STAGING_KERNEL_DIR Jul 08 20:25:50 hi all Jul 08 20:25:57 jkp__: not the variable, bzImage and syslinux do Jul 08 20:26:30 ok - so change my kernel .bb file to force the install in STAGING_KERNEL_DIR ? Jul 08 20:26:39 do_deploy there Jul 08 20:27:05 heh - it has not do_deploy section...i guess it is inheritted Jul 08 20:27:11 which makes chaning it a pain Jul 08 20:27:25 no, it is likely empty Jul 08 20:27:36 and then there is always do_deploy_append/_prepend Jul 08 20:27:45 Isn't there something in kernel.bbclass which determines where the resulting image is? Jul 08 20:28:09 you are the kernel master :) Jul 08 20:29:02 KERNEL_OUTPUT Jul 08 20:29:05 hi Jul 08 20:29:08 03cbrake 07org.oe.dev * r01860128... 10/packages/python/python-2.4.3-manifest.inc: python-netclient: add Cookie* files to package Jul 08 20:29:09 03cbrake 07org.oe.dev * r3307181a... 10/packages/python/python-cheetah_1.0.bb: python-cheetah 1.0: add package to OE Jul 08 20:29:12 hi hrw Jul 08 20:29:12 zecke:i am writing a keypad to thumbscript userspace keyboard driver at the moment it is nearly compleate. cause you found thumbscript interesting you could be my first tester? Jul 08 20:29:13 03cbrake 07org.oe.dev * re24156fa... 10/packages/python/python-webpy_0.138.bb: python-webpy 0.138: add missing RDEPENDS Jul 08 20:29:15 ~kill linux Jul 08 20:29:16 * ibot shoots a hyper-charged proton gun at linux Jul 08 20:29:23 RP - what i dont get is that STAGING_KERNEL_DIR seems to be different in the two cases Jul 08 20:29:43 fro the kernel.bbclass its one place, and for bootimg.bbclass its another Jul 08 20:30:27 why does linux cannot get ANY option to kill 'D' process... or make their resources freed... I do not like to being forced to reboot just because my MUA died Jul 08 20:31:14 hrw: It is annoying - I had that issue with my laptop the other day (wouldn't suspend any longer due to it) Jul 08 20:31:39 Any idea how I can determine how many bits my target has? Jul 08 20:31:46 RP: I have it too often Jul 08 20:31:46 jkp__: I not seen bootimg.bbclass before and its very new. I suspect its a bug or it needs features adding Jul 08 20:32:03 RP: right, i see the variable - shouldnt the bootimg.bblass, inherit from kernel.bbclass Jul 08 20:32:05 therefore inheritting that var Jul 08 20:32:08 therefore able to use it to locate the kernel? Jul 08 20:32:15 wouldnt that be a nicer solution? Jul 08 20:32:39 jkp__: That depends what bootimage.bbclass is doing. It certainly sounds nicer... Jul 08 20:32:54 jkp__: Where possible we do reuse classes, often quite heavily Jul 08 20:33:19 i think that sounds like the way forward to me Jul 08 20:33:38 if the kenrel is installed there everytime, it makes sense to look it up there Jul 08 20:33:51 cleaner than some nasty hacking using overrides or some shit Jul 08 20:34:09 * jkp__ tries his idea to see if it would work Jul 08 20:34:15 jkp__: You need to ask the author of that calss about it Jul 08 20:34:23 CosmicPenguin Jul 08 20:34:33 i was talking to him about it last night - i will mention it Jul 08 20:34:41 (when hes on again) Jul 08 20:39:17 heh..my home machine has C degrees inside of box and getting segfgfaults ;( Jul 08 20:39:42 he hrw Jul 08 20:42:14 so, the STAGING_KERNEL_DIR and the KERNEL_OUTPUT dir - why are they different? Jul 08 20:44:22 jkp__: KERNEL_OUTPUT is where the kernel build tree generates the kernel. STAGING_KERNEL_DIR is where the kernel is staged two. Totally different things Jul 08 20:44:56 jesus Jul 08 20:44:57 http://gnumonks.org/~laforge/weblog/2006/07/06/#20060706-shanghai-armzone Jul 08 20:45:28 is the kernel removed from the staging dir after it is built? Jul 08 20:46:31 jkp__: Its more likely it gets placed in staging after its built Jul 08 20:46:39 jkp__: staging != workdir Jul 08 20:47:14 im lost, i just dont understand why the bootimg class is trying to look it up in a place where it is not i guess :/ Jul 08 20:47:18 noobie i now Jul 08 20:47:21 *know Jul 08 20:47:39 woglinde: awesome Jul 08 20:49:54 hrw yes Jul 08 20:50:26 RP: ok so accepting i need to hack Jul 08 20:50:30 hrw reminds how stupid we believe the price for soemthing is fair Jul 08 20:50:43 i tried to set KERNEL_STAGING_DIR to the place where my kernel is Jul 08 20:50:57 no go, it still looks in the old place...is there a special syntax for an override? Jul 08 20:51:05 damn, i've spent whole day trying to make sound out of poodle with 2.6 Jul 08 20:51:30 I hate whoever invented that damn LoCoMo chip Jul 08 20:51:43 ~locomo Jul 08 20:51:44 i guess locomo is a completely undocumented chip in the Sharp Zaurus Jul 08 20:51:52 hm Jul 08 20:52:06 ixoth has some written to sharp for details? Jul 08 20:52:06 ixoth: poodle use wm8731 for audio - we have driver for it as same one is in c7x0 Jul 08 20:52:08 what's FILES_${PN} about? Jul 08 20:52:27 tkp_: it means: PN will contain FILES Jul 08 20:52:29 tkp_: PACKAGES and FILES relate Jul 08 20:52:40 tkp_: so you PACkAGE = " Jul 08 20:52:58 and PN is? Jul 08 20:52:59 tkp_: so you have PACkAGES = " ${PN} ${PN}-doc ${PN}-dev " Jul 08 20:53:15 Package Name? Jul 08 20:53:18 tkp_: and with FILES_${PN} you say for package ${PN} install these files Jul 08 20:53:23 tkp_: exactly Jul 08 20:53:39 nite Jul 08 20:53:52 hrw: but the problem is that poodle uses Locomo chip for i2s access to wm8731, and c7x0 uses scoop Jul 08 20:54:18 well, in fact i've got some results, but still no sound :( Jul 08 20:55:16 ah Jul 08 20:55:38 ixoth: so you are hacking poodle/2.6 sound support? Jul 08 20:56:19 http://wiki.openzaurus.org/KernelInfo26/MachineStatus Jul 08 20:56:34 well, i'm trying to do that... it seemed simple at the beginning, but then it became harder & harder Jul 08 20:58:20 RP: w00t - i have a bootimage Jul 08 20:58:27 * jkp__ learned how to use OVERRIDES Jul 08 20:58:44 awesome Jul 08 20:58:50 zecke: :D Jul 08 21:03:38 is there some trick to loop mount jffs2 images? Jul 08 21:03:47 emte: yes Jul 08 21:03:57 emte: mtdram Jul 08 21:04:00 emte: mtdram, dd and mount :) Jul 08 21:04:17 google: mount jffs2 mtdram Jul 08 21:04:19 mtdram .. Jul 08 21:04:35 yeah i've tried ~ 10 permutations of mount byitself Jul 08 21:06:13 maybe i should add that link to Eriks pivot wiki page ... Jul 08 21:13:04 hardly OT - does anyone know how the key on the american keyboard(http://www.kachold.de/tastbel.html) on top left under the esc is called(not the wave the sign under the wave which looks like a apostrophe)? or even better knows the keycode? many thx. Jul 08 21:13:29 tilde? Jul 08 21:13:30 backtick? Jul 08 21:13:39 hey hrw Jul 08 21:13:42 hi koen Jul 08 21:13:59 mr_nice_slacker: its ` (backtick) and ~ (tilde) Jul 08 21:15:20 03freyther 07org.oe.dev * rdff694f0... 10/classes/tinderclient.bbclass: Jul 08 21:15:20 class/tinderclient.bbclass: Send bigger files in smaller chunks to avoid broken firewalls Jul 08 21:15:20 Thanks to the firewall fascists of A???? longer HTTP requests were lost in the corporate firewall. Avoid this situation by sending the data in smaller chunks. Jul 08 21:15:24 03freyther 07org.oe.dev * r9e87109e... 10/packages/xerces-c/xerces-c_2.6.0.bb: Jul 08 21:15:24 packages/xerces-c/xerces-c_2.6.0.bb: Use configure directly instead of runConfigure Jul 08 21:15:25 To fix bug #1154 where CXX="ccache arm-linux-g++ ..." got mangled - thanks to Jul 08 21:15:28 getopt - into CXX=ccache. Using ./configure directly and setting CFLAGS,CXXFLAGS, Jul 08 21:15:29 LIBS and other variables we totally avoid this issue. Only one issue remains Jul 08 21:15:32 we have to set BUILDBITS and these are set to 32. We need a way to know how many Jul 08 21:15:33 bits our target has. Jul 08 21:16:12 thx - something from this(http://pastebin.ca/82723) list must be it but there is no backtick :( Jul 08 21:16:56 did you look in your keyboard map? Jul 08 21:17:33 i have german keyboard map the list is from /usr/include/linux/input.h Jul 08 21:18:48 what about /usr/share/keymaps ? Jul 08 21:19:06 in the keyboard map the keys are called equal or are there signs? - i will try Jul 08 21:20:23 mr_nice_slacker: on my pc: keycode 41 = grave asciitilde Jul 08 21:20:52 hmm i needto reinstal gcat Jul 08 21:21:22 hrw: thx Jul 08 21:21:30 hrw: grave is in the list :) Jul 08 21:22:48 np Jul 08 21:23:11 * koen is going to name his first device 'umbaumba' Jul 08 21:23:26 or I could convince mickeyl into giving angstrom releases codenames Jul 08 21:25:07 okay ... i dd the jffs2 image into a mtd device, then try to mount the device correct? ( all the indicated mtd modules are loaded ) Jul 08 21:25:52 i am still getting the same errors Jul 08 21:27:33 there must be an easier way to extract /bin and /sbin Jul 08 21:28:30 koen: I was thinking that angstrom 2006.9 had to be called 'umbaumba' Jul 08 21:32:26 goodbye everyone, have to go Jul 08 21:34:40 hey guys - i could do with a hand writing my first .bb file for a new package Jul 08 21:34:57 i dont need anything right now, but beware...ill be asking annoying questions im sure ;) Jul 08 21:35:07 (whats new eh? :p) Jul 08 21:35:14 I'm started to get annoyed in preparation Jul 08 21:35:34 zecke: hhehe - i lie, its not my first, ive written meta packages Jul 08 21:35:42 but nothing that actually builds and installs sw Jul 08 21:36:03 i need a native version which i need to use in another package, and also a target version of the same package Jul 08 21:36:17 can you think of a simple example of this i could copy? Jul 08 21:36:45 ipkg maybe i guess Jul 08 21:36:50 nano_1.3.12.bb - most famous example Jul 08 21:36:50 or zlib Jul 08 21:36:59 ok :) Jul 08 21:37:05 i need to use zlib so thats good Jul 08 21:37:22 i take it the native version of zlib puts the headers in the right place? Jul 08 21:37:50 oh wait Jul 08 21:37:59 doesnt have a target version :) Jul 08 21:38:01 I've actually written a howto for that Jul 08 21:38:08 koen: w00t! Jul 08 21:38:09 jkp__: http://www.openembedded.org/user-manual&dpage=ch02s04 Jul 08 21:40:21 am i allowed to BSD license my efforts? Jul 08 21:40:27 or is that frowned upon? Jul 08 21:40:38 oe metadata is MIT licensed Jul 08 21:40:38 (its my preferred license) Jul 08 21:40:41 we prefer the .bb files themself to be MIT Jul 08 21:41:01 im not familiar with this license...which is it most similar to? Jul 08 21:41:26 to BSD Jul 08 21:41:33 http://www.opensource.org/licenses/mit-license.php for details Jul 08 21:41:42 ta Jul 08 21:41:55 I do not remember diffs between BSD/MIT/X license Jul 08 21:42:32 "do what you like, but mention us and don't sue" Jul 08 21:42:54 heh :) Jul 08 21:43:08 koen: there was also BSD without 'mention us'? Jul 08 21:43:21 iirc yes Jul 08 21:43:32 BSD minus advertising clause Jul 08 21:44:00 the bsd license also forces you to grow a beard and wear sandals Jul 08 21:44:07 * koen hides Jul 08 21:44:22 if i put something in RDEPENDS, it will build automatically right? i dont need to put it in DEPENDS too right? Jul 08 21:44:35 * jkp__ notes he currently has a beard and was wearing sandles approx 2 hours ago ;) Jul 08 21:44:52 jkp__: the short answer is "yes" Jul 08 21:45:22 * jkp__ wonders what the long answer is but refrains from asking :) Jul 08 21:45:46 everything is dynamically linked on OE right? Jul 08 21:45:58 ie - if something needs zlib to build its an RDEPEND Jul 08 21:46:06 if it's needed for the build proces it's needs to be in DEPENDS as well, in case we start writing a multithreaded bitbake Jul 08 21:46:25 jkp__: by default we use dynamic linking Jul 08 21:46:37 ok, so in this case it needs to be in both Jul 08 21:46:41 jkp__: but you can instruct OE to do static linking Jul 08 21:46:53 i wont bother ;) Jul 08 21:46:56 tkp_: xerces-c should work Jul 08 21:47:08 tkp_: for dosfstools-native I would need an AMD64 machine Jul 08 21:47:27 zecke: just updating my db Jul 08 21:47:32 thanks for that Jul 08 21:47:33 tkp_: or could you run gcc with -E instead of -c Jul 08 21:47:49 tkp_: to see what it thinks about __u8 Jul 08 21:48:10 zecke:where would I need to set that? Jul 08 21:48:37 tkp_: e.g. in your(?) bugreport foo.c fails to compile Jul 08 21:48:58 tkp_: you need to use this compile line, substitute -c with -E pastebin the resulting output Jul 08 21:51:01 zecke: oe_runmake? Jul 08 21:51:21 tkp_: no :) Jul 08 21:52:02 tkp_: is that your bug report?: http://bugs.openembedded.org/show_bug.cgi?id=1155 Jul 08 21:52:14 'night all Jul 08 21:52:14 make[1]: Entering directory Jul 08 21:52:14 `/home/tom/stv/oe-stv/build/tmp/work/x86_64-linux/dosfstools-native-2.10-r2/dosfstools-2.10/mkdosfs' Jul 08 21:52:18 | ccache gcc Jul 08 21:52:19 zecke: yes Jul 08 21:52:20 -isystem/home/tom/stv/oe-stv/build/tmp/staging/x86_64-linux/include -O2 -c Jul 08 21:52:23 mkdosfs.c -o mkdosfs. Jul 08 21:52:32 tkp_: you see that from within ...-2.10-r2/dosfstools-2.10/mkdosfs Jul 08 21:52:41 ccache gcc ... was invoked Jul 08 21:53:04 now substitute -c with -E, run the command Jul 08 21:53:10 ~pastebin the preprocessed file Jul 08 21:53:18 zecke: is this in the tmp dir? Jul 08 21:53:29 I'm looking at the .bb file atm Jul 08 21:53:43 tkp_: bitbake was so nice to print the path for you Jul 08 21:53:51 tkp_: just open your eyes, try Jul 08 21:54:06 sorry... I have severall things on the go right now. Jul 08 21:54:13 I'll take a proper look now. hold on Jul 08 21:54:53 * emte curses jffs2 Jul 08 21:57:38 zecke, i dont suppose you know of a link off hand that deals with mounting jffs2 on a host using udev? all the ones i've found sofar use methods that dont work Jul 08 21:57:56 oh Jul 08 21:58:04 modinfo mtdram Jul 08 21:58:14 modprobe mtdram PARAM=SIZE works? Jul 08 21:58:22 zecke: I think that did it Jul 08 21:58:24 it does create mtd devices? or is this the issue? Jul 08 21:58:35 no output with E instead of c Jul 08 21:58:35 tkp_: well, it is only preprocessing Jul 08 21:58:41 ~pastebin Jul 08 21:58:43 hmm... pastebin is a place to paste your stuff without flooding the channel - try http://pastebin.com/, or http://pastebin.ca, or http://channels.debian.net/paste, or for #oe use http://oe.pastebin.com, or http://bzflag.pastebin.ca/ Jul 08 21:58:47 cu Jul 08 21:58:48 tkp_: you know what preprocessing is? Jul 08 21:58:57 hrw|gone: cya Jul 08 21:59:00 it creates the devices but /dev is mounted with a 10MB size which the jffs2 exceeds Jul 08 21:59:00 zecke: nope :p Jul 08 21:59:18 zecke: I ran ccache gcc -isystem/home/tom/stv/oe-stv/build/tmp/staging/x86_64-linux/include -O2 -E mkdosfs.c -o mkdosfs.o Jul 08 21:59:22 and got no output Jul 08 21:59:35 emte: then your dd does something wrong Jul 08 21:59:48 tkp_: well the output is in mkdosfs.o Jul 08 22:00:01 zecke, how so? dd doesnt control the size of /dev udev does .... Jul 08 22:00:12 tkp_: you should get a book about the C Programming Language, and should get familiar with make, autotools, compiling and preprocessing Jul 08 22:00:23 emte: why is the size of /dev important? Jul 08 22:00:33 zecke: yes, I'm still learning... Jul 08 22:00:41 emte: you just create a handfull of mtdblockX and mtdX nodes Jul 08 22:00:44 zecke: im teaching him Jul 08 22:00:47 but I know shit loads more than I did two weeks ago ;) Jul 08 22:00:48 bit by bit Jul 08 22:00:49 ah :) Jul 08 22:00:53 he's a web programmer Jul 08 22:00:54 all the instructions ive tried reference dd if=image.jffs2 of=/dev/mtd0 Jul 08 22:00:56 i do the c stuffs Jul 08 22:01:02 :) Jul 08 22:01:06 but hes getting there Jul 08 22:01:13 he knows how headers and includes work now :) Jul 08 22:01:17 hehe, I wondered about the'kp_' similiarity Jul 08 22:01:18 and linking Jul 08 22:01:20 zecke: you you want me to paste the entire output contents of that file? Jul 08 22:01:23 hehehe Jul 08 22:01:23 it's prety largs Jul 08 22:01:26 large* Jul 08 22:01:31 tkp_: ~pastebin Jul 08 22:01:38 we are say about 2 meters away from each other zecke Jul 08 22:01:39 tkp_: we will kill you if you paste it here Jul 08 22:01:44 hahahaha Jul 08 22:01:44 hehehe Jul 08 22:01:51 tkp_: using pastebin is okay :) Jul 08 22:01:53 ~pastebin Jul 08 22:01:55 i guess pastebin is a place to paste your stuff without flooding the channel - try http://pastebin.com/, or http://pastebin.ca, or http://channels.debian.net/paste, or for #oe use http://oe.pastebin.com, or http://bzflag.pastebin.ca/ Jul 08 22:02:39 or rafb.net/paste if pastebin is too slow Jul 08 22:03:08 last time I tried to paste something that large it timed out Jul 08 22:03:30 "Maximum execution time of 30 seconds exceeded in" Jul 08 22:03:31 :( Jul 08 22:04:21 anyway I will move to bed now Jul 08 22:04:32 zecke: http://bzflag.pastebin.ca/82776 Jul 08 22:04:37 I will consider upgrading and testing dosfstools 2.11 Jul 08 22:04:55 tkp_: thanks, I will bookmark it and look into it tomorrow Jul 08 22:05:19 mm. my FC5 box has 2.11 on it and that works with gcc 4.1.1 Jul 08 22:05:28 night zecke . Thanks Jul 08 22:05:34 dd: writing to `/dev/mtd0': No space left on device Jul 08 22:05:34 <- my eternal issue Jul 08 22:06:13 emte: try to mount it now :) Jul 08 22:06:23 there is nothing in it Jul 08 22:06:44 i was hoping it would atleast have a partial Jul 08 22:06:46 hmm it has been years since I last mounted jffs2 loopback Jul 08 22:07:11 emte: is mtd0 a charachter node? Jul 08 22:07:16 or just a normal file? Jul 08 22:07:47 -rw-r--r-- 1 root root 0 2006-07-08 15:27 /dev/mtd0 Jul 08 22:07:51 looks like its cahr Jul 08 22:07:55 char* Jul 08 22:08:17 it does? Jul 08 22:08:36 I'm not good as these things but I would expect to see a 'c' and major and minor number Jul 08 22:08:42 0 is a char device is it not? Jul 08 22:08:54 3 is a block if i remember correct Jul 08 22:09:10 tkp_: okay, my x86 preprocessed file has a typedef for __u8; you lack it Jul 08 22:09:30 tkp_: mine is coming from asm-i386/types.h, mkdosfs should use stdints.h anyway... Jul 08 22:09:42 * emte looks it up Jul 08 22:10:23 zecke: you going to patch it? Jul 08 22:10:34 (to use stdints.h) Jul 08 22:10:42 jkp__: I will compile and try dofstools 2.11 tomorrow Jul 08 22:10:47 sure Jul 08 22:10:51 no rish :) Jul 08 22:10:55 *rush Jul 08 22:11:06 ewww. that must mean something like im tired Jul 08 22:11:11 where are you zecke btw? Jul 08 22:11:25 germany Jul 08 22:11:33 not far then Jul 08 22:11:50 not a football fan then? :p Jul 08 22:11:51 zecke, i am wrong ... it is missing the 'c' before the permissions Jul 08 22:12:18 hehe Jul 08 22:12:25 well I have watched the game Jul 08 22:13:38 ouch the workaround in 2.11 is to #define __KERNEL__ before including the types.h :} Jul 08 22:13:43 morons Jul 08 22:13:50 heh Jul 08 22:16:23 how do i make my build use fakeroot? Jul 08 22:16:32 or does that happen automatically? Jul 08 22:16:48 jkp__: fakeroot do_stuff() {} Jul 08 22:16:55 jkp__: should make it use fakeroot Jul 08 22:17:17 hmm - getting odd warnings in the build Jul 08 22:17:29 ill need to get the log from somewhere in a mo to see what they were Jul 08 22:18:25 cya Jul 08 22:21:16 can anyone tell me why i might be seeing multiple repetitions of the error "Couldn't find shared library provider for libgcc_s.so.1"? Jul 08 22:22:10 multiple things are looking/depend on it? Jul 08 22:22:32 emte: i dont follow sorry :/ Jul 08 22:22:50 anything to do with having things both in DEPENDS and RDEPENDS at the same time? Jul 08 22:22:54 i would thing you'd get the message everytime something went looking for it Jul 08 22:22:59 think Jul 08 22:23:15 ok, and we dont have it? Jul 08 22:23:17 well DEPENDS is buildtime Jul 08 22:23:18 seems odd Jul 08 22:23:27 RDEPENDS is actual runtime on the device Jul 08 22:23:54 and perhaps at image creation time ... Jul 08 22:23:56 right, but if something needs to link to a lib then it needs to be in both right? Jul 08 22:24:01 no Jul 08 22:24:04 no? Jul 08 22:24:27 but how can you ensure it is built before the package you are buidling, and is present on the device? Jul 08 22:24:34 it would depend on library linking and if it expected another app ti be present Jul 08 22:25:12 hang on - this is a simple case, just tell me which one i need to use and why if thats cool... Jul 08 22:25:19 DEPENDS is for build time requirements Jul 08 22:25:33 so, i have an app that links to libcurl, zlib and readline Jul 08 22:25:34 RDEPENDS is for actual running Jul 08 22:25:43 right, but the two are the same surely? Jul 08 22:25:47 no Jul 08 22:26:01 DEPENDS does not do anything after OE/bitbake are done Jul 08 22:26:05 to buld it, the libs need to be present Jul 08 22:26:14 RDEPENDS is used by ipkg Jul 08 22:26:32 right, which is why i say i need both Jul 08 22:26:37 i need to link to them at build time Jul 08 22:26:46 and to use RDEPENDS to make sure they are packaged properly Jul 08 22:26:50 no? Jul 08 22:26:52 no Jul 08 22:27:11 so RDEPENDS alone is enough? Jul 08 22:27:12 rdpends has nothing to do with packaging, but the required package for it to run Jul 08 22:27:34 will packages in RDEPENDS be built first? Jul 08 22:27:38 best example i can think of would be ssl gnupg Jul 08 22:27:58 those are split into application, libraries, documentation etc Jul 08 22:28:19 do dont generically say the ssl package Jul 08 22:28:36 you woudl rdepends most likly on the library package Jul 08 22:29:03 right, coz you want it on the box Jul 08 22:29:31 well you need it if you compile dymanic libraries Jul 08 22:29:40 sure Jul 08 22:29:51 otherwise it would be compiled in already Jul 08 22:30:04 ok - i think i get that anyway...just about... Jul 08 22:30:19 im still confused by the libgcc_s.so.1 error Jul 08 22:30:23 that seems like bad news Jul 08 22:30:33 can you explain what you think is causing it? Jul 08 22:30:47 your rdepends minght be openssl-libraries while your depends would be openssl Jul 08 22:30:48 presumably something is not right Jul 08 22:31:11 ok - now i get it Jul 08 22:31:35 so you build the whole openssl package (ie headers etc as well) for your actual build, but for running you only need the libs Jul 08 22:31:45 that is as far as i understand the depends rdepends differences anyway Jul 08 22:31:53 yes Jul 08 22:32:03 ok - good, i think i follow that now Jul 08 22:32:15 so are those targets defined within one package? Jul 08 22:32:25 which? Jul 08 22:32:28 ie - there is a package called openssl Jul 08 22:32:33 the packages split out ? Jul 08 22:32:42 but you can say RDEPENDS=openssl-libs Jul 08 22:32:48 are they seperate .bb files? Jul 08 22:32:49 look in your packages dir Jul 08 22:33:04 there are some 3k+ packages you can use Jul 08 22:33:11 yes Jul 08 22:33:28 have you looked in your deploy dir or at one of the feeds? Jul 08 22:33:38 the feeds? nope Jul 08 22:33:47 i'll get you a link Jul 08 22:34:19 http://ftp.handhelds.org/distributions/familiar/releases/v0.8.2/feed/world/ Jul 08 22:34:58 right...lots of ipks? :p Jul 08 22:35:06 things are optimaly split appart so you can use the least amount of required items Jul 08 22:35:29 the locale stuff is probably the most obvious Jul 08 22:35:54 since every locale is split out for each package taht needs/provides them Jul 08 22:36:20 ok - that all makes sense, but i want to know how to do the same in my own packages now :) Jul 08 22:36:35 most of it is automated by default Jul 08 22:36:47 based on the Makefile Jul 08 22:37:06 sometimes manual manipulation is needed etc Jul 08 22:37:43 probably more often than not ... Jul 08 22:38:09 ah - now looking at the openssl package Jul 08 22:38:30 it puts stuff in ${STAGING_INCDIR}, ${STAGING_LIBDIR} etc Jul 08 22:38:46 is it by doing that that you define the seperate component ipks? Jul 08 22:38:52 no Jul 08 22:39:02 taht is for the buildsystem includes Jul 08 22:39:04 i dont see where that happens Jul 08 22:39:16 ahhh - so that other stuff can use it after Jul 08 22:39:24 you only push stuff to staging for building Jul 08 22:39:28 right Jul 08 22:39:31 someone with kernel 2.6 evdev uinput and no fear of resets will try my key2thumb driver? it is qnd at the moment (i am on lerning c). but could be tested. if you don't know thumbscript you can see it here (thumbscript.com). Jul 08 22:39:43 it is userspace Jul 08 22:39:46 so if i was going to write a -native package Jul 08 22:39:59 i wouldnt inherit ipk , id just install stuff to staging Jul 08 22:40:00 ? Jul 08 22:40:04 you would push things to staging Jul 08 22:40:07 yes Jul 08 22:40:10 ahhhha Jul 08 22:40:13 the penny drops Jul 08 22:40:32 i stil dont understand how to split things out into seperate targets from one .bb Jul 08 22:40:47 you will find out if you need to do it Jul 08 22:40:53 well Jul 08 22:40:59 i have one case where i do Jul 08 22:40:59 ( missing files/rdepends ) Jul 08 22:41:03 but its not urgent Jul 08 22:41:19 its only removing three two binaries from the final system Jul 08 22:41:26 you will have to manually code it in to happen/bundle properly Jul 08 22:41:41 there are three in the package and i only want one in the final system Jul 08 22:41:47 ok - ill worry about it later :) Jul 08 22:41:57 there are lots of examples around :) Jul 08 22:42:03 3k+ Jul 08 22:42:04 im still worried about that gcc error i pasted Jul 08 22:42:06 should i be? Jul 08 22:42:09 hehe :) Jul 08 22:42:19 wlan-ng-modules splits itself into several packages Jul 08 22:42:37 if it is an error, maybe, if its a warning, not until something breaks because of it Jul 08 22:42:52 seems to be a warning Jul 08 22:42:59 are you planning to use gcc on the target? Jul 08 22:43:02 i mention it because i was having problems with gpe-image not seeing its wlan-ng-modules-usb, even though everything seems to build correctly ;) Jul 08 22:43:20 * zwelch is still trying to figure that one out Jul 08 22:43:54 i think that supports the "worry about it later" train of thought :) Jul 08 22:53:30 * Harvy is back Jul 08 23:44:15 i still cant get this to work Jul 08 23:44:28 maybe i'll have to email Erik ... Jul 08 23:48:00 * thejapa sends good vibes to emte Jul 08 23:48:12 lol Jul 08 23:48:34 i dont think that will help my trying to mount a jffs2 image Jul 08 23:49:11 okay Jul 08 23:49:14 maybe this will help Jul 08 23:49:37 mount -o loop image.jffs2 /mnt/temp Jul 08 23:49:39 hehehe Jul 08 23:49:50 cant do that with jffs/jffs2 Jul 08 23:50:00 j/k :) Jul 08 23:50:13 you have to use mtd Jul 08 23:50:31 my problem is i cant copy my image to an mtd device Jul 08 23:50:38 i dont have enough room in /dev Jul 08 23:50:53 * thejapa remembers you had to erase it first, using mtdtools something like that. Jul 08 23:51:00 I did it only once. :( Jul 08 23:51:18 i suspect its a udev issue on my host Jul 08 23:51:33 since /dev is mounted with a size of 10MB Jul 08 23:52:01 * emte contemplates remounting /dev .... Jul 08 23:52:18 i know very little of udev Jul 08 23:52:31 i knew a bit of devfs though, only as a user Jul 09 00:21:28 weird ... Jul 09 00:21:55 is anyone aware of strange datestamps in OE images? Jul 09 00:22:06 tar: bin: implausibly old time stamp 1969-12-31 16:00:00 Jul 09 00:23:27 i finally figured out what i was doing wrong Jul 09 00:28:26 wow oe kept going without a touch until gpe-image.bb itself failed ... Jul 09 00:28:36 that is yet impressive Jul 09 00:46:33 Anybody out there remember how to pull multiple branches at once in monotone? Jul 09 01:07:33 night all Jul 09 01:11:52 hvontres|home, its in the wiki Jul 09 01:12:00 in the monotone phrasebook page **** ENDING LOGGING AT Sun Jul 09 02:59:56 2006