**** BEGIN LOGGING AT Tue Aug 22 02:59:56 2006 Aug 22 06:48:43 bonjour ! Aug 22 07:21:06 morning Aug 22 07:39:03 morning Aug 22 07:39:14 hi XorA Aug 22 07:39:50 RP: bitbake trunk + 'bitbake bootstrap-image' == nothing built Aug 22 07:44:10 hrw|work: speed optimsation :-) Aug 22 07:45:21 morning all Aug 22 07:45:23 yep Aug 22 07:45:41 hey RP Aug 22 07:46:16 RP: what I have to do to build image with trunk? Aug 22 07:46:42 hrw|work: bitbake bootstrap-image should work fine... Aug 22 07:47:40 hrw|work: What was the output from bitbake? Aug 22 07:47:46 RP: nope. it do boostrap-image/do_fetch then do_rootfs and fails due to lack of makedevs Aug 22 07:48:01 re Aug 22 07:48:12 hrw|work: So it did try to build something :) Aug 22 07:48:16 RP: will pastebin log Aug 22 07:48:25 hrw|work: Is this with a clean tmp? Aug 22 07:48:29 clean Aug 22 07:48:36 sounds like hrw may be hitting what i hit when i tried it last week Aug 22 07:48:45 btw - why .dev still has 2.4 zaurus kernels?? Aug 22 07:49:18 kergoth: I checked the new dependency code into trunk on Sunday night so maybe, maybe not Aug 22 07:49:20 http://pastebin.ca/143944 is enough? Aug 22 07:49:26 ah, cool Aug 22 07:49:34 hrw|work: I'd like them left in case we need them for dev purposes... Aug 22 07:49:58 kergoth: It also has basic multithreading capability Aug 22 07:50:13 * kergoth has a bitbake trunk poky build chugging along as we speak Aug 22 07:50:35 hrw|work: Is this the oz354 branch? Aug 22 07:50:37 RP: then we can take them from history - its SCM... Aug 22 07:50:39 RP: .dev Aug 22 07:51:46 hrw|work: It didn't build anything else? Aug 22 07:52:00 nope Aug 22 07:52:15 hrw|work: and you have the changes I checked into .dev for the new bitbake? Aug 22 07:52:21 RP: thats whole log - first it complain about lack of gcc295 for 2.4 kernels Aug 22 07:52:41 RP: probably now - checkouting .dev again due to monotone fscking Aug 22 07:53:05 RP: retrying with recent .dev Aug 22 07:53:09 hrw|work: Its as if the dependency code in image_ipk.bbclass is missing Aug 22 07:53:42 * zecke has found a free Quad Xeon in a remote rack Aug 22 07:53:54 RP: hrw|work : I think I can continue the builds there Aug 22 07:54:18 * kergoth got his new workstation built, is running 3 simultanious oe builds atm without a hikkup :) Aug 22 07:55:11 kergoth: Try multithreading with bitbake head - it runs with poky :) Aug 22 07:55:13 nice zecke Aug 22 07:56:23 zecke: nice toys you find Aug 22 07:57:49 RP: now it goes Aug 22 07:57:51 RP: restarting the build with that set to 3.. *crosses fingers* Aug 22 07:58:23 RP: why it try openzaurus-sa at all when MACHINE=tosa? "NOTE: Removing failed build target openzaurus-sa" Aug 22 07:58:38 RP: same with few other !tosa kernels Aug 22 07:59:29 hrw|work: I suspect it evaluates all providers for virtual/kernel and removes the broken ones (e.g. ones requiring 2.95.3) Aug 22 07:59:38 now I need to transfer my scripts... Aug 22 07:59:50 koen: probably Aug 22 07:59:56 zecke: 'svn checkout'? Aug 22 07:59:59 NOTE: package gnu-config-native-0.1+cvs20050701-r4: task do_package: completed Aug 22 08:00:08 funny... native and packaging.. Aug 22 08:00:09 and a good morning to all as well Aug 22 08:00:15 hrw|work: my build scripts, with auto cleanup, etc... Aug 22 08:00:16 hi koen too Aug 22 08:00:33 zecke: put them in svn or use rsync? Aug 22 08:00:48 hrw|work: I need to touch the sample local.conf Aug 22 08:01:32 native.bbclass should not set PACKAGES=""? Aug 22 08:01:51 hm.. it does it. Aug 22 08:03:00 RP: usbhost on cx00 is ohci-hcd? Aug 22 08:03:12 hrw|work: yes Aug 22 08:03:46 RP: hm. I want to autoload it on tosa but probably autoloading it on cx00 is not quite what we want? Aug 22 08:03:54 hrw|work: ATM I'm a bit scared we need to move the machine into a different net... Aug 22 08:04:10 koen, hrw|work: correct, it evaluates all providers - it computes the dependency tree in advance rather than on the fly Aug 22 08:04:44 hrw|work: It will conflict with usb gadgets like g_ether which people probably use more often Aug 22 08:08:13 module_autoload_ohci-hcd_tosa = "ohci-hcd" Aug 22 08:08:14 then Aug 22 08:09:02 hey Aug 22 08:09:08 hrw|work: that would do it Aug 22 08:09:11 ~lart wndows Aug 22 08:09:13 * ibot slams wndows against a large cement Tux Aug 22 08:11:10 RP: kind of getting status of working tasks would be nice Aug 22 08:11:49 RP: it is hard to find what is doing from output Aug 22 08:11:55 hrw|work: It would. We really need to get some kind of UI to report it nicely though... Aug 22 08:13:46 some curses gui with seperate boxes for the various tasks? Aug 22 08:14:29 03hrw 07org.oe.oz354x * rda9e7689... 10/ (1 packages/linux/linux-openzaurus.inc): linux-openzaurus: added dummy do_rm_work to not remove WORKDIR because wlan-ng stuff needs it Aug 22 08:14:38 03hrw 07org.oe.oz354x * rd2658c57... 10/ (4 files in 2 dirs): linux-openzaurus: autoload USB Host on Tosa Aug 22 08:15:48 Hi Aug 22 08:16:53 Is it possible to habe two different ld-linux.so.2 on a system and that one binary is using the standard one and another binaries are using a speacial ld-linux.so.2? Does exist an envirnment variable for that? Aug 22 08:17:07 hrw|work: iirc the 2.4 kernels are still present for sharprom-compatible stuff Aug 22 08:17:28 hrw|work: how to make rm_work happen after do_deploy in zaurus-updater.bb as rm_work breaks the build for that .bb Aug 22 08:18:14 XorA: hmm will pluck it from branch Aug 22 08:18:53 argh.. pluck is not perfect yet... Aug 22 08:19:00 * XorA needs to find space for branch Aug 22 08:19:30 XorA: did you see gpe-fbpanel already? Aug 22 08:19:56 koen: florian was talking about it a couple of days ago, did he checkin when I wasnt looking? Aug 22 08:20:01 * koen is trying catching up Aug 22 08:20:02 hrw@bitbake:~/devel/oe/org.openembedded/packages/linux$ mtn pluck -r65cddc99c679a5ae5c62989d9fc821a6dcc878d6 Aug 22 08:20:03 XorA: yes Aug 22 08:20:05 mtn: applied changes to workspace Aug 22 08:20:08 hrw@bitbake:~/devel/oe/org.openembedded/packages/linux$ mtn dif . Aug 22 08:20:10 # Aug 22 08:20:13 # no changes Aug 22 08:20:15 ;(( Aug 22 08:20:22 With the new bitbake, you could declare do_latebuild as the default task and insert do_rmwork between do_build and do_latebuild Aug 22 08:20:51 koen: that in gpe cvs, I dont tend to monitor that Aug 22 08:21:10 XorA: http://projects.linuxtogo.org/plugins/scmsvn/viewcvs.php/trunk/base/gpe-fbpanel/?root=gpe Aug 22 08:22:42 hrm Aug 22 08:22:56 I suspect that will give an 'access denied' Aug 22 08:22:59 * koen stabs gforge Aug 22 08:23:08 koen: no group Aug 22 08:24:14 XorA: ah well, florian put it in the experimental svn repo for gpe Aug 22 08:24:34 koen: goodo, his changes sounded rockin Aug 22 08:28:17 XorA: svn co svn://projects.linuxtogo.org/gpe/trunk/base/gpe-fbpanel Aug 22 08:28:34 koen: fork of fbpanel for gpe? Aug 22 08:28:45 hrw|work: it seems that way Aug 22 08:28:53 hrw|work: I discovered it this morning Aug 22 08:29:22 I'll ask florian when he's online Aug 22 08:34:32 RP: 2.6.16/.dev differ from 2.6.1.6/.oz354x in few patches: /zlib_inflate-r3.patch /logo_rotate_fix-r1.patch /poodle_partsize-r0.patch /jffs2_longfilename-r1.patch Aug 22 08:35:02 okay solved the pr Aug 22 08:35:06 oblem Aug 22 08:37:41 RP: building with multitask bitbake need more diskspace when rm_work used Aug 22 08:37:48 03pH5 07org.oe.dev * r8d5edb15... 10/ (10 files in 5 dirs): Aug 22 08:37:48 udev: add 097 Aug 22 08:37:48 - drop support for kernels <2.6.15, udevsynthesize replaced by udevtrigger Aug 22 08:37:48 - split out libvolume_id into its own package Aug 22 08:37:48 - I set PKG_libvolume-id-dev="libvolume-id-dev" because otherwise Aug 22 08:37:49 libvolume-id-dev would be renamed to libvolume-id0, too. Aug 22 08:37:51 - DEFAULT_PREFERENCE="-1", it's not run-time tested yet. Aug 22 08:38:11 koen: florian has added autotools, and support for .desktop files from what he said to me Aug 22 08:39:48 hey mikearthur Aug 22 08:39:59 hey koen Aug 22 08:40:15 RP: can you check it? I want to push some changes from branch to dev Aug 22 08:42:19 hey mikearthur Aug 22 08:43:53 yo XorA Aug 22 08:45:58 hrw|work: I'll have a look shortly Aug 22 08:46:19 hrw|work: Sadly, more diskspace is a side effect Aug 22 08:47:29 cr Aug 22 08:47:40 RP: acceptable it must be Aug 22 08:47:49 RP: accepted by users I mean Aug 22 08:48:12 03hrw 07org.oe.dev * r82b20eb0... 10/ (1 packages/linux/linux-openzaurus_2.6.17.bb): linux-openzaurus: bump PR to get dependencies in modules recreated for feeds Aug 22 08:48:17 03pH5 07org.oe.dev * r3a1c2d0d... 10/ (1 packages/udev/udev.inc packages/udev/udev_097.bb): udev: stop overwriting description in udev.inc Aug 22 08:48:22 03hrw 07org.oe.dev * r1f2f012d... 10/ (1 packages/linux/linux-openzaurus.inc): linux-openzaurus: added dummy do_rm_work to not remove WORKDIR because wlan-ng stuff needs it Aug 22 08:48:28 03pH5 07org.oe.dev * rc58d6c1d... 10/ (1 packages/udev/udev_097.bb): udev: add description for libvolume_id Aug 22 08:49:31 hrw|work: Yes, I hadn't thought of that problem. I guess we need an option to force the older more linear behaviour Aug 22 08:51:45 RP: number-of-tasks=1 should be fine? Aug 22 08:53:06 hrw|work: I suspect it will still jump from task to task a bit... Aug 22 08:54:05 hrw|work: Its all to do with the task weighting algorithm and I have some ideas which might improve that based on this comment... Aug 22 08:54:56 RP: does it really needs to run (or print) all the do_fetch stuff? Aug 22 08:55:01 RP: clean/rm_work/fetch can be anytime, unpack/patch/configure/compile take diskspace Aug 22 08:58:17 time to pluck some things but first meeting :( Aug 22 09:04:21 any ideas why I'd be getting this times a hundred? Aug 22 09:04:25 /usr/bin/jade:../rsvg-docs.sgml:14:16:E: element "BOOK" undefined Aug 22 09:04:37 package librsvg-2.6.5 Aug 22 09:04:50 1-2h wasted now Aug 22 09:05:41 mikearthur: what was the result/conclusion of the /usr/lib/libXau.so from yesterday? Aug 22 09:05:47 zecke: I don't know Aug 22 09:05:56 zecke: I just kept rebuilding stuff till it worked :D Aug 22 09:06:10 zecke: think it could have been *-native wierdness Aug 22 09:06:17 that is not good Aug 22 09:06:28 nah, it was my fault Aug 22 09:06:41 iirc, I was trying to use some system stuff rather than *-native stuff Aug 22 09:06:48 and went overboard Aug 22 09:07:44 zecke: any ideas about the sgml stuff? Aug 22 09:07:49 no Aug 22 09:10:16 :s Aug 22 09:11:48 does someone experiment glfbdev ( mesa ) on OE ? Aug 22 09:11:57 koen: Its safe to remove do_fetch[nostamp]=1 from base.bbclass which fixes that Aug 22 09:12:44 koen: It will force everyone to rebuild everything in tmp so I' Aug 22 09:12:50 ve not checked it into OE yet Aug 22 09:13:01 hmmm Aug 22 09:13:28 how does that handle SCM based recipes Aug 22 09:13:29 ? Aug 22 09:14:02 koen: Dates based ones have the date in PV Aug 22 09:14:11 ok Aug 22 09:17:52 zecke: fixed it by using 2.9.5 Aug 22 09:18:46 of what? Aug 22 09:19:28 zecke: the librsvg stuff Aug 22 09:19:59 03koen 07org.oe.dev * r447bcfa8... 10/ (1 classes/base.bbclass): Aug 22 09:19:59 base.bbclass: remove nostamp for fetch Aug 22 09:19:59 * this will force a full rebuild, so have some coffee + cookies ready Aug 22 09:20:16 * koen gets some coffee and cookies Aug 22 09:22:37 koen: You're braver than me :) Aug 22 09:23:00 koen: Someone should write a script to cope all the do_unpack stamps to do_fetch ;-) Aug 22 09:24:22 http://www.openembedded.org/~zecke/BUILDS.tar.bz2 Aug 22 09:24:35 <- my build scripts that used to compile poky and oz354x Aug 22 09:26:14 hngr Aug 22 09:26:19 * koen fights monotone on OSX Aug 22 09:26:34 or rather my own stupidity Aug 22 09:35:02 load average: 11.74, 4.07, 1.65 Aug 22 09:35:11 ~hail multithreaded bitbake Aug 22 09:35:13 * ibot bows down to multithreaded bitbake and chants, "I'M NOT WORTHY!!" Aug 22 09:35:44 * koen should get access to the altix next to ewi546 Aug 22 09:36:59 RP: you should try to convince mickeyl to add a selfcontained webserver to bitbake that can show the status of the threads via some AJAXy interface Aug 22 09:37:50 * kergoth thinks itd be fun to just attach a bitbake -i's stdin & stdout to a unix domain socket, then write the client as a seperate app that talks to the bitbake shell Aug 22 09:38:07 * koen wonders where mickeyl is Aug 22 09:38:10 koen: I do want to talk to mickeyl about the UI options :) Aug 22 09:38:19 Frankfurt Aug 22 09:38:54 kergoth: It'd be easy to split as the multithreading code just calls tryBuild Aug 22 09:39:05 03koen 07org.oe.dev * r3fb5d51c... 10/ (1 packages/linux/LAB-kernel_cvs.bb): LAB-kernel cvs: add hhv version check Aug 22 09:39:11 There is no other function overlap Aug 22 09:39:56 zecke: I'll be interested to look at those scripts at some point btw, thanks - I've saved a copy :) Aug 22 09:40:25 RP: the OZ354x are more advanced as they have a SCM and Builder spec Aug 22 09:41:53 later guys Aug 22 09:42:40 load average: 104.52, 77.61, 36.93 Aug 22 09:42:51 * koen wonders how that happened with 3 threads Aug 22 09:42:56 koen: :) Aug 22 09:43:44 For development purposes I was running it with 5 and on a single processor and never saw that... Aug 22 09:44:02 I have a zillion tar instances Aug 22 09:45:06 I did wonder how long it'd be before a bitbake went mad and became a bitkiller... Aug 22 09:45:22 probably redoing all the fetches to get a stamp Aug 22 09:51:44 hey woglinde Aug 22 09:52:23 hi ! Aug 22 09:52:29 hallo koen Aug 22 09:52:32 hi ifaistos Aug 22 09:53:18 koen how was lowlands? Aug 22 09:53:19 RP: Zecke has done some work on the icecream class and i am setting up a cluster of 10 p4@2.5 Ghz to test it Aug 22 09:53:53 Ifaistos: I saw the photos :) Aug 22 09:55:09 RP: yeah . that was 4-5 years ago Aug 22 09:55:54 woglinde : hi ! Aug 22 09:56:35 The idea is to 'donate' the spare building capacity to the OE developers to test things Aug 22 09:57:33 Ifaistos: Sounds useful :) Aug 22 09:57:51 RP: Do you have any (ball park) benchmarks on the new multithread bitbake ? Aug 22 09:58:03 Ifaistos: What kind of benchmarks? Aug 22 09:58:58 RP: time to build :) Aug 22 09:59:32 Ifaistos: I've not tried it on a decent multi CPU box yet ;-) Aug 22 09:59:55 I have one though so I will have to do some benchmarking :) Aug 22 10:00:36 RP: I have a dual PIII@1Ghz is this considered decent ? Aug 22 10:01:45 Ifaistos: It'd at least give an idea of the speed improvement :) Aug 22 10:02:14 Ifaistos: I have a Dual Xeon 2,8Ghz I will test on Aug 22 10:03:08 woglinde: it was great Aug 22 10:03:18 koen :) Aug 22 10:04:43 RP: I just made OE IO-bound with that nostamp commit Aug 22 10:07:20 koen: certain tasks are always going to be IO bound - this is why we need better control of the types of threads and the existing algorithm needs improvement Aug 22 10:08:31 anyone here used the stock sharp firmware for the tosa? Aug 22 10:09:16 koen: e.g. a max of: 4 fetch, 2 patch, 1 unpack, 2 compile, 1 configure, 1 package and a max total of 4, a max total of 2 from the group (unpack, configure, compile, package) Aug 22 10:11:23 also, how do I get things, such as kexec-tools and bash to be included in my gpe-image? Aug 22 10:16:23 does anyone have a second to help give a BitBake n00b some direction? Aug 22 10:17:11 bigd0g: I could try, depends if you are more of a noob than me :D Aug 22 10:18:01 i'm working with the org.openembedded.dev tree and am trying to compile linux from a .bb file ... i believe my 'local.conf' file is setup correctly, but now i get an obscure error: Aug 22 10:18:26 ERROR: Nothing provides dependency linux-omap1_2.6.12-rc2.bb Aug 22 10:18:48 sounds like you're trying to pass a filename on the bitbake commandline rather than a provider. Aug 22 10:18:59 what exactly are you typing at your shell to initiate the build? Aug 22 10:19:11 bitbake linux-omap1_2.6.12-rc2.bb Aug 22 10:19:14 ^ that Aug 22 10:19:16 thats incorrect. Aug 22 10:19:22 perhaps you wanted bitbake linux-omap1 Aug 22 10:19:41 hrm. Aug 22 10:19:42 * bigd0g tries Aug 22 10:20:04 negative Aug 22 10:20:05 ERROR: Nothing provides dependency omap1 Aug 22 10:20:26 what? Aug 22 10:20:43 that's the error i get when i tried 'bitbake linux-omap1' Aug 22 10:20:51 oh... wait Aug 22 10:21:12 excellent. Aug 22 10:21:17 thanks, kergoth. :) Aug 22 10:21:48 bitbake -can- accept a specific filename, using the -b commandline option, but it isnt for the faint of heart, as it bypasses all dependencies. Aug 22 10:23:46 yeah, i was reading about the switches... but the sample executions that i had seen online warned against it. this is really my first time using bitbake. Aug 22 10:25:26 hrm. the build failed. must be the proxy i'm forced to sit behind. Aug 22 10:25:46 anyone know how I can get packages, such as kexec-tools and bash to be included in my gpe-image? Aug 22 10:26:53 re Aug 22 10:27:01 mikearthur: add them into IPKG_INSTALL of gpe-image Aug 22 10:28:51 hrw|work: no way of doing that via my local.conf? Aug 22 10:30:09 mikearthur: BOOTSTRAP_EXTRA_RDEPENDS? Aug 22 10:30:16 cool, thanks Aug 22 10:31:38 9.5M free, 3.5M swap free Aug 22 10:36:08 shit.. Aug 22 10:36:12 Revision: 5ea9eb6f097fe0b646889d1b3a1312d9d7fc796e Aug 22 10:36:20 ~lart me for Revision: 5ea9eb6f097fe0b646889d1b3a1312d9d7fc796e Aug 22 10:36:21 * ibot raises middle finger to hrw|work for Revision: 5ea9eb6f097fe0b646889d1b3a1312d9d7fc796e Aug 22 10:40:15 RP: unpack should invalidate the patch stamp Aug 22 10:45:18 RP: ingore that, just some minor quilt breakage Aug 22 10:46:19 03hrw 07org.oe.dev * ra1d8cff4... 10/ (3 files in 3 dirs): base-files: (un)load ohci-hcd on tosa to get usb working after suspend/resume Aug 22 10:46:29 03hrw 07org.oe.dev * rc72d9bc6... 10/ (1 packages/netbase/netbase_4.21.bb): Aug 22 10:46:29 netbase: bump PR because of change of tosa interfaces pushed in 5ea9eb6f097fe0b646889d1b3a1312d9d7fc796e Aug 22 10:46:29 - disable usage of usbctl on tosa/2.6 until we get 2.6 version Aug 22 10:48:49 next thing for XorA Aug 22 10:49:02 NOTE: Handling BitBake files: / (1573/3948) [39 %]NOTE: package glibc-2.4-r10: task do_compile: completed Aug 22 10:49:05 NOTE: Handling BitBake files: \ (1581/3948) [40 %]NOTE: package glibc-2.4: completed Aug 22 10:49:16 those lines for RP ofcourse Aug 22 10:49:39 RP: I broke build when it was in glibc-2.4/do_compile()... Aug 22 10:50:02 XorA: pushed fix to zaurus-updater Aug 22 10:50:33 hrw|work: It looks like it failed to wait for all tasks to exit :-/ Aug 22 10:50:46 RP: could you have a look at http://rafb.net/paste/results/X342Dj90.html ? Aug 22 10:51:44 03pH5 07org.oe.dev * r45bac0f6... 10/ (1 packages/dbus/dbus_0.92.bb): dbus: ad 0.92, last commit was missing the actual recipe Aug 22 10:51:50 03hrw 07org.oe.dev * rbb05117c... 10/ (1 packages/zaurus-updater/zaurus-updater.bb): Aug 22 10:51:50 zaurus-updater: move do_deploy so it does not conflict with rm_work Aug 22 10:51:50 applied changes from dc69aab02eb323fa6a030fa05f7b63be3642fb52 Aug 22 10:51:50 through 41767e9744f20a6d01a4cede1ef96c50c85d78f6 Aug 22 10:52:47 RP: and I broke to get fetch/nostamp change in Aug 22 10:55:37 03koen 07org.oe.dev * rca20876b... 10/ (1 conf/distro/angstrom-2006.9.conf): angstrom-2006.9: prefer dbus 0.92 Aug 22 10:56:59 03rpurdie * r578 10bitbake/lib/bb/runqueue.py: runqueue.py: Make sure depends is initialised Aug 22 10:57:15 koen: fixed, hopefully ;-) Aug 22 10:58:08 svnup... Aug 22 10:58:28 13:10 hrw@work:mt$ time nice bitbake -cfetch bootstrap-image Aug 22 10:58:50 lets check will it work... Aug 22 11:03:11 RP: how to force bitbake to fetch everything needed for ANY-image now? bitbake -cfetch ANY-image does not work. Aug 22 11:04:18 hrw|work: We need a new option for this... Aug 22 11:04:57 RP: bitbake -f -cfetch ANY-image? Aug 22 11:04:59 hrw|work: The code is right - bootstrap-image's do-fetch task has no dependences on the any other fetch tasks. We just need a way to make it do what you want Aug 22 11:05:42 ok Aug 22 11:06:58 started bootstrap-image build.. Aug 22 11:07:09 NOTE: Machine ID assigned by tinderbox: 104 Aug 22 11:10:37 heh.. need to clean tmp again ;( Aug 22 11:22:06 woglinde: ping Aug 22 11:26:47 I want to run a webrowser on my embedded unit for configuration perposes. The LCD I have is not very large so I would want a web browser that could run in fullscreen mode. Does anyone have a recommendation for such a browser that is already supported by OE? Aug 22 11:31:17 firefox? Aug 22 11:35:16 Gerrath: that's more a windowmanager issue Aug 22 11:35:44 hrw|work, I thought firefox was pretty large, is there a more light weight one? Aug 22 11:35:48 links -g Aug 22 11:35:57 will run on framebuffer or X Aug 22 11:36:24 Gerrath: either konqueror embedded or gpe-mini-browser Aug 22 11:36:39 koen: what would be the lightest weight window manager that would allow me to run a web browser? Aug 22 11:36:52 Gerrath: matchbox? Aug 22 11:36:57 zecke, ok, I will look into konqueror embedded or gpe-mini-browser. Aug 22 11:36:58 wm2? Aug 22 11:37:27 Gerrath: gpe-mini-browser used Gtk/X11 where konqueror-embedded uses Qt(X11 or Framebuffer) Aug 22 11:37:37 Gerrath: both use KHTML/WebKit Aug 22 11:38:42 RP: bitbake do only one task here now - binutils-cross/do_compile Aug 22 11:39:08 zecke, ok, thanks.. I was even thinking of just installing gpe or opie but I think it is pretty over kill for my app however it might just be the easiest thing to do. My app currently will only support to modes. One is a webbrowser for configuration and the second is still images for messages. The webbrowser mode must support a touch screen. Aug 22 11:39:09 RP: http://ewi546.ewi.utwente.nl/tinderbox/showlog.pl?machine_id=104&logfile=20060822133243.log Aug 22 11:39:54 that poor tinderclient Aug 22 11:40:15 hrw|work: tinderclient isn't really compatible with the new bitbake :-/ Aug 22 11:40:44 zecke: kergoth pointed out the pkg start and pkg completed events are totally broken :-( Aug 22 11:41:07 RP: ;) Aug 22 11:41:18 turned it off Aug 22 11:41:40 RP: in what regard? Aug 22 11:41:56 boken in what regard Aug 22 11:42:11 zecke: we fire pkg completed events after each task completes in the new bitbake... Aug 22 11:42:35 ah, I thought tinderclient was broken Aug 22 11:43:00 zecke: no, the new code is in that regard. I'm not sure what it should do though... Aug 22 11:43:02 I currently wonder where I want to do the linearisation of the build log Aug 22 11:43:31 but I will improve other parts of the box first Aug 22 11:44:37 zecke: I don't know enough about tinderclient to know how best to handle it... Aug 22 11:48:01 zecke, wow Konqueror Embedded is nice, you can run it with out a window manager! it seems it is all I need, I can use it for the web configuration and also to display system images. Aug 22 11:48:22 Gerrath: it should work and without X11 Aug 22 11:49:44 zecke, I guess the touchscreen would be the hard to implement since there is no windows manager running, I guess I may still want a WM. Aug 22 11:50:54 Hi! Aug 22 11:51:31 Gerrath: You don't need a wm for a working touchscreen? Aug 22 11:52:02 koen, hrw|work: Guys, can you please process #1266, #1351 (h4000 patches)? http://bugs.openembedded.org/show_bug.cgi?id=1351 Aug 22 11:52:51 h4000 is h415x? Aug 22 11:53:26 hrw|work: yes, 4150 & 4350 ;-) Aug 22 11:53:34 RP, how would the TS interface get tied back to Konqueror-embedded with out a WM, is it somehow handled by the libqt-embedded? Aug 22 11:54:39 psokolovsky: pushing 1351 Aug 22 11:54:58 Gerrath: qte handle tslib Aug 22 11:55:11 hrw|work: thanks, 1266 is machine conf, so needed too ;-) Aug 22 11:56:38 ok Aug 22 11:57:16 koen: your zombie python processes suck cpu on ewi Aug 22 11:57:38 hrw|work, would that be a lot of effort to incorporate into konqueror-embedded using fb instead of a WM? Aug 22 11:58:04 Gerrath: konq-emb with qt/e? you run it: 'konqueror -qws' and it will became own wm Aug 22 11:58:21 hrw|work, thanks, that is great. Aug 22 11:58:46 hrw|work, sounds like exactly what I need. Aug 22 12:01:35 hrw|work: that's the new and improved bitbake Aug 22 12:01:49 koen: ah. Aug 22 12:02:11 koen: Z state made me thinking about zombies Aug 22 12:02:34 no, Z state means that it's building for a zaurus ;) Aug 22 12:02:50 ;D Aug 22 12:03:14 wonw, mtn 0.29 is out. anybody tried yet? Aug 22 12:05:05 psokolovsky: I'm using it on ewi546 Aug 22 12:05:21 I'll wait until njs^Hmaintainer will add it to debian Aug 22 12:05:23 psokolovsky: it's just a small time-based release Aug 22 12:05:41 ok Aug 22 12:05:58 (which fixes the hang-on ctrl-c bug) Aug 22 12:06:18 14:20:53 up 56 days, 3:26, 18 users, load average: 148.78, 136.11, 75.07 Aug 22 12:06:22 awesome.. Aug 22 12:08:41 koen: how much tasks it has? Aug 22 12:08:50 2 Aug 22 12:08:56 koen: seems to fix mtn log as well, as you promised ;-) Aug 22 12:09:35 0.30 should speed up annotate as well and contain workspace-merge Aug 22 12:10:21 cool, pity it's another month to wait ;-) Aug 22 12:13:51 RP, kergoth: for bitbaked: http://webpy.org/ Aug 22 12:15:21 * koen restarts apacke Aug 22 12:15:26 apache* Aug 22 12:15:35 http://lists.openembedded.org/cgi-bin/mailman/listinfo/openembedded-devel is working now :) Aug 22 12:17:50 Having issues with liblxutil, that mkg3states patch isnt working for me Aug 22 12:19:30 koen: I don't mind that Aug 22 12:19:48 zecke: don't mind what? Aug 22 12:20:05 ah Aug 22 12:20:08 * koen slow Aug 22 12:21:11 my bitbake build is failing because cvs can't access the network due to the corporate proxy i'm behind. my http_proxy variables are set correctly in the shell and i haven't been able to find a good resource detailing how to force cvs through an http or socks4 proxy. anyone have any ideas/resources? Aug 22 12:22:43 bigd0g: no Aug 22 12:22:55 bigd0g: socksify or the like should make most stuff tunnel socks Aug 22 12:23:08 or dante-client Aug 22 12:23:14 koen: who is Kerkhoff? Aug 22 12:23:23 hrw|work: ? Aug 22 12:23:46 I don't know any Kerkhoff Aug 22 12:23:59 koen: you have CC'ed him Aug 22 12:24:00 only a girl I went to highschool with Aug 22 12:24:03 koen: "Kerkhoff,H.G. \(EWI\)" Aug 22 12:24:32 socksify? have a link, XorA? Aug 22 12:24:34 meh Aug 22 12:24:51 03pmiscml 07org.oe.dev * r43c44091... 10/ (1 packages/linux/handhelds-pxa-2.6_cvs.bb): handhelds-pxa-2.6: add h4000 to supported list -close #1351 Aug 22 12:24:52 hrw|work: thunderbird autocompleted kergoth to the wrong person Aug 22 12:24:56 03pmiscml 07org.oe.dev * r06621039... 10/ (1 conf/distro/angstrom-2006.9.conf): angstrom.conf: added h4000 support Aug 22 12:25:00 using the uni LDAP server Aug 22 12:25:01 03pmiscml 07org.oe.dev * r744ab7b1... 10/ (3 files in 3 dirs): tslib: added h4000 support - close #1351 Aug 22 12:25:05 bigd0g: part of the socks packages on most distros I think Aug 22 12:25:07 03pmiscml 07org.oe.dev * rbd9490f3... 10/ (1 packages/matchbox-panel/matchbox-panel.inc): matchbox-panel: added h4000 support - close #1351 Aug 22 12:25:25 I cant close 1351 - oe@hh is unknown for bugzilla... Aug 22 12:25:47 later again Aug 22 12:26:29 hrw|work: refresh Aug 22 12:26:40 XorA: ok, i'll see whati can find out for socksify then. thanks. Aug 22 12:28:16 psokolovsky: pushed 1266 and 1351 Aug 22 12:28:33 hrw|work: Thanks! Aug 22 12:29:56 03kristoffer 07org.oe.dev * r336c2756... 10/ (1 conf/distro/jlime-donkey.conf): Aug 22 12:29:56 conf/distro/jlime-donkey.conf: Major cleanups and safe versions added Aug 22 12:29:56 * Updated safe versions for GPE,GTK,... Aug 22 12:29:56 * Cleaned up some typos, made generally cleaner Aug 22 12:30:01 03kristoffer 07org.oe.dev * r92c2882d... 10/ (1 conf/distro/jlime-donkey.conf): conf/distro/jlime-donkey.conf: Added specific liblbxutil version Aug 22 12:30:06 03kristoffer 07org.oe.dev * r7977b8ce... 10/ (3 files in 3 dirs): Aug 22 12:30:06 xorg-lib/liblbxutil : Addition of 1.0.0 patch Aug 22 12:30:06 * Addition of mkg3states-1.0.0 patch that works with Aug 22 12:30:06 liblbxutil_X11R7.0-1.0.0 (correct directory layout for patch). Aug 22 12:30:11 03pmiscml 07org.oe.dev * r476ae042... 10/ (1 conf/machine/h4000.conf): h4000.conf: changes from #1266 Aug 22 12:32:14 bbl Aug 22 12:32:45 XorA et al: since cvs is being called by bitbake, where would i modify the calls to cvs to use the http proxy i'm behind? Aug 22 12:33:21 bigd0g: does cvs not just use the http_proxy envvars? Aug 22 12:33:42 it doesn't appear to... it can't connect to the repo and fails the bitbake build Aug 22 12:33:54 http_proxy, https_proxy, etc ... are all set as envvars Aug 22 12:34:11 just get someone to send you the cvs drop file Aug 22 12:35:01 if i'm doing a build, isn't it going to want multiple pulls from the repo? Aug 22 12:38:20 hrm. maybe proxychains could resolve the issue? hrm. Aug 22 12:44:27 who here has a tosa? Aug 22 12:44:29 morning Aug 22 12:45:45 hrw|work: http://bugs.openembedded.org/show_bug.cgi?id=1304 Aug 22 12:52:00 ops Aug 22 12:58:57 RP: I have to scroll up too much to see what recipe failed Aug 22 12:59:26 koen: I know. Its crying out for a UI... Aug 22 12:59:49 ~summon mickeyl Aug 22 12:59:58 apt takes out 20 clean, identical-looking phones, some extra hands, and pretends to be a telemarketer for a large corporation, so he gets delivered a phonelist containing mickeyl's coordinates Aug 22 13:01:45 RP: I get the feeling bitbake doesn't honor the max threads and tries to paralize it all Aug 22 13:02:18 RP: can I output the maxnumber of threads bitbake thinks it has somewhere? Aug 22 13:02:45 koen: lib/bb/runqueue.py - look at active_builds Aug 22 13:07:32 RP: I'm affraid I found a bug: http://rafb.net/paste/results/oUleQA18.html Aug 22 13:08:28 koen: ouch :-/ Aug 22 13:08:50 koen: What does it say number_tasks is? Aug 22 13:09:05 testing that right now Aug 22 13:09:31 03hrw 07org.oe.dev * r55b7a5bb... 10/ (3 files in 3 dirs): keymaps: add h2200 keymap - finally close #1304 Aug 22 13:11:44 NOTE: Active builds increased to 7NOTE: Max active builds: 2 Aug 22 13:12:19 I can't seem to manage to add new packages to a gpe-image. I've tried adding GPE_EXTRA_RDEPENDS += "kexec-tools bash altboot" to my local conf and the gpe-image.bb and rebuilding gpe-image, neither seems to work Aug 22 13:12:20 bugzilla: The product Zaurus Kernel does not exist or you don't have access to it. The following is a list of the products you can choose from. Aug 22 13:12:31 mikearthur: edit gpe-image.bb Aug 22 13:12:49 hrw|work: I did. Do I need to do a clean of gpe-image too? Aug 22 13:13:00 rebuild it Aug 22 13:13:42 just bitbake gpe-image or bitbake gpe-image -c clean; bitbake gpe-image? Aug 22 13:14:19 koen: You'd agree that doesn't make sense given the code? Aug 22 13:15:31 RP: there's no 'else' in the check for active>number Aug 22 13:16:20 koen: As it should fall through to the waitpid? Aug 22 13:19:35 koen: In that paste of yours, the count decreases at one point with no decrease message which is strange... Aug 22 13:20:37 I noticed that as well Aug 22 13:21:05 koen: The exception handler is the only other code that touches that Aug 22 13:23:59 koen: You could also add some oe.getpid()s to make sure we're getting the messages from one process... Aug 22 13:24:13 os.getpid() Aug 22 13:24:13 RP: I still don't get the if statement, but that might be my lack of python voodoo Aug 22 13:24:47 koen: You mean "if active_builds < number_tasks:"? Aug 22 13:25:00 yes Aug 22 13:25:33 Its in a while 1: loop. The top of it spawns another task, the bottom waits for tasks to finish Aug 22 13:26:04 "continue" moves to the top of the loop, if we haven't enough builds running Aug 22 13:26:05 ah, I though continue was just a fancy way of saying NOP Aug 22 13:26:07 hrw|work: did a rebuild, still not there :( Aug 22 13:26:47 koen: continue means skip the rest of this while loop and start the next one Aug 22 13:27:20 afk Aug 22 13:27:33 hrw|work: np Aug 22 13:27:44 anyone else know who I would add packages to my gpe-image? Aug 22 13:28:49 mikearthur: You have the right approach - perhaps something is overwriting the variable you're adding them to instead of appending? Aug 22 13:29:02 do13: ping Aug 22 13:32:39 RP: clean *native still bombs out Aug 22 13:32:52 depdata = taskData.build_targets[depid][0]IndexError: list index out of range Aug 22 13:33:46 hi all Aug 22 13:33:51 openzaurus.org down? Aug 22 13:34:00 or just deadly slow? Aug 22 13:34:43 Kristoffer: slow Aug 22 13:34:56 ERROR: Error Method already seen: get_glibc_fpu_setting in' glibc_cvs.bb' now in 'glibc_2.4.bb' Aug 22 13:35:13 RP: such things will need cleaning metadata or changing bitbake - right? Aug 22 13:35:59 hrw|work: or comment out the lines in question ;-) Aug 22 13:37:10 hrw|work: http://www.rpsys.net/openzaurus/temp/bitbake_hacks-r0.patch Aug 22 13:37:11 could someone give openzaurus.org some speed :D Aug 22 13:37:34 if I clean a package, should it get build if its depended on by another package? Aug 22 13:38:04 mikearthur: how you mean? Aug 22 13:38:38 If I do bitbake bash -c clean and, say, gpe-image depends on bash, should it build bash first on bitbaking gpe-image? Aug 22 13:38:53 yes Aug 22 13:39:31 mikearthur: clean doesn't touch the .ipk (yet) Aug 22 13:39:57 koen: package index clears that up though right Aug 22 13:40:13 koen: so it wouldn't? Aug 22 13:40:31 koen: I'm finding it very hard to actually get gpe-image including bash/kexec-tools/altboot Aug 22 13:40:34 mikearthur: but yes, it should build it Aug 22 13:40:42 koen: doesn't seem to.... Aug 22 13:41:02 mikearthur: does gpe-image depend on bash?? Aug 22 13:41:12 it should do, I added it Aug 22 13:41:25 too bootstrap depend or into the image/ Aug 22 13:41:37 RP: http://ewi546.ewi.utwente.nl/tmp/threads.log Aug 22 13:42:07 mikearthur: just make sure you either rebuild the task that contains your extra stuff, or bump its PR Aug 22 13:42:36 otherwise OE won't rebuild the ipk containing the (changed) task and hence do_rootfs will not include your stuff Aug 22 13:42:39 koen: can I put GPE_EXTRA_DEPENDS in my local.conf Aug 22 13:43:11 mikearthur: I suspect you want GPE_EXTRA_INSTALL Aug 22 13:43:33 RP: can that go in my local.conf? Aug 22 13:43:35 RP: does OPIE_EXTRA_INSTALL exist? Aug 22 13:43:46 Kristoffer: grep is your friend Aug 22 13:43:53 koen: :D Aug 22 13:44:15 mikearthur: I don't know. It should be able to but its possible a machine.conf file does GPE_EXTRA_INSTALL= and breaks things Aug 22 13:44:15 koen: :D Aug 22 13:44:30 RP: that's using this patch: http://rafb.net/paste/results/rmF4mc22.html Aug 22 13:44:38 mikearthur: Find any references to GPE_EXTRA_INSTALL = and change them to GPE_EXTRA_INSTALL += Aug 22 13:45:07 koen: Its running several zillion copies of bitbake? Aug 22 13:45:09 RP: where do you want me to drop in the getpid? Aug 22 13:46:04 koen: You should only ever see "OE Build Configuration:" once Aug 22 13:47:25 RP: that's probably the redirect screwing up, I only get that once in the actual outpu to console Aug 22 13:48:57 ~lart gimp Aug 22 13:48:58 * ibot takes large quantities of Krispy Kream donuts and stuffs them one after another down gimp's throat until gimp puts on 150lbs Aug 22 13:49:01 koen: ever thought of setting up the /package dir to something similiar to the portage dir? like sys-apps/ opie/ gpe/ xorg-server/ ...and so on Aug 22 13:49:16 Kristoffer: nope Aug 22 13:49:29 Kristoffer: tbh, I would dream of oe moving to portage :p Aug 22 13:49:34 Kristoffer: g3nt00 is for ricers Aug 22 13:49:39 :D Aug 22 13:49:47 Kristoffer: as it was originally a fork, iirc Aug 22 13:49:49 just asking, its alot easier to find stuff that way Aug 22 13:50:07 Kristoffer: actually it isn't Aug 22 13:50:11 but I guess grep is my friend :P Aug 22 13:50:17 is just enlarges the subdirs you have to look in Aug 22 13:50:34 yeah, I guess your right :) just gentoo damaged Aug 22 13:50:50 A bag of bananas for the first one that fixes openzaurus.org lag Aug 22 13:51:01 that involving fixing sourceforge Aug 22 13:51:06 involves* Aug 22 13:51:14 oki, 2 bags of bananes to whoever fixes sourceforge :D Aug 22 13:54:48 03rpurdie * r579 10bitbake/lib/bb/runqueue.py: runqueue.py: Make sure number_tasks is a number Aug 22 13:54:57 :) Aug 22 13:54:57 koen: fixed it ;-) Aug 22 13:56:06 RP: yay! Aug 22 14:00:15 hrw|work: http://wiki.openzaurus.org/ReleaseNotes/Oz3542 is ambiguous Aug 22 14:00:19 hrm. is there a listing of valid handheld.org repos somewhere? Aug 22 14:00:24 koen: fix it then Aug 22 14:00:31 hrw|work: all models: SD/MMC cards larger then 1GB are working Aug 22 14:00:40 ah Aug 22 14:00:50 or... better yet. can someone confirm cvs://anoncvs:anoncvs@anoncvs.handhelds.org/cvs;module=ipkg-utils is valid? Aug 22 14:01:00 hrw|work: ah, I can edit it Aug 22 14:01:09 does image_ipk work differently when DISTRO_TYPE = "release"? Aug 22 14:01:12 03rpurdie * r580 10bitbake/lib/bb/runqueue.py: runqueue.py: Fix the case where BB_NUMBER_THREADS is unset Aug 22 14:04:12 or, more to the point... what triggers remove_init_link() Aug 22 14:04:32 it doesn't seem to be being run in my image, leaving me with an unbootable image Aug 22 14:05:03 NOTE: Fetch cvs://anoncvs:anoncvs@anoncvs.handhelds.org/cvs;module=ipkg-utils Aug 22 14:05:04 cvs checkout: Updating ipkg-utils Aug 22 14:05:04 U ipkg-utils/.cvsignore Aug 22 14:05:04 U ipkg-utils/COPYING Aug 22 14:05:19 bigd0g: so yes, it is a valid uri Aug 22 14:05:37 tkp: That was something added for the nokia770 machine, You need to trigger that function to run on a per machine basis Aug 22 14:06:03 thanks, koen. i tried patching CVS to go through the proxy, but that doesn't appear to have worked. Aug 22 14:06:15 i'll have to try building from home :-/ Aug 22 14:06:20 tkp: ROOTFS_POSTPROCESS_COMMAND += " remove_init_link; " Aug 22 14:06:35 RP: right, ok Aug 22 14:11:48 tkp: nokia770 does not boot when /sbin/init is symlink - remove_init_link is for it Aug 22 14:13:40 hrw|work: have you considered tagging the repo for each RC to make it easier to view the differences? Aug 22 14:14:08 koen: not yet Aug 22 14:14:39 koen: rc2 works ok on poodle and tosa - need some love for collie ;( Aug 22 14:15:22 somtimes I could mistake the channel for #bestiality Aug 22 14:15:49 XorA: You're a regular there? Aug 22 14:16:05 * XorA declines to comment Aug 22 14:16:40 lol Aug 22 14:16:55 hrw|work: I have none of those :) Aug 22 14:17:02 hrw|work: I've made some headway with the touch nonsense on tosa Aug 22 14:18:01 hrw|work and RP: was it one of you who fixed the touch glitches for 2.6 on collie? Aug 22 14:18:10 hrw|work and RP: was it one of you who fixed the touch glitches for 2.6 on corgie, even Aug 22 14:18:16 koen: you also does not have collie iirc Aug 22 14:18:26 mikearthur: corgi == RP Aug 22 14:18:31 hrw|work: correct Aug 22 14:18:38 only a c700 Aug 22 14:18:38 mikearthur: I was the one who wrote the 2.6 c7x and cxx00 code Aug 22 14:18:58 RP: did you manage to completely eliminate the touch y-sync glitches or just reduce them and did the rest in userspace Aug 22 14:19:16 koen: then rc2 does not affect you Aug 22 14:19:22 koen: or even 3.5.4.2 ;D Aug 22 14:19:41 RP: can we give 2.6.17-current for clamshell users? Aug 22 14:19:48 mikearthur: Put it this way: with the standard userspace filters, it never showed a problem... Aug 22 14:20:12 hrw|work: I'm not sure if spectrum cards work but other than that, yes Aug 22 14:21:30 RP: patch to usb subsystem which will printk('Device xyz, abc need %d mA, but host can provide only %d mA.') would be nice to have Aug 22 14:21:51 hrw|work: It doesn't do that atm? Aug 22 14:22:00 thanks for the help guys. off to try the bitbake build from a residential connection to see if that prevents the build failure. Aug 22 14:22:59 CosmicPenguin: multhitheading works with bitbake trunk, so breakout your 16-way opterons :) Aug 22 14:23:09 ohhhhhh Aug 22 14:23:11 RP: 2.6.17 do that Aug 22 14:23:11 super sexy Aug 22 14:23:16 RP: ok, I'll need to play more with this then :s Aug 22 14:23:35 koen: I'll upgrade and take it for a test run Aug 22 14:23:46 koen: do I need to update the oe metadata too? Aug 22 14:24:13 CosmicPenguin: no idea Aug 22 14:24:29 CosmicPenguin: I just added "BB_NUMBER_THREADS=" to local.conf Aug 22 14:24:41 CosmicPenguin: you need to Aug 22 14:25:23 CosmicPenguin: There was a patch to .dev to add the correct dependency info - it will eaily apply to a branch Aug 22 14:27:04 RP: http://42.pl/u/kWY is patch - applied after 2616. Aug 22 14:27:25 RP: and 3541 users complain about it. Aug 22 14:27:54 RP: I'm considering 3.5.4.2 images for clamshells Aug 22 14:28:16 hrw|work: I'm happy enough to apply that to 2.6.16 kernels Aug 22 14:28:43 RP: I too - but how many people will update kernel just because they are told to do it... Aug 22 14:29:08 hrw|work: The spectrum_cs issue is the only bug I can remember for 2.6.17 Aug 22 14:29:34 ok Aug 22 14:37:38 * XorA has run 2.6.17 without trouble for ages now Aug 22 14:37:48 though I dont have a spectrum_cs Aug 22 14:39:53 morning all Aug 22 14:41:03 hey lrg Aug 22 14:41:17 hey lrg Aug 22 14:43:10 hi Liam Aug 22 14:44:38 RP, XorA: I may have a fix for corgi static. Something obvious that I missed. Aug 22 14:44:50 lrg: What's that? Aug 22 14:45:34 RP: there is a codec active bit. It wasn't being de-activated when the stream stopped Aug 22 14:46:07 lrg: ah. That sounds promising :) Aug 22 14:46:21 RP: hopefully :) Aug 22 14:54:14 psokolovsky: iirc there is code in .oz354x as well to cope with mice Aug 22 14:54:24 psokolovsky: try asking CoreDump|home Aug 22 14:54:48 koen: ok, will have a look Aug 22 14:55:04 it should be in .dev as well, but I haven't checked Aug 22 15:06:00 lrg: cool, I didnt get time yet to test your last patch so I new one I could test tomorrow Aug 22 15:16:11 RP: you tweaked ipkg-utils some time ago to get it faster. but how to preserve Packages.filelist without regenerating Packages from scratch... Aug 22 15:16:33 hrw|work: That's what the tweak did Aug 22 15:17:30 "ipkg-make-index -p Packages -v -l Packages.filelist ." == empty Packages.filelist Aug 22 15:18:05 Ah, Packages,filelist. sorry. hmm :-/ Aug 22 15:18:22 03rpurdie 07org.oe.oz354x * r7e952968... 10/ (1 packages/linux/linux-openzaurus_2.6.16.bb): linux-oz-2.6: Sync wth .dev, fixing what looks like a merge error Aug 22 15:18:43 RP: thx for fixing Aug 22 15:20:10 hrw|work: np Aug 22 15:20:19 hrw|work: What uses Packages.filelist? Aug 22 15:20:43 RP: I do sometimes Aug 22 15:22:08 hrw|work: ipkg-utils will probably need more logic adding to handle adding/removing packages from that file Aug 22 15:22:28 RP: pacakage searchtools like http://www.angstrom-distribution.org/repo/repo.php Aug 22 15:23:19 I'm trying to do a first release of my distro Aug 22 15:23:50 I have altered the DISTRO_VERSION in DISTRO.conf, but now loads of things seem to be missing from the rootfs Aug 22 15:23:57 koen: right. I guess for now, these tools will have to regenerate Packages from scratch until someone fixes ipkg-utils :-/ Aug 22 15:24:00 RP: after I add the logic to parse the filelist :) Aug 22 15:24:19 are there some docs details how to do a release? Aug 22 15:24:25 03hrw 07org.oe.oz354x * rf09d3ca4... 10/ (5 files in 3 dirs): ipkg-utils: Fix a bug in the md5sum field handling and add some extra checks (md5sum, size) to ipkg-make-index when reusing data from the previous package file. (from poky) from .dev Aug 22 15:24:31 03hrw 07org.oe.oz354x * r2fe0885c... 10/ (1 classes/rootfs_ipk.bbclass packages/meta/package-index.bb): Aug 22 15:24:31 rootfs_ipk.bbclass, package-index: With the fixed ipkg-utils there is no need to remove the old Packages file when regenerating the Packages index. Aug 22 15:24:31 This gives a big speed improvement when regenerating images as it doesn't have to Aug 22 15:24:31 open every ipk for its metadata. (from poky) Aug 22 15:24:32 from .dev Aug 22 15:24:35 Do I need to rebuild everything? Aug 22 15:24:46 or some specific packages? Aug 22 15:24:47 tkp: 1. generate images 2. boot images 3. test images Aug 22 15:25:00 4. get drunk Aug 22 15:25:03 heh Aug 22 15:25:23 5. add the url to the releasenotes + install instructions to a hotkey Aug 22 15:25:23 tkp: if images work then 4. tag metadata 5. build feed 6. get drunk for few days before feed will build Aug 22 15:25:25 yeah...but since changing the version number in DISTRO.conf, things are missing from the image Aug 22 15:25:41 tkp: then you have to fix configs Aug 22 15:26:11 right... I was hoping for some pointers as to which ones! Aug 22 15:26:48 litterally all I did was change the version number Aug 22 15:28:33 tkp: distro ones Aug 22 15:29:23 I set: Aug 22 15:29:23 DISTRO_TYPE = "release" Aug 22 15:29:23 DISTRO_VERSION = '0.2.1' Aug 22 15:29:27 in my distro.conf Aug 22 15:29:36 (from) #DISTRO_TYPE = "debug" Aug 22 15:29:36 #DISTRO_VERSION = ".dev-snapshot-${SRCDATE}" Aug 22 15:30:04 I don't have anything else that refers to the version number Aug 22 15:32:52 what is actually missing from the image? Aug 22 15:33:08 most things Aug 22 15:33:15 busybox Aug 22 15:33:20 * koen polishes the crystalball Aug 22 15:33:30 almost every package Aug 22 15:33:47 it doesn't even have /bin Aug 22 15:33:52 tried a clean rebuild? Aug 22 15:34:21 I cleaned task-stv, stv-image and stv-version and removed all the associated ipk files Aug 22 15:36:33 stv-image now does do_fetch in a split second... from the log: Aug 22 15:36:36 NOTE: package stv-image-1.0-r0: task do_fetch: started Aug 22 15:36:37 NOTE: package stv-image-1.0-r0: task do_fetch: completed Aug 22 15:36:50 it's fetching no packages Aug 22 15:38:36 that probably means it's already present Aug 22 15:39:55 ah... I see something: Aug 22 15:40:01 several messages like: Aug 22 15:40:02 Package stv-base-depends md5sum mismatch. Either the ipkg or the package index are corrupt. Try 'ipkg update'. Aug 22 15:40:39 x29a: ping Aug 22 15:40:43 one thing I added recently to local.conf was: DEPLOY_KEEP_PACKAGES = 1 Aug 22 15:40:47 could that be related? Aug 22 15:41:45 cu Aug 22 15:42:30 tkp: remove deploy/Packages* and try again Aug 22 15:42:37 if that doesn't work, remove tmp Aug 22 15:43:32 I really wanted to aviod hasving to rebuild my toolchain to do a release Aug 22 15:43:37 having* Aug 22 15:44:03 you should bump PR of the tasks when you change something that affects the outcome Aug 22 15:44:33 I have not changed anything in tasks... ony the version number in DISTRO.conf Aug 22 15:46:54 well I bumber the PR in the task just for the sake of it, but I still get messages like "ackage stv-base-depends md5sum mismatch" Aug 22 15:47:08 03florian 07org.oe.oz354x * rba209c9f... 10/ (4 files in 3 dirs): gpe-screenshot: Add patch to fix segfault uploading files in current release. Aug 22 15:47:15 03florian 07org.oe.dev * r7bf39cf9... 10/ (4 files in 3 dirs): gpe-screenshot: Add fix for segfault uploading files in current release. Aug 22 15:49:25 stv-image still tries to "Downloading file:/home/tom/stv/oe-stv/build/tmp/deploy/ipk/stv-x11-base-depends_1.0-r1_i586.ipk" even tho it should now be r2 Aug 22 15:49:51 is r2 present in the Package file? Aug 22 15:50:30 nope Aug 22 15:50:51 so I do I get it to rebuild that file? Aug 22 15:50:56 did you do a 'bitbake package-index' after removing that files? Aug 22 15:51:17 no Aug 22 15:51:22 I have never run that command Aug 22 15:51:33 that rebuilds the Package file? Aug 22 15:51:38 yes Aug 22 15:51:47 (like I said a few lines up) Aug 22 15:54:16 seems to have sorted it Aug 22 15:59:18 thanks koen Aug 22 16:08:35 hey hvontres|poodle Aug 22 16:40:06 03hrw 07org.oe.oz354x * rd8cbfd9b... 10/ (6 files in 4 dirs): zudoku: added 1.1 from .dev Aug 22 16:49:18 03hrw 07org.oe.oz354x * r6f64da16... 10/ (4 files in 2 dirs): collie.conf:split to 2.4/2.6 configs Aug 22 17:07:58 http://www.rpsys.net/openzaurus/temp/make_fetch_stamps will add missing fetch stamps to a stamps directory Aug 22 17:20:02 re Aug 22 17:20:10 hrw: wb Aug 22 17:20:35 RP: mc tmp/stamps/ then select *_unpack*, press F6, from, to, enter Aug 22 17:21:02 collie/3.5.4.2-rc2-opie = 13.7M :( Aug 22 17:22:17 hrw: That doesn't make the date slightly earlier and bitbake does a < compare, not a <= Aug 22 17:22:30 hrw: Yes, there are alternatives :) Aug 22 17:22:52 hrw: I have a fix for the spectrum_cs problem btw Aug 22 17:22:59 cool Aug 22 17:23:54 RP: so 3.5.4.2-rc2 can get it? (not yet released) Aug 22 17:25:51 hrw: I think so, just the question of which asoc release to use Aug 22 17:26:03 (which is why I've not checked anything into .dev yet) Aug 22 17:26:10 RP: the one which work on tosa/poodle? Aug 22 17:26:18 the one without static Aug 22 17:26:20 ? Aug 22 17:28:13 we don't have one without static yet but Liam should be providing one soon... :) Aug 22 17:28:20 fingers crossed :) Aug 22 17:28:35 is there an equivalent dcop system for konqueror-embedded? I need to control it from the arm system. Aug 22 17:28:52 I want to do a slide show and be able to change the pictures via the system. Aug 22 17:29:36 Gerrath: qcop? Aug 22 17:31:25 RP: how about the resume on removing/inserting headphone stuff in asoc? Aug 22 17:31:46 koen, can it just simply be installed with out a WM along side of konqueror and allow shell commands to control konqueror? a quick read of it looks like it can... Aug 22 17:32:10 no idea Aug 22 17:32:16 I don't use qt/e crap Aug 22 17:32:21 (expect on my moto) Aug 22 17:33:09 Koen, lol, I'm only using it because it seems like the best way to have a web browser with out having a full blown WM since konqueror can use a framebuffer direct. Aug 22 17:34:35 12887900 for minimal-opie-image/collie Aug 22 17:37:26 hrw: why did opie get so big? Aug 22 17:37:36 unstripped kernel modules? rtti? Aug 22 17:37:40 uhm, dillo does look like a nice option as well. Aug 22 17:38:50 koen: checking how much plain bootstrap takes Aug 22 17:40:14 . o O (working connect+ drivers) Aug 22 17:40:22 . o O (working SD) Aug 22 17:43:29 hvontres|poodle: I need to look at that Aug 22 17:44:51 koen: bootstrap is 4.3M, minimal-opie is 13M Aug 22 17:45:03 checking gpe now Aug 22 17:48:06 Gerrath: links -g is also good for framebuffer use Aug 22 17:48:58 * Necronom wonders where the tags for the OZ releases are. Aug 22 17:54:47 Necronom: in .oz354x Aug 22 17:54:58 koen: 13M for gpe Aug 22 17:57:03 hrw: you could try this: Aug 22 17:57:05 FULL_OPTIMIZATION = "-fexpensive-optimizations -fomit-frame-pointer -frename-registers -Os" Aug 22 17:57:05 BUILD_OPTIMIZATION = "-Os" Aug 22 17:57:22 koen: do not want to rebuild whole feed.. Aug 22 17:57:39 just the opie bits to see how much it differs Aug 22 17:58:58 and -fvisibility-inlines-hidden Aug 22 17:59:04 and QtE with visibility Aug 22 18:00:59 mr_nice_slacker, have you used links2 in fb mode? Can you control it remotely? Aug 22 18:02:04 any caveats for -fvisibility-inlines-hidden ? Aug 22 18:04:43 kergoth: have you worked with tan/tap? Aug 22 18:04:47 tun/tap? Aug 22 18:07:20 Gerrath: I only used links on x86 - i have no clue about if it is possible to use it remotely Aug 22 18:13:53 zecke: we need chroot Aug 22 18:15:41 * chouimat is away: Extensive session with Mr. Clean Aug 22 18:18:27 Ania use Zaurus again Aug 22 18:18:51 I show her c7x0 with Zudoku Aug 22 18:18:56 :) Aug 22 18:20:17 btw - end of google job Aug 22 18:23:18 end of? Aug 22 18:25:18 zecke: remember offer which I got from google? Aug 22 18:25:26 sysadmin Aug 22 18:25:30 or senior engineer Aug 22 18:25:34 yep Aug 22 18:25:41 and this offer is now void? Aug 22 18:26:01 yep Aug 22 18:26:07 sad? Aug 22 18:26:21 not too much Aug 22 18:28:30 but it pushed me to write kind of resume for my website Aug 22 18:33:30 hrw: then be happy :) Aug 22 18:44:31 flasing collie... Aug 22 18:47:53 aaargh Aug 22 18:47:57 ENOM&MS Aug 22 18:59:08 "If you can explain how you do something, then you're very very bad at it." Aug 22 19:08:23 hehe Aug 22 19:08:42 I don't agree, but I can't explain why Aug 22 19:09:11 * chouimat is back. Aug 22 19:16:35 zecke: biff Aug 22 19:39:44 collie minimal opie image fails Aug 22 19:40:10 argh Aug 22 19:40:37 looks like i need to do 3542-collie distro Aug 22 19:42:04 * mr_nice_slacker dreams of a really usable SIMpad Aug 22 19:42:17 altboot lack 2.6 config, keymaps need to be added.. Aug 22 19:49:07 SoC next year: teach OE use use a chroot for compiling Aug 22 19:49:34 koen: I want to do this before next SoC Aug 22 19:50:50 zecke: let's gather funds after OEDEM :) Aug 22 19:51:08 for that? Aug 22 19:51:23 koen: we need to find out if we need FUNDS at all Aug 22 19:51:40 koen: I have decided to get married to an old and rich lady... Aug 22 19:51:47 I don't know where to find them though Aug 22 19:52:02 zecke: hosting costs for example Aug 22 19:53:07 * koen has a look at http://www.openembedded.org/wiki/OEDEM Aug 22 19:54:38 * koen inserts easter egg Aug 22 19:54:57 the busy box compile ussing all available memory is annoying Aug 22 19:55:41 koen: /wiki/oedem == /oedem - right? Aug 22 19:56:07 YEAH, YEAH, YEAH Aug 22 19:56:24 hrw|tv: yes Aug 22 19:56:56 leffe? Aug 22 19:58:43 sssh! Aug 22 19:58:49 :) Aug 22 20:01:31 hey hey hey Aug 22 20:01:37 the bearded man has arrived Aug 22 20:01:45 * koen waves at mallum Aug 22 20:01:53 hey koen Aug 22 20:02:03 koen: my beard is coming back again, as I'm too lazy to shave Aug 22 20:02:15 zecke: tsk Aug 22 20:02:32 * koen shaved this morning Aug 22 20:02:47 a beard iches too much Aug 22 20:03:23 i have to it.. Aug 22 20:08:00 beard is good Aug 22 20:08:04 shave is bad Aug 22 20:08:20 mallum: http://www.openembedded.org/~zecke/Web-Site/Me.html Aug 22 20:09:22 zecke: http://www.flickr.com/photos/mallum/117649319/ Aug 22 20:10:02 lol Aug 22 20:10:04 still collie no boot.. Aug 22 20:10:52 * CosmicPenguin was hoping it was a baby picture Aug 22 20:11:53 re Aug 22 20:11:56 CosmicPenguin: not just yet Aug 22 20:12:21 CosmicPenguin: think were getting a bit bored waiting; http://www.flickr.com/photos/mallum/222135187/ ? Aug 22 20:12:38 ohhhhh.... cutre Aug 22 20:13:03 http://pastebin.ca/145065 <- how to make dvbsnoop a complete DVB-H stack Aug 22 20:13:15 plan9 bunny! Aug 22 20:13:29 man loves bunny Aug 22 20:13:32 RP: is RMK purposely ignoring that 2GB SD patch? Aug 22 20:15:09 i hate to have use erial cables.. Aug 22 20:15:15 mallum: http://www.flickr.com/photos/mallum/168437566/ <- nice shirt Aug 22 20:15:47 zecke: employee only shirts Aug 22 20:16:09 mallum: hello; sorry I'm catching you here; question about mb-panel - when I once do gtk_widget_hide on plug, which is connected to panel, it won't get shown anymore... . Aug 22 20:17:21 mmp: check matchbox-panel-2 Aug 22 20:18:14 shit.. collie/3.5.4.2-rc2 does not send kernel logs to serial.. Aug 22 20:18:15 mallum: :) Sorry, I did not expect newer version of mb-panel:); I'll check it out asap. Aug 22 20:18:17 mallum: http://www.flickr.com/photos/mallum/90287916/ luckily you don't look insane Aug 22 20:20:47 * koen also notices mb-wm2 Aug 22 20:21:39 wm2? l33t Aug 22 20:21:45 aare there screenshots? Aug 22 20:22:18 CosmicPenguin: build it, install it, make screenshots :) Aug 22 20:22:34 hah Aug 22 20:22:46 I would but there are too many children without laptops Aug 22 20:23:03 yup Aug 22 20:23:03 lol Aug 22 20:24:39 go ewi, go... Aug 22 20:24:55 * hrw|tv want debug kernel for collie but compiling takes time... Aug 22 20:25:16 * koen hits ctrl-z Aug 22 20:25:28 ;F Aug 22 20:25:53 RP: tried 2.6 on collie? it looks like it does not send anything to serial port;( Aug 22 20:26:58 hrw|tv: tried the printascii hack? Aug 22 20:27:10 koen: no Aug 22 20:27:23 looks like I will release 3.5.4.2-rc2 without collie Aug 22 20:27:39 CoreDump|home: where are you........ Aug 22 20:30:12 CosmicPenguin: wm2 dont do much useful yet Aug 22 20:30:50 * koen fg Aug 22 20:31:33 time for thumb trick to flash collie again Aug 22 20:32:19 RP: what do you think about putting the bb.note("increased active builds to %s" % active_builds) into bitbake? Aug 22 20:32:26 RP: I like it very much :) Aug 22 20:33:07 ~lart sharp for releasing collie Aug 22 20:33:08 * ibot puts sharp into a headlock and administers a mighty noogie, rubbing half of sharp's hair of for releasing collie Aug 22 20:33:21 ~lart sharp for poodle Aug 22 20:33:22 * ibot judo chops sharp for poodle Aug 22 20:33:26 ~lart sharp for corgi Aug 22 20:33:26 * ibot rm -rf's sharp for corgi Aug 22 20:33:35 ~lart sharp for zaurus line Aug 22 20:33:36 * ibot follow's sharp with a gauntlet and ... scratch ... HUMILIATION for zaurus line Aug 22 20:33:44 that was goo lart ibot Aug 22 20:33:51 s/goo/good Aug 22 20:34:41 time to bed Aug 22 20:34:42 cu Aug 22 20:45:18 03koen 07org.oe.dev * r9c24baa5... 10/ (1 packages/wpa-supplicant/wpa-supplicant_0.4.7.bb): wpa supplicant 0.4.7: change do_install to work with OE's install from coreutils-native Aug 22 21:02:12 tmbinc_: do you know any DVB-T receiver that can receive all pids (or supports -tsraw of dvbsnoop?) Aug 22 21:04:48 * chouimat is away: PUB Aug 22 21:15:05 nite Aug 22 21:32:42 CosmicPenguin: No idea. I can't predict RMK.... :) Aug 22 21:33:29 that patch has been out there forever Aug 22 21:33:34 he's being stubborn ^ 2 Aug 22 21:33:59 CosmicPenguin: That figures. There are several issues like this I need to address - I just lack the time :-/ Aug 22 21:36:55 RP: multithreading changed bitbake behaviour Aug 22 21:37:10 koen: Which change do you mean? Aug 22 21:37:14 RP: 'bitbake angstrom-gpe-images' now builds everything in task-angstrom.bb Aug 22 21:37:17 -s Aug 22 21:37:32 koen: It has to. blame debian.bbclass Aug 22 21:37:37 grrr Aug 22 21:37:51 ~lart debian.bbclass Aug 22 21:37:52 * ibot whacks debian.bbclass upside the head Aug 22 21:37:52 The pervious understanding was a bug :-( Aug 22 21:38:31 and yes, I dislike debian.bbclass ;-) Aug 22 21:51:41 koen: Any distro doesn't have to inherit debian.bbclass ;-) Aug 22 21:52:13 koen: We could also implement a cache and avoid the nasty depends implications of debian.bbclass of course Aug 22 21:52:13 yeah Aug 22 21:52:26 or banish shared libs Aug 22 21:52:59 That wouldn't help in this case ;-) Aug 22 21:53:14 heh Aug 22 21:53:15 banishing RDEPENDS would :) Aug 22 21:53:22 CC=arm-linux-gcc Aug 22 21:53:35 * koen declares drifnet to be buggy Aug 22 21:54:16 we should lart the person who added that Aug 22 21:54:18 ~lart me Aug 22 21:54:18 * ibot slaps koen around with a large trout Aug 22 21:54:36 03koen 07org.oe.dev * r5c6bd922... 10/ (1 packages/angstrom/task-angstrom.bb): task-angstrom: mask out broken packages Aug 23 00:52:22 * chouimat is back. **** ENDING LOGGING AT Wed Aug 23 02:59:57 2006