**** BEGIN LOGGING AT Fri Apr 06 03:00:00 2007 Apr 06 07:32:03 good morning Apr 06 07:34:48 psokolovsky: you're here? Apr 06 07:44:19 Hi! Apr 06 07:44:25 Bernardo: yep Apr 06 07:47:05 morning Apr 06 07:47:36 good morning hrw Apr 06 07:48:37 psokolovsky: I'm (still) having problems with opie-image in angstrom. I managed to also build console-image and x11-image with my env, and only opie-image gives me a "no init found" error Apr 06 07:49:01 Bernardo: hm, weird Apr 06 07:49:25 Bernardo: well, start with checking if you actually have init package in your image, or no Apr 06 07:49:50 most my previous errors went away when I found out I had dash as default again, but this one won't quit Apr 06 07:50:35 ok, what is the best way to check? mount the ext2 image as loopback? Apr 06 07:51:17 Bernardo: yes, if you build ext2 image Apr 06 07:52:54 untar the tar.gz image? Apr 06 07:54:02 koen: ok Apr 06 07:55:16 good morning all Apr 06 07:55:49 good morning koen Apr 06 07:56:03 ok, sbin only has ldconfig - that might be a problem Apr 06 07:59:29 now to find out why Apr 06 08:00:06 the log.image_cmd shows only sbin added in ldconfig... :( Apr 06 08:04:05 morning Apr 06 08:05:22 XorA: at work? Apr 06 08:05:35 hrw: yeah Apr 06 08:06:05 XorA: scotland does not have free day today? Apr 06 08:06:49 hrw: yeah, but I work as its a waste of holidays, just millions of annoying kids everywhere Apr 06 08:06:57 hrw: I shall take the day when kids are in schools Apr 06 08:07:25 lol Apr 06 08:07:42 I did that when I didn't have kids too... ;) Apr 06 08:07:52 Bernardo: what about sysvinit package? consider posting the whole log somewhere Apr 06 08:08:23 the image_cmd log? I can pastebin it in a couple of minutes Apr 06 08:08:58 Bernardo: yeah, cool. just make sure you post the latest in case you have few in the dir Apr 06 08:11:45 http://pastebin.ca/426605 Apr 06 08:15:13 hrw: check the next commit Apr 06 08:16:10 ok Apr 06 08:16:25 03koen 07org.oe.dev * r76a890c9... 10/ (1 packages/gcc/gcc-cross-sdk_4.1.2.bb): gcc-cross-sdk: add 4.1.2 with its own SRC_URI to remove zecke-xgcc-cpp and failfast patches Apr 06 08:16:37 nice Apr 06 08:16:49 koen: anyway it will fail on amd64 probably Apr 06 08:18:52 hrw: Still no luck with nunome and friends. Same error as always. Can you do me a favor and try for MACHINE=spitz or collie? Preferably from clean tmp. Apr 06 08:20:10 * Laibsch reboots. will brb. Apr 06 08:20:43 Laibsch: lets see Apr 06 08:20:53 !oebug 2066 Apr 06 08:20:54 * * Bug 2066, Status: ASSIGNED, Created: 2007-03-30 04:56 Apr 06 08:20:55 * * dp(AT)xora.org.uk: 2.6.20 pxa-serial-hack.patch prevents 8250 loading Apr 06 08:20:56 * * http://bugs.openembedded.org/show_bug.cgi?id=2066 Apr 06 08:21:23 Bernardo: err, that's not right log. I need bitbake's do_rootfs log Apr 06 08:22:15 03koen 07org.oe.dev * r85b3b5a3... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom: prefer gcc 4.1.2 for the sdk Apr 06 08:22:16 ah, ok Apr 06 08:23:34 10:23:25 (328.85 KB/s) - `feeds.db.bz2' saved [4474902/4474902] Apr 06 08:23:41 3Mbps at home ;) Apr 06 08:24:35 psokolovsky: it's too big for pastebin (300k). I'm putting it on my web site now Apr 06 08:24:42 ok Apr 06 08:26:33 http://www.computerworld.com/action/article.do?command=viewArticleBasic&taxonomyName=security&articleId=9015618&taxonomyId=17 Apr 06 08:26:58 wow that security researcher is worth his weight in gold, wow if you attach gdb to JTAG you can debug the CPU amazing discovery that Apr 06 08:27:10 psokolovsky: http://www.bandos.homelinux.com/log.do_rootfs.10733 Apr 06 08:27:31 * XorA hates fake security morons like that Apr 06 08:27:32 "By taking advantage of a standard integrated circuit testing interface, called JTAG (Joint Test Action Group) Jack was able to sneak a peek at the systems' processors and get a close-up look at how they worked." Apr 06 08:27:48 looks like a ignobel prize nominee Apr 06 08:27:52 Beg: 430 Apr 06 08:27:55 Beg: 403 Apr 06 08:28:14 wow if I have physical access to a PC I can disconnect the RAM and dump it on a different PC Apr 06 08:28:52 wow if I dismantle my xbox I can put in a chip that replaces the BIOS so I can run my own code Apr 06 08:28:54 * XorA sighs Apr 06 08:29:38 that guy has obviously never heard of scraping the top off CPU's to use SEMs to look at the register contents Apr 06 08:30:00 * koen gets back to gcc-cross-sdk and llvm-gcc4-cross Apr 06 08:30:09 Apr 06 08:31:01 XorA: like I said, a very good nominee for a ignobel Apr 06 08:31:07 Bernardo: yeah Apr 06 08:31:28 Bernardo: but I bet he will get a few $$$ doing a couple of lectures at "security" shows Apr 06 08:31:50 Bernardo: the ones where 2600 cant get in to heckle Apr 06 08:32:05 Bernardo: you're listening? 403 on that url Apr 06 08:32:35 ah, didn't understand, let me check Apr 06 08:32:56 03koen 07org.oe.dev * rcc52573a... 10/ (1 packages/gcc/gcc-cross-sdk_4.1.2.bb): gcc-cross-sdk: actually add the SRC_URI Apr 06 08:33:23 psokolovsky: check now, should work Apr 06 08:33:33 yeah, ok Apr 06 08:34:12 I'm sorry, I forgot to check perms after uploading the file Apr 06 08:36:02 03koen 07org.oe.dev * r6cafad3c... 10/ (5 files in 2 dirs): gcc 4.1.x: move fail-fast to gcc-cross since it's only needed there Apr 06 08:38:02 Bernardo: np. Tell me please again exact command you used to build image? Apr 06 08:40:16 bitbake opie-image Apr 06 08:40:43 I've also tried later to build opie-kdepim-image, but that would fail building kdepimpi Apr 06 08:40:46 Bernardo: ok, please post you local.conf too Apr 06 08:41:14 Bernardo: kdepimpi does not build with qte-mt Apr 06 08:41:41 ipkg-cl -f /media/other/oe/oe/akita/tmp/work/akita-angstrom-linux-gnueabi/opie-image-1.0-r21/temp/ipkg.conf -o /media/other/oe/oe/akita/tmp/rootfs install task-opie-extra-games task-opie-extra-apps task-opie-extra-styles ipkg ipkg-collateral Apr 06 08:41:41 hrw: it fails with a missing include Apr 06 08:41:51 this clearly lacks a task-base Apr 06 08:41:58 NOTE: package gcc-cross-sdk-4.1.2-r1: task do_compile: started Apr 06 08:42:45 http://pastebin.ca/426617 Apr 06 08:43:57 koen: this does not inform. do_compile: completed is more important Apr 06 08:44:14 ;) Apr 06 08:44:37 koen: on amd64 it fails at do_compile Apr 06 08:45:52 gcc-cross-sdk-4.1.2-r1: task do_install: failed Apr 06 08:46:25 good! Apr 06 08:47:46 NOTE: Creating the CheckSum parser failed Apr 06 08:48:21 ERROR: Architecture did not match (40 to 3) on /work/i686-angstrom-linux-gnueabi/gcc-cross-sdk-4.1.2-r1/install/gcov/usr/local/arm/oe/bin/arm-angstrom-linux-gnueabi-gcov Apr 06 08:50:14 03koen 07org.oe.dev * r6a408a17... 10/ (1 packages/gcc/gcc-package-sdk.inc): gcc-package-sdk.inc: libg2c isn't always there, so don't error out on stripping Apr 06 08:50:59 koen: it can be escaped in if HAS_LIBG2C Apr 06 08:51:01 Bernardo: what bitbake do you use? Apr 06 08:51:48 psokolovsky: 1.6.7 Apr 06 08:51:58 hrw: ok Apr 06 08:52:46 hrw: no I have to fix insane .bbclass :( Apr 06 08:54:55 siliconmotion does not support randr ;( Apr 06 08:55:29 Bernardo: well, I kinda found where's your issue, and I have idea to fix it globally, but first want to reproduce it Apr 06 08:56:48 so it's not just pebcak? Apr 06 08:57:33 :) Apr 06 08:57:34 what's that? ;-) Apr 06 08:58:05 "problem exists between chair and keyboard" - what I used to tell some users their problem was when i was a sysadmin Apr 06 08:58:37 * XorA should really get that entered as a closing state in bugtracker Apr 06 08:58:42 Bernardo: ;-) nope, appears bug/corner case in bitbake Apr 06 09:01:02 I don't know why I have such a tendency to find those... :( Apr 06 09:05:59 bbl - daughter awake Apr 06 09:33:44 re Apr 06 09:35:12 Laibsch: wb Apr 06 09:35:19 hrw: how did the collie test go? Apr 06 09:35:41 none Apr 06 09:35:54 my 2.6 kernel does not boot Apr 06 09:37:22 I believe Thomas Kunze claims his collie booted. Did you make any changes? Apr 06 09:37:56 I will rebuild kernel with some other changes Apr 06 09:39:31 today I have 'bitbake world' day Apr 06 09:40:06 Laibsch: angstrom-2007.1-oabi/spitz from scratch -> nunome is in progress Apr 06 09:41:42 hrw: thank you VERY much Apr 06 09:53:36 zecke: I moved your include check to gcc-cross Apr 06 09:53:47 (for gcc 4.1.x, don't care enough for 3.x) Apr 06 09:55:21 gnome-vfs-dbus is more or less up-to-date with dbus 1.x? Apr 06 09:56:12 hrw: wanna funky quiz? please have a look at opie-image.bb and tell, why INSTALL_PACKAGES_*_append overwrites, not appends, INSTALL_PACKAGES? Apr 06 09:57:43 psokolovsky: because it should be machine_append instead of append_machine? Apr 06 09:58:05 hrw: exactly what it has, sorry for confusion Apr 06 09:59:45 psokolovsky: looks like I fscked it Apr 06 10:00:28 hrw: what you mean? syntax appears to be correct, just bitbake mis-intereprets it, it seems Apr 06 10:01:34 tried reverse it? Apr 06 10:04:24 hrw: gnome-vfs-dbus is dead Apr 06 10:04:31 ok Apr 06 10:04:50 hrw: gnome-vfs upstream isn't using orbit anymore Apr 06 10:04:53 ok Apr 06 10:05:35 * koen wishes gconf-dbus would die as well Apr 06 10:07:58 hrw: well, my idea is to kill those lines at all. But I'd like to understand why bitbake bugs out. Apr 06 10:08:24 psokolovsky: then provide other way of 'my device has bigger rootfs then yours' Apr 06 10:08:52 hrw: yeah, exactly. that's why I want to remove them. still wonder why bitbake chokes. Apr 06 10:29:25 03pfalcon 07org.oe.dev * rb13026a6... 10/ (1 packages/irda-utils/irda-utils_0.9.16.bb): Apr 06 10:29:25 irda-utils 0.9.16: Don't add irattach to startup services by default. Apr 06 10:29:25 * This just wastes power (to have IrDA on all the time), and conflicts with Apr 06 10:29:25 other userspace wanting to control the port. Apr 06 10:29:25 * Per RFC on teh OE mailing list. Apr 06 10:29:29 03pfalcon 07org.oe.dev * r38adb5ed... 10/ (1 packages/irda-utils/files/init): Apr 06 10:29:29 irda-utils: Make service script detect default IrDA port on PXA2xx devices. Apr 06 10:29:31 * Some issue is that some PXA2xx devices use ttyS2 for non-IrDA, these should Apr 06 10:29:33 be added to case *before* '"XScale-PXA2"*'. This should not be a problem as Apr 06 10:29:35 irattach is no longer started on boot, only manually. Apr 06 10:29:37 * Otherwise, assume that default is no IrDA, not some random serial port. Apr 06 10:29:39 * A bit of reformatting, too. Apr 06 10:29:43 03pfalcon 07org.oe.dev * rf4ff76f0... 10/ (1 packages/irda-utils/files/init): irda-utils: Reformat init script using tabs instead of spaces to save precious embedded space. Apr 06 10:35:02 Laibsch: qte or qte-mt? Apr 06 10:35:43 03pfalcon 07org.oe.dev * rf4c0e147... 10/ (1 packages/images/opie-image.bb): Apr 06 10:35:43 opie-image: Drop machine appends for opie-image, make it be the same on all machines. Apr 06 10:35:43 * Those appends were found to not work at all (overwrire INSTALL_PACKAGES instead) Apr 06 10:35:43 * Let's have default opie-image clean and lean, and the same for all machines, which Apr 06 10:35:43 helps support and QA. Use package manager, folks! Apr 06 10:35:44 * If there's a need for more packaged images, let's create opie-big-image. Apr 06 10:36:19 Laibsch: ping Apr 06 10:42:18 psokolovsky: don't forget to follow up in your RFC on the mailinglist Apr 06 10:43:44 koen: regarding irda-utils? I assume that easy and non-controversial thing, and I don't have anything else in q regarding that for now. Apr 06 10:44:09 not everyone tracks the commits Apr 06 10:44:55 hmmm Apr 06 10:45:06 llvm-gcc4 needs access to the llvm buildtree Apr 06 10:45:30 which conflicts with rm_work Apr 06 10:50:25 03rwhitby 07org.oe.dev * r8d8e263c... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: Demoted madfu - fails to build Apr 06 11:04:13 03rwhitby 07org.oe.dev * r2b9491b7... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: Demoted task-native-sdk - glibc fails to build Apr 06 11:06:37 where can i find examples to include some task into the build process of oe based distro Apr 06 11:07:17 mmmmm, dfu update is so nice and easy Apr 06 11:07:42 XorA: as long as you do not update wrong uboot Apr 06 11:10:37 hrw: Im safe, I got jtag :-) Apr 06 11:10:50 03marek.vasut 07org.oe.dev * r865854b3... 10/ (3 files in 2 dirs): Apr 06 11:10:50 libwpd: Add missing DEPENDS on libgsf. Apr 06 11:10:50 * Closes #2089. Apr 06 11:10:54 03marek.vasut 07org.oe.dev * r6c57393e... 10/ (1 packages/btscanner/btscanner_2.0.bb): btscanner 2.0: Add missing DEPENDS on libxml2. Apr 06 11:11:29 hrw: BTW ever get anywhere with collie? Apr 06 11:13:27 XorA: kernel boots, image not Apr 06 11:14:16 XorA: A friend of mine got OpenZaurus 3.5.5 with an opie-image running yesterday. He said he only needed some small fixes. Apr 06 11:14:39 amaldo: can you get him to file a bug with fixes then? Apr 06 11:15:00 * rwhitby updates tzdata to 2007e Apr 06 11:15:26 XorA: Yep, I told him already. He's testing things right now, and will submit later. Apr 06 11:15:41 nice Apr 06 11:16:25 amaldo: BTW opie is a thing I do not speak :-) Apr 06 11:16:31 amaldo: better off asking hrw Apr 06 11:17:22 XorA: He told me that GPE is a bit memory hungry for that machine, that he could only run one or two apps at the same time. That's why he switched to opie. Apr 06 11:20:45 03rwhitby 07org.oe.dev * r1c7cc034... 10/ (5 files in 3 dirs): tzdata: Updated to 2007e Apr 06 11:23:11 XorA: sorry, false alarm: he has a poodle. Apr 06 11:35:28 hrw: sorry to keep you waiting. I don't mind either way. You said you successfully built with both last time. Apr 06 11:37:48 It would be nice if you let me know your steps so I can reproduce them Apr 06 11:37:49 NOTE: package nunome-1.0.1-r1: task do_build: completed Apr 06 11:37:57 I don't get this. Apr 06 11:38:22 Laibsch: MACHINE=spitz DISTRO=angstrom-2007.1-oabi PALMTOP_USE_MULTITHREADED_QT = "yes" bitbake qte-mt;bitbake nunome Apr 06 11:46:13 For the OE bugdays: http://dominion.kabel.utwente.nl/koen/cms/the-bugs-are-out-there-again Apr 06 11:47:37 !oebug 2087 Apr 06 11:47:39 * * Bug 2087, Status: ASSIGNED, Created: 2007-04-06 03:27 Apr 06 11:47:40 * * marek.vasut(AT)gmail.com: btscanner-2.0 wont build Apr 06 11:47:41 * * http://bugs.openembedded.org/show_bug.cgi?id=2087 Apr 06 11:47:42 !oebug 2088 Apr 06 11:47:43 * * Bug 2088, Status: NEW, Created: 2007-04-06 03:33 Apr 06 11:47:44 !oebug 2089 Apr 06 11:47:45 * * marek.vasut(AT)gmail.com: Groff-1.19.2 installs to host system Apr 06 11:47:46 !oebug 2090 Apr 06 11:47:47 * * http://bugs.openembedded.org/show_bug.cgi?id=2088 Apr 06 11:47:49 * * Bug 2089, Status: RESOLVED (FIXED), Created: 2007-04-06 03:36 Apr 06 11:47:51 * * marek.vasut(AT)gmail.com: libwpd wont build Apr 06 11:47:53 * * http://bugs.openembedded.org/show_bug.cgi?id=2089 Apr 06 11:47:54 errrr .... Apr 06 11:47:55 * * Bug 2090, Status: NEW, Created: 2007-04-06 04:04 Apr 06 11:47:57 * * marek.vasut(AT)gmail.com: gkrellm-2.2.7 wont install Apr 06 11:47:59 * * http://bugs.openembedded.org/show_bug.cgi?id=2090 Apr 06 11:48:00 any ideas? Apr 06 11:48:06 slapin_nb, like this? Apr 06 11:48:14 Marex, :) Apr 06 11:48:32 Marex, probably not that fast... :) Apr 06 11:48:38 slapin_nb, ok ... :) Apr 06 11:51:43 hi, all! Apr 06 11:52:06 which IMAGE_TYPE could I set to get compressed ext2 image? Apr 06 11:52:29 and where to find all existing IMAGE_TYPEs? Apr 06 11:52:32 ext2.gz? Apr 06 11:52:33 slapin_nb: RTFC Apr 06 11:53:19 slapin_nb, RTFM Apr 06 11:53:56 aaargh Apr 06 11:54:06 RTFIL Apr 06 11:54:10 why do 'sdk' or 'sbox' target drag in gtk? Apr 06 11:54:35 koen: libgtk-dev? Apr 06 11:57:01 Should I send an email to the list about cleaning up the uboot files? Apr 06 11:57:48 uboot versus u-boot Apr 06 11:57:56 and uboot-utils Apr 06 11:58:07 03koen 07org.oe.dev * r636bac75... 10/ (1 packages/tasks/task-sdk-sbox.bb packages/tasks/task-sdk.bb): task-sdk-sbox, task-sdk: clean out base sdk packages, they should only contain a toolchain Apr 06 11:58:22 Crofton: please do Apr 06 11:58:27 (send mail and clean up) Apr 06 11:58:33 OK, it may be a few days Apr 06 11:58:44 rewriting some moldy old code this week Apr 06 12:03:11 another dumb question - how do you people run rsync to upload images - using special bitbake class or by script which runs bitbake? Apr 06 12:04:12 why use bitbake to rsync? Apr 06 12:04:28 rsync [options] images host:images Apr 06 12:05:29 koen: do you remember -lgcc_eh and why it happens? Apr 06 12:05:40 hrw, so it rsyncs stuff when build is finished (and optionally writes mails) Apr 06 12:05:46 I remember it, but I don't recall the solution Apr 06 12:06:21 03koen 07org.oe.dev * r55afd742... 10/ (1 packages/meta/meta-sdk-sbox.bb packages/meta/meta-sdk.bb): Apr 06 12:06:21 meta-sdk, meta-sdk-sbox: these are general targets, so they should only contain a basic toolchain Apr 06 12:06:21 * attempt to make them generic instead of hardcoding 'arm-linux' Apr 06 12:06:39 * koen hugs sed Apr 06 12:11:31 koen: I see.. -lgcc_eh == static Apr 06 12:12:06 slapin_nb: ^^^ note this and never ask again ;-) Apr 06 12:12:28 psokolovsky, what? Apr 06 12:12:43 slapin_nb: what hrw wrote Apr 06 12:12:58 psokolovsky, I have some patch lying smewhere but can't remeber... Apr 06 12:13:28 slapin_nb: you had that issue once if you don't remember Apr 06 12:13:45 psokolovsky, but anyway that was for some loder version of gcc. yes, I know. Apr 06 12:13:56 s/lodier/older Apr 06 12:13:58 hrw: ah, right Apr 06 12:14:28 psokolovsky, but it was just symlinc making.. Apr 06 12:14:35 symlink Apr 06 12:14:43 03koen 07org.oe.dev * rd14d459e... 10/ (1 packages/meta/meta-sdk-sbox.bb packages/meta/meta-sdk.bb): meta-sdk*: repair sed damage Apr 06 12:14:49 I dunno how hacky it is. Apr 06 12:14:56 mickey|zzZZzz: meta-sdk and meta-sdk-sbox should now be useable for everyone (including openmoko) Apr 06 12:16:53 ~lart util-linux Apr 06 12:16:54 * ibot whips out a sword and chops util-linux in half Apr 06 12:20:09 ~lart florian for making 'meta-sdk' include gpe, gpephone and more by default Apr 06 12:20:09 * ibot smacks florian up side the head with a clue-by-4 for making 'meta-sdk' include gpe, gpephone and more by default Apr 06 12:21:20 hi all Apr 06 12:21:34 hey mr_nice Apr 06 12:22:20 koen: hi Apr 06 12:24:18 hrw: it seems we might actually have a working sdk and sbox-sdk before ELC :) Apr 06 12:25:27 which is good Apr 06 12:25:29 hrw: I did as you said. Same problem as before (undefined reference to `main'). But I had not cleaned out tmp to see if this builds before going to sleep. I will clean tmp and retry but I doubt it will work. I just wonder what is the difference between your setup and mine Apr 06 12:25:32 koen: nice slides btw Apr 06 12:25:37 * Laibsch goes rm -Rf ../../tmp/angstrom/ && MACHINE=spitz DISTRO=angstrom-2007.1-oabi PALMTOP_USE_MULTITHREADED_QT="yes" bitbake qte-mt && MACHINE=spitz DISTRO=angstrom-2007.1-oabi PALMTOP_USE_MULTITHREADED_QT="yes" bitbake nunome Apr 06 12:25:56 Laibsch: set those vars in local.conf Apr 06 12:27:29 OK, just to make sure. I think they get picked up, though. DISTRO *is* set in local.conf. Apr 06 12:35:45 tmp/cross/lib/gcc/i586-poky-linux/4.1.1/libgcc_eh.a Apr 06 12:35:51 but no such for arm.. hmm Apr 06 12:37:14 rebuilding.. Apr 06 12:37:30 btw, what tarket meta-sdk actually produces? Apr 06 12:40:04 slapin_nb: sh5 for ultrasparc3 Apr 06 12:41:32 hrw, sorry for wrong formulation - I meant is it produces package of SDK or something else? Apr 06 12:41:42 try it? Apr 06 12:41:52 hrw, ok Apr 06 12:43:03 or look into recipe Apr 06 12:43:32 hrw, thanks Apr 06 12:51:23 hrw: although I'm still uncomfortable handing out toolchains to users Apr 06 12:51:59 koen: s/users/external developers/ too? Apr 06 12:52:18 koen: in OH we see a need for external toolchains Apr 06 12:53:22 hrw: I'd rather hand out a qemu images or a scratchbox toolchain Apr 06 12:53:36 external toolchains are a loaded gun pointed at your foot Apr 06 12:53:47 with a malfunctioning safety pin Apr 06 12:54:00 koen: qemu is fscking slow Apr 06 12:54:43 hrw: on a opteron it is twice as fast as a real underclocked NSLu2 Apr 06 12:54:49 Every minute I spend on SDK generation is a minutes not spent on packaged-staging :( Apr 06 12:56:02 time to drill some holes - cu Apr 06 13:09:33 time to make hrw unhappy Apr 06 13:11:59 03koen 07org.oe.dev * rafe622ad... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom: move gcc-cross to 4.1.2 as well Apr 06 13:13:07 hmm, looks like I need to rebuild the world again .... Apr 06 13:14:28 I started a build from scratch this morning Apr 06 13:18:15 zecke: what's the best way to disable package_qa_check_arch() in insane.bbclass for cross packages? Apr 06 13:23:29 koen: teach insane.bbclass about it Apr 06 13:23:57 ok Apr 06 13:24:10 * koen pushes a workaround Apr 06 13:24:46 03koen 07org.oe.dev * r633e68c0... 10/ (1 classes/insane.bbclass): insane.bbclass: short circuit architecture check for cross packages (e.g. gcc-cross-sdk) Apr 06 13:25:19 ~lart gstreamer Apr 06 13:25:19 * ibot gets a hotmal account and SPAMs gstreamer Apr 06 13:25:50 tmbinc: on the dreambox do you resize the video sink? Apr 06 13:28:00 hmmm Apr 06 13:28:10 libstdc++-dev_4.1.2-r1_i686.ipk is a bit wrong for a cross toolchain Apr 06 13:28:47 zecke: http://lists.linuxtogo.org/pipermail/openembedded-commits/2007-April/004848.html <- acceptable as short term solution? Apr 06 13:31:01 zecke: resize? Apr 06 13:31:56 tmbinc: on the fly resizing of the video playback, e.g. 640x480->320x240 Apr 06 13:34:51 Hello Apr 06 13:36:24 hey sirfred Apr 06 13:36:32 zecke: you can resize the video plane arbitrary on DM7025, if that's what you meant :) Apr 06 13:36:56 otherwise it will scale everything to fullscreen Apr 06 13:37:02 but you can select any box Apr 06 13:37:06 tmbinc: it has a BES? Apr 06 13:37:10 (well , down to a certain size) Apr 06 13:37:14 BES Apr 06 13:37:28 sweet Apr 06 13:37:30 Aren't all the gtk stock icons included in distributions? Apr 06 13:37:42 (gpe-mplayer:10983): Gtk-WARNING **: Could not find the icon 'media-playback-start'. The 'hicolor' theme ... Apr 06 13:38:08 Should they be present? Is it a bug? Apr 06 13:38:11 sirfred: we don't ship the hicolor-icon-theme by defaults Apr 06 13:38:31 and different versions of gtk+ ship different subsets of icons... Apr 06 13:38:39 icon hell Apr 06 13:38:45 correct Apr 06 13:39:10 So, what should be better? Put a depend against hicolor-icon-theme? Provide my own icons? Apr 06 13:39:24 I'm feeling lazy for the latter Apr 06 13:39:24 koen: sorry. BES? Apr 06 13:39:31 (i missed the '?') Apr 06 13:39:45 tmbinc: back end scaler Apr 06 13:40:21 it scales in realtime, so it'sa frontend scaler - isn't it? Apr 06 13:40:32 it could also scale into memory, but we aren't using that Apr 06 13:40:43 (as it already offer 3 seperate planes) Apr 06 13:41:28 if the video chipset scales it for you, I call it a BES :) Apr 06 13:41:36 ah ok Apr 06 13:42:04 ATI makes some distinction here - they try to avoid frontend (=realtime) scaling, in order to save bandwidth, as much as possible Apr 06 13:42:14 because realtime bandwidth is so much more expensive than normal Apr 06 13:42:22 but we don't care Apr 06 13:42:48 (that poses a certain restriction for downsampling, but that doesn't matter for us) Apr 06 13:43:02 ah, ok Apr 06 13:43:43 if we had that to-memory-scaling working, and would implement the mpeg decoder timeslicing (which is nearly working - but the remaining bits are very hard) then we could decode&display 16x MPEG2 SD Apr 06 13:44:12 but that would be an immense amount of work, and not even the ATI drivers suppport that at the moment.. Apr 06 13:44:47 (so if anybody wants to do an internship for low level nearly undocumented hardware coding... ;) Apr 06 13:47:01 tmbinc: What ATI chip are you talking about, If I'm allowed to ask? Apr 06 13:47:17 xilleon Apr 06 13:47:54 tmbinc: Never heard about that. Apr 06 13:48:07 * sirfred got stucked in the 90's, and the Imageon w100 Apr 06 13:48:12 Some embedded DVR chipset Apr 06 13:48:37 sirfred: can you copy&paste already? Apr 06 13:48:56 sirfred: or do you have to practice it some more in window mobile? ;) Apr 06 13:49:05 koen: :) Apr 06 13:49:14 koen: I'm afraid to say that I've abandoned that. Apr 06 13:49:33 I've decided to focus my idle time into my zaurus. Sorry. Apr 06 13:49:39 np Apr 06 13:50:02 The fact is that the problems, the differences between w100 and w3220 are not easy to figure out. Apr 06 13:50:16 We had aticore as source for the w100, and even with that, it was a hard task. Apr 06 13:51:03 sirfred: do you have a patch that updates the cursor to 32x32 on w3220? Apr 06 13:51:19 For the w3220 is even harder, because pocketpc dlls are using microcode, so, it's very difficult to find the way. Apr 06 13:51:21 koen: No. Apr 06 13:51:38 koen: I was trying for a while, but didn't got the cursor to work correctly. Apr 06 13:52:11 koen: My functions to map from xcursor to 2 bpp as ati cursor demands, didn't like the 32x32 size. Apr 06 13:52:38 koen: Anyway, hardware cursors are not going to make a difference. The great problem is xrandr. Apr 06 13:53:16 koen: We could just probably disable them for w3220, but disabling cursors, and disabling xrandr and xv, we have less more than Xw100 Apr 06 13:54:10 koen: Perhaps if someone owning a w3220 powered pda is interested... but with the little spare time I have, I prefer to work on my zaurus, sincerely. Apr 06 13:54:44 Well, I could change my mind if I win the lottery. ;-) Apr 06 13:55:36 I could live with disabled xrandr and cursors Apr 06 13:55:57 koen: So, why not Xw100 ? Apr 06 13:56:04 I think we should swap names Apr 06 13:56:11 Xw100 <-> Ximageon Apr 06 13:56:19 Xw100 isn't using videomemory, right? Apr 06 13:57:37 koen: The problem with Xw100 is a one-line-fix bug, as kdrive, if you don't set the alignment needed for the offscreen memory chunks, it just gives you always zero bytes. Apr 06 13:57:51 I would be happy with that fix :) Apr 06 13:57:53 koen: What Xw100 is not going to do, is to use the internal 320Kb of the card. Apr 06 13:58:01 koen: I can tell you what to change. Apr 06 14:01:39 koen: Ok, I've found it. Apr 06 14:01:54 koen: THis is how kaa calculates the amount of memory needed for every pixmap scanline: Apr 06 14:01:57 pitch = (w * bpp / 8 + pKaaScr->info->pitchAlign - 1) & Apr 06 14:01:57 ~(pKaaScr->info->pitchAlign - 1); Apr 06 14:02:26 koen: Note that when pKaaScr->info->pitchAlign is zero, pitch is always zero. I think it's rather a kdrive bug. Apr 06 14:03:43 So, the only needed thing is to set some pitchAlign in the w100 driver. The best place to do that is in ati_draw.c , function ATIDrawInit, in hw/kdrive/w100/ati_draw.c Apr 06 14:04:00 atis->kaa.pitchAlign = 16; Apr 06 14:04:23 koen: That should be enough to make Xw100 use offscreen memory, only external. But it's better than nothing. Apr 06 14:06:50 * koen makes a note of that Apr 06 14:08:38 koen: That way , Xw100 is the best option for hx4700 nowadays, as you still will have xrandr using the framebuffer emulation. Apr 06 14:09:15 03koen 07org.oe.dev * r59e33c8e... 10/ (4 files in 3 dirs): glib-2.0: update to 2.12.11n Apr 06 14:09:19 03koen 07org.oe.dev * rd06fd2cd... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom: prefer glib-2.0 2.12.11 Apr 06 14:20:37 ~lart ftp://ftp3.easysw.com/ for being slow Apr 06 14:20:38 * ibot breaks out the Hoover and sucks up ftp://ftp3.easysw.com/ for being slow Apr 06 14:31:21 sirfred: can you explain a bit more regarding "atis->kaa.pitchAlign = 16;"? where 16 is from? Apr 06 14:31:42 psokolovsky_: Empirical ;-) Apr 06 14:32:36 psokolovsky_: Perhaps a lower value can be used, but that seems to be safe for blit operations. Apr 06 14:32:44 sirfred: well, I'm seriously. where's that value from? is it arbitrary or it's w100 hardware parameter? Apr 06 14:33:16 psokolovsky_: It need to be the minimal value the w100 likes for every scanline in blitting for example. Apr 06 14:33:30 psokolovsky_: As I've said, it probably can be lower than that. But never zero. Apr 06 14:33:45 sirfred: ok. so, you say it is not initalized for w100 at all and stays at 0? Apr 06 14:33:54 psokolovsky_: Yes. Apr 06 14:34:06 psokolovsky_: So, offscreen requested memory is always zero. Apr 06 14:34:06 03koen 07org.oe.dev * r10a90319... 10/ (1 packages/cups/cups_1.2.10.bb): cups: add 1.2.10 Apr 06 14:34:40 sirfred: so it's safe to put that assignment unconditionally, and it's going to work on any w100-compatible chip? Apr 06 14:34:48 psokolovsky_: Probably. Apr 06 14:35:13 psokolovsky_: But take into account that actually Xw100 is not using offscreen memory. If you change that, it's going to use it. Apr 06 14:35:31 psokolovsky_: You can probably trigger new bugs. Apr 06 14:35:47 psokolovsky_: But probably not, as the blitting functions were already used for onscreen-onscreen copy Apr 06 14:36:29 psokolovsky_: Take into account that most of knowledge about Imageon chipsets is from reverse engineering and .. just guessing. Apr 06 14:36:39 Try, fail, restart... Apr 06 14:36:41 koen: you didn't edit checksums.ini? Apr 06 14:36:51 sirfred: ok, I probably make patch, so your wisdom is not lost, or you didn't haev to repeat it again and again ;-) Apr 06 14:36:53 zecke: I didn't Apr 06 14:37:07 koen: bad boy Apr 06 14:37:20 psokolovsky_: Actually, I think it's a kdrive bug. Apr 06 14:37:34 psokolovsky_: They should shift to 1 that value, never use zero. Apr 06 14:38:03 sirfred: for it not to check pitchAlign != 0? I see for example that ati/ati_draw.c sets it Apr 06 14:38:15 ok Apr 06 14:38:45 psokolovsky_: Yes, they should check it and shift to one. As it's made in other parts of kdrive code with similar parameters. Apr 06 14:38:52 sirfred: maybe it's intentionally, it == 0 => don't use offscreen ;-) Apr 06 14:38:57 But a pitchAlign of zero is just disabling offscreen memory. Apr 06 14:39:00 psokolovsky_: Could be. ;-) Apr 06 14:39:31 But a great comment should be nice. Apr 06 14:43:24 sirfred: now if you only could hint me how will I know that offscreen stuff is being used? ;-) Apr 06 14:43:53 psokolovsky_: There's a define in the kdrive/src/koffscreen.c Apr 06 14:44:00 psokolovsky_: To enable some logs, let me see... Apr 06 14:44:18 #define DEBUG_OFFSCREEN 0 Apr 06 14:44:23 Change that with 1. Apr 06 14:45:34 psokolovsky_: You should see some stderr logs for every offscreen request, with size and something more, look for DBG_OFFSCREEN calls in that file. Apr 06 14:45:54 sirfred: ok, thanks! Apr 06 14:46:32 DBG_OFFSCREEN (("Alloc 0x%x -> EMPTY\n", size)); Apr 06 14:46:49 That is the log that happens when the requested size was zero Apr 06 14:47:18 psokolovsky_: This one for a good alloc: Apr 06 14:47:18 [16:38] sirfred: for it not to chec Apr 06 14:47:20 oops Apr 06 14:47:23 DBG_OFFSCREEN (("Alloc 0x%x -> 0x%x\n", size, area->offset)); Apr 06 14:47:31 Cut&Paste, my great enemy Apr 06 14:47:41 ok Apr 06 14:49:39 re Apr 06 14:51:14 hrw, btw, what is RTFIL? could not find it on wikipedia... :) Apr 06 14:55:00 lol Apr 06 14:57:24 rtf irc logs? Apr 06 15:03:27 hrw, where to rtf IRC logs? Apr 06 15:04:46 ~logs Apr 06 15:04:48 well, logs is apt/ibot/infobot/jbot/purl all log daily to http://ibot.rikers.org// where channelname is html encoded ie: %23debian | lines that start with a space are not shown | some channels have stats at http://ibot.rikers.org/stats/.html.gz Apr 06 15:05:37 03xora 07org.oe.dev * r8936e843... 10/ (9 files in 5 dirs): Apr 06 15:05:37 linux-smdk2440_2.6.20+git : add a new version updated to 2.6.21-rc6+asoc Apr 06 15:05:37 remove older versions that are in various states of disrepair. Apr 06 15:17:24 bye Apr 06 15:20:58 03xora 07org.oe.dev * r091a66f4... 10/ (1 packages/mplayer/files/mplayer-imageon-svn.patch): mplayer-imageon-svn.patch : forgot to add this long ago Apr 06 15:52:05 morning Apr 06 15:58:08 sirfred, koen: how this looks: http://pastebin.ca/427042 Apr 06 15:59:35 psokolovsky_: Looks fine, can see an offscreen alloc Apr 06 16:00:04 psokolovsky_: At 0x96000, where the visible framebuffer ends Apr 06 16:00:19 sirfred: ok, so it show something too, and even doesn't segfault ;-). will play a bit more and commit. Apr 06 16:00:30 sirfred: any specific test I shoudl perform? Apr 06 16:01:03 psokolovsky_: Humm, I expected to see some copy logs Apr 06 16:01:15 psokolovsky_: Perhaps it doesn't emit logs, let me see... Apr 06 16:01:44 sirfred: that from manual run of Xw100 ;-) Apr 06 16:02:08 psokolovsky_: Try to run, for example gpe-login Apr 06 16:02:22 psokolovsky_: That should make some allocs and copies. Apr 06 16:03:40 psokolovsky_: You will see things like Enter AtiPrepareSolid with some offsets. Apr 06 16:03:46 sirfred: http://pastebin.ca/427048 Apr 06 16:03:54 hi Apr 06 16:04:12 psokolovsky_: from your commit it seems I can try again building opie-image, right? Apr 06 16:04:23 Bernardo: yeah! Apr 06 16:04:24 psokolovsky_: Very well, it's performing offscreen blitting. Apr 06 16:04:29 psokolovsky_: Do it looks fine? Apr 06 16:04:35 psokolovsky_: On the screen, I mean Apr 06 16:04:36 sirfred: yeah ;-) Apr 06 16:04:40 can I try using qte-mt (PALMTOP_USE_MULTITHREADED_QT="yes") or should I leave that out of local.conf for now? Apr 06 16:04:49 psokolovsky_: So, congrats, you've fixed it. :-) Apr 06 16:05:04 Bernardo: you should have only require opie-1.2.3-pre Apr 06 16:05:10 psokolovsky_: Performance should be better now, but... who knows Apr 06 16:05:21 sirfred: you! ;-) Apr 06 16:05:27 psokolovsky_: ok, I've got that, will try now. Apr 06 16:05:47 psokolovsky_: It's a pitty that in rotated mode, the offscreen memory is disabled by Xfbdev routines. Apr 06 16:06:11 psokolovsky_: But since the kernel framebuffer level is not able to rotate properly the w3220 screen, you will have to live with that. Apr 06 16:06:27 sirfred: in such a way that it affects even Xw100? Apr 06 16:06:48 psokolovsky_: Yes. Xw100 relays in fbdev routines to perform rotation. Apr 06 16:07:09 It's made in software, just with a shadow framebuffer and an update function to rotate in software. Apr 06 16:07:45 That way, framebuffer needs double amount of memory Apr 06 16:08:17 And, in a perhaps weird decission, fbdev decides to set offscreen memory to zero Apr 06 16:09:18 Well, perhaps it makes sense, as the rotation is made in software, who cares how are we copying to the primary framebuffer? It's a disaster anyway Apr 06 16:10:24 psokolovsky_: On the other hand, Ximageon just tells to the w100 (using the framebuffer level) to rotate: It's fast, transparent and doesn't consume additional memory Apr 06 16:11:41 sirfred: nice. let's leave for another time though, I'm not ready for full-fledged X hacking yet ;-) Apr 06 16:12:03 psokolovsky_: Well, who's ready? ;-) Apr 06 16:12:19 sirfred: but to get to it some day, and hope I'll have chance to ask you more questions. ;-) Apr 06 16:12:38 psokolovsky_: Of course. Apr 06 16:13:34 thanks! Apr 06 16:14:44 psokolovsky_: np Apr 06 16:27:18 03pfalcon 07org.oe.dev * rb359807f... 10/ (3 files in 3 dirs): Apr 06 16:27:18 xserver-kdrive 1.2.0: Add patch to actually enable offscreen pixmap acceleration in Xw100. Apr 06 16:27:18 * Courtesy of Manuel Teira (sirfred). Apr 06 16:27:18 * Tested on hx4700. Apr 06 16:29:54 03pfalcon 07org.oe.dev * r9bb2e14e... 10/ (1 packages/xorg-xserver/xserver-kdrive-common.inc): Apr 06 16:29:54 xserver-kdrive: remove bogus PACKAGE_ARCH overrides for Palm machines. Apr 06 16:29:54 * They don't use any adhoc patches to require that. Apr 06 16:29:54 * TODO: do the same for poodle. Apr 06 16:37:20 03koen 07org.oe.dev * r741da7c0... 10/ (3 files in 2 dirs): task-sdk: move non base stuff to task-sdk-extras to avoid building half of OE when inheriting debian.bbclass Apr 06 16:54:01 update-modules is generating modules.conf, but modprobe is looking for modprobe.conf on my target system. Anyone know what is going on here? Apr 06 17:03:16 looks like /etc/modprobe.d/options is now the correct place Apr 06 17:08:22 Curious, if I start X manually and then matchbox-session, window size policies are not the same than when started from gpe-dm. Apr 06 17:08:43 For example, the filechooser shows fullscreen, what's really better. What could be the cause of this? Apr 06 17:16:40 sirfred: gpe-conf(d) changes some xsetttings Apr 06 17:17:51 puzzles broken again :( Apr 06 17:18:02 hrw|gone, Apr 06 17:18:41 maybe make svn checkout default (with appropriate revision no)? Apr 06 17:22:28 !oebug 2091 Apr 06 17:22:30 * * Bug 2091, Status: NEW, Created: 2007-04-06 10:28 Apr 06 17:22:31 * * slapinid(AT)gmail.com: puzzles package is non-fetchable again Apr 06 17:22:32 * * http://bugs.openembedded.org/show_bug.cgi?id=2091 Apr 06 17:22:41 slapin_nb: do that, submit patch Apr 06 17:23:33 psokolovsky_, RP is against it, I want comments. Apr 06 17:24:47 slapin_nb: comments on what? what was his argumentation? Apr 06 17:25:09 hey mickey|meeting Apr 06 17:25:15 psokolovsky_, look at some IRC logs archive. Apr 06 17:25:41 koen, do you feel anything bad with suggestion on oebug 2091? Apr 06 17:25:50 slapin_nb: I suggest yoy make a patch and submit it. Apr 06 17:26:10 I have no real opinion on that Apr 06 17:26:39 psokolovsky_, not a problem for me, but RP thinks it is wrong thing to do. Apr 06 17:27:09 psokolovsky_, I myself is not sure, because I could misunderstand something. Apr 06 17:27:37 slapin_nb: patches welcome Apr 06 17:28:07 it is good to have ability to download stuff using tarball, but a problem is they make it unavailable too often, so I suggest SVN checkout as solution. Apr 06 17:28:27 psokolovsky_, patch is trivial, whole idea needs discussion. Apr 06 17:28:51 IIRC, RP is maintainer of it. Apr 06 17:29:09 chers Apr 06 17:29:12 cheers, eve Apr 06 17:29:13 n Apr 06 17:36:56 mickey|meeting: how's the meeting coming along? Apr 06 17:40:36 awesome Apr 06 17:40:59 coffee and cookies :) Apr 06 17:41:01 time is running too fast, but other than that, we're talking about the most exciting stuff atm. Apr 06 17:41:25 * mickey|meeting whispers future devices Apr 06 17:41:34 ~v12 Apr 06 17:41:39 mickey|meeting: stop chating, focus Apr 06 17:41:42 absolutely Apr 06 17:41:44 [v12] the next must-have in the Neo1973, http://www.ultimatestupidity.com/pics/1/diesel/ Apr 06 17:41:44 direct injection Apr 06 17:41:55 zecke: Sean is drawing atm., i have a minute Apr 06 17:43:47 mickey|meeting: did you see my mail about the branch? Apr 06 17:44:06 koen: got it, yet Apr 06 17:44:09 s/yet/yeah/ Apr 06 17:45:01 mickey|meeting: you need to start thinking about wo to give OE access for that as well :/ Apr 06 17:45:23 * koen hasn't been impressed by non-mickeyl commits to the openmoko svn tree Apr 06 17:47:14 zecke: you have cookies? Apr 06 17:47:21 zecke: I'm all out of cookies Apr 06 17:47:42 koen: don't talk about food, I'm about to throw over... Apr 06 17:55:35 psokolovsky_: opie-image is booting now, no more complains about not finding init Apr 06 17:56:14 Bernardo: cool Apr 06 17:57:51 ok, it seems my regular opie desktop - just kdepim is missing Apr 06 17:58:58 time to build more stuff... iqnotes, opie-reader are mandatory, all the rest I can live without. :) Apr 06 17:59:05 hmm I need last.fm for my neo Apr 06 17:59:21 as you put on the wiki, no sound Apr 06 17:59:49 Bernardo: yep, that needs to be researched Apr 06 18:00:21 it is using alsa, right? or expecting oss emulation? Apr 06 18:02:09 Bernardo: OSS emulation Apr 06 18:02:11 ouch... hard lock when I inserted my sandisk plus wifi card Apr 06 18:02:36 is there a alt+sysreq equiv on the akita? Apr 06 18:04:14 alt+home Apr 06 18:04:34 Bernardo: yeah dont use a connect+ is oopses the kernel Apr 06 18:04:56 I can't afford a new wifi card now... :( Apr 06 18:05:24 I can work without it, though Apr 06 18:05:32 Bernardo: got a serial cable? Apr 06 18:05:58 a usb/serial one, I hacked from a nokia cable to use with a wifi router Apr 06 18:08:10 I don't have the plug to connect to the akita port - will check on ebay if I can find one. ANd also try if I can reproduce the lockup on a x86 2.6.20 kernel Apr 06 18:08:29 if so, it might be easier to trace Apr 06 18:11:05 woot!! the "Easter Bunny" just gave me a picture of Ben Franklin @ work :) Apr 06 18:12:06 koen: how are you gonna fit that v12 in a phone tho? Apr 06 18:12:27 hvontres|poodle: magic Apr 06 18:12:44 herbal one? Apr 06 18:17:14 can't hang it on x86... but no longer works, on 2.6.17 it did work Apr 06 18:22:15 psokolovsky_, I added that patch. Apr 06 18:22:42 psokolovsky_, a bit rough, probably. please comment if you find it bad in some way. Apr 06 18:25:28 !oebug 2091 Apr 06 18:25:29 * * Bug 2091, Status: NEW, Created: 2007-04-06 10:28 Apr 06 18:25:30 * * slapinid(AT)gmail.com: puzzles package is non-fetchable again Apr 06 18:25:31 * * http://bugs.openembedded.org/show_bug.cgi?id=2091 Apr 06 18:28:15 mtn: 741da7c046302cd2a55436a833775a51b3118e29 koen@openembedded.org 2007-04-06T16:34:26 Apr 06 18:28:15 mtn: bd2a1b3443e4c99866c718f6aaa29ffc02817dd0 pfalcon@openembedded.org 2007-04-06T16:28:31 Apr 06 18:28:15 m Apr 06 18:28:19 2 heads again Apr 06 18:30:02 slapin_nb: read the monotone manual Apr 06 18:30:09 :) Apr 06 18:30:36 koen, I know about -r Apr 06 18:30:51 slapin_nb: so stop complaining about it then Apr 06 18:31:03 hello all Apr 06 18:31:10 koen, I'm not complaining. just curious. Apr 06 18:31:28 * koen plans to inform the channel that water is still wet on regular intervals Apr 06 18:31:55 http://rafb.net/p/9OSMeM76.html <- six hours and now I'm forced to hack around... Apr 06 18:33:37 koen, thanks for live analogy Apr 06 18:36:25 slapin_nb: fwiw, the mtn server automerges every hour or so, so multiple heads should go away automagically Apr 06 18:36:56 is updating to 0.34 worth it? Apr 06 18:37:25 koen, thanks for explanation Apr 06 18:39:21 * cbrake moves his compulab builds to EABI :-) Apr 06 18:49:06 Is the guy reporting the bug in uboot-utils around? Apr 06 18:57:09 What too can I use to find out various raw/cooked keycodes on linux? Apr 06 18:59:21 psokolovsky_: take a look at the QtE keyboard driver Apr 06 19:01:07 zecke: nice. exactly to check its sanity I need separate tool ;-) Apr 06 19:02:40 psokolovsky_: do you know goedel? Apr 06 19:03:21 zecke: theorem of incompleteness of arithmetic? ;-) Apr 06 19:03:36 sounds like a proper translation Apr 06 19:03:49 psokolovsky_: completeness and incompleteness Apr 06 19:04:13 psokolovsky_: there is this huge keymap and the first line states a python tool Apr 06 19:04:27 psokolovsky_: these numbers where used to generate the table (and then hand edited... :}) Apr 06 19:04:48 damn! python! that's what I need! Apr 06 19:04:56 lol Apr 06 19:05:11 psokolovsky_: alternatively you can spy at kdrive Apr 06 19:05:39 zecke: I just hope some package includes some nice utility for that ;-) Apr 06 19:06:17 hoped Apr 06 19:07:06 psokolovsky_: alternatively SYSRQ+X is your friend Apr 06 19:07:33 psokolovsky_: change your terminal into raw mode, print the keycodes and if something goes bad you can safely return to XLATE Apr 06 19:08:15 psokolovsky_: http://handhelds.org/~zecke/downloads/python_keytable_creator.py Apr 06 19:08:42 zecke: yeah, found it, thanks. that py is *big* ;-) Apr 06 19:08:58 psokolovsky_: it is mostly my documentation Apr 06 19:09:06 psokolovsky_: self executable documentation :) Apr 06 19:09:22 ok ;-) Apr 06 19:09:52 I think this 'self' is wrong as you need to import and then type the statements you want Apr 06 19:10:02 psokolovsky_: but why do you suspect that the driver is broken? Apr 06 19:11:07 zecke: well, not really broken, just doesn't work for ipaqs. specifically, power and record buttons are not recognized, so I look for ways to add them. Apr 06 19:21:31 hrw|gone: What are you using to generate our e-mail sigs? Those things are *FUNNY* Apr 06 19:24:18 hvontres|poodle: its perl script which use external textfile with quotes Apr 06 19:24:34 hvontres|poodle: and I am updating quotes from time to time Apr 06 19:25:10 slapin_nb: RP is against using unversioned tarball which puzzles author provides. Apr 06 19:25:36 slapin_nb: we can move to SVN version instead but it should be set to working revision instead of being floating Apr 06 19:26:09 happy easter everyone (or have a nice weekend for !christians) Apr 06 19:26:29 hrw|gone, please look at oebug 2091 Apr 06 19:26:39 slapin_nb: submit patch Apr 06 19:26:43 hrw|gone, there is patch Apr 06 19:27:02 hrw|gone, happy easter, too! Apr 06 19:27:41 slapin_nb: look ok for me Apr 06 19:28:04 slapin_nb: atleast it looks like you start learning how to report bugs ;) Apr 06 19:29:09 bye Apr 06 19:29:45 hrw|gone, bye, I always new but I'm usually lazy or have no time :) Apr 06 19:47:36 hello all Apr 06 19:48:11 vivijim: hi Apr 06 19:48:26 hi likewise Apr 06 20:37:06 rwhitby: write a Makefile that helps to use MokoMakefile :) Apr 06 20:37:07 scnr Apr 06 20:41:03 hi koen Apr 06 20:41:14 hey likewise Apr 06 20:41:25 koen: I read on OE ml you are interested in 'upstart'? Apr 06 20:41:47 koen: I was just investigating using it for OE, being bitten by initng unstableness. Apr 06 20:41:50 not really, just looking ahead Apr 06 20:42:27 if we are going to redo initscripts we might as well try to future proof it Apr 06 20:43:07 zecke: heh :-) Apr 06 20:43:30 * rwhitby senses disapproval of using a makefile to control bitbake in the OE core team :-) Apr 06 20:43:45 rwhitby: not disapproval Apr 06 20:43:58 rwhitby: pure hate and anger ;) Apr 06 20:44:07 with fire and flames Apr 06 20:44:40 as long as it's *pure*, then that's ok :-) Apr 06 20:44:43 seems like zecke had the same CS course about makefiles as I did Apr 06 20:44:53 don't want any of that impure hate and anger ... Apr 06 20:44:59 rwhitby: but make qemu-flash-official sounds like something I have to try Apr 06 20:45:02 hehe Apr 06 20:45:30 rwhitby: no, I have grown up, I can offer the full pure anger Apr 06 20:45:46 rwhitby: no "I can't believe its impure hate and anger" Apr 06 20:46:05 zecke: I'm going to add "make flash-neo-official" next week sometime so I don't have to remember the dfu-util cmdline options to exclude my bluetooth dongle from the upgrade ... Apr 06 20:47:05 rwhitby: the only time MokoMakefile annoys is when stupid people use it. "You don't have bzip2 installed, please install it" "Oh MokoMakefile breaks and I don't know what to do" :) Apr 06 20:47:22 koen: note that I don't say that makefiles can replace bitbake - makefiles have limits. I just believe that makefile limits are further along than script limits. Apr 06 20:47:40 zecke: the same people would give the same bug report s/MokoMakefile/bitbake/ Apr 06 20:47:55 rwhitby: my only problems with makefiles like mokomakefile is what zecke says Apr 06 20:48:10 the question is whether you want those people running the build system _at_all_ Apr 06 20:48:22 rwhitby: that is the question. actually you mention the two questions Apr 06 20:48:29 with mokomakefile, you get a lower barrier to entry. which can be a good or bad thing. Apr 06 20:48:32 rwhitby: and even projects like X.org face that Apr 06 20:49:29 rwhitby: I think lowering the barrier is good, and I generally agree with MokoMakefile Apr 06 20:49:39 rwhitby: could make setup make a turing test or such? Apr 06 20:49:51 my hope is that MokoMakefile allows some coding guru who has never used bitbake or OE before to get productive in developing OM applications without having to cross over the largish barrier to entry. Apr 06 20:50:19 rwhitby: and you have my support there Apr 06 20:50:37 the turing test should be in bitbake sanity.bbclass, right? Apr 06 20:50:58 and when first compiling OpenSlug I have used the MaterMakefile even if I knew OE Apr 06 20:51:10 rwhitby: isn't the makefile curing symptons rather than the problem? Apr 06 20:51:34 bbl Apr 06 20:52:26 koen: good question. I guess the same question could be asked of external toolchains. Do you give an application developer a toolchain RPM or .deb to install, or do you make them run bitbake to get it? Apr 06 20:53:11 Even with mokomakefile support for building qemu, there will still be a very large number of people who won't use it until there is a windows installer ... Apr 06 20:54:05 rwhitby: A few hours ago I said something like "giving out external toolchains is giving people a loaded gun without a safety pin, aimed at their foot" Apr 06 20:54:24 koen: how far is your foot away? Apr 06 20:54:32 yep, I saw that. that's why I used that example :-) Apr 06 20:55:19 but PalmOS would not have it's huge application developer base if each one of them had to install bitbake and OE to build their toolchain from scratch. Apr 06 20:55:29 right Apr 06 20:55:52 it's a tiered system Apr 06 20:56:42 that's why I rather work on packaged-staging that driving people with a hole in their foot to the ER Apr 06 20:57:15 at one end of the spectrum are those that fix toolchain problems. then you go through those to build toolchains, those who package and support toolchains, and those who simply want to use toolchains and ipkg upgrade if the toolchain changes, cause they are only interested in the application development. Apr 06 20:57:51 right now, I'd hand out qemu images with gcc + friends installed Apr 06 20:57:54 ah, now you're talking about the mechanism of delivering external toolchains, rather than the principle of external toolchains itself. Apr 06 20:58:22 yeah Apr 06 20:58:28 those are tied together pretty close Apr 06 20:58:34 and on the mechanism, I agree with you - you need something which is easily and automatically updateable. Apr 06 20:58:54 (and isolated from host configuration somewhat) Apr 06 20:59:00 yes Apr 06 20:59:10 Anyone know which channel I should bother with programming questions (regarding shared library development)? Apr 06 20:59:12 everyone should switch to 64bit :) Apr 06 20:59:27 likewise: here? Apr 06 20:59:38 or rather bigendian 128bit Apr 06 20:59:50 zecke: Can I define variables in a .so that are shared between all users? Apr 06 21:00:07 likewise: globally? btw multiple processes? Apr 06 21:00:18 zecke: Yes. C language. Apr 06 21:00:38 likewise: not on UNIX Apr 06 21:01:07 likewise: or only constant variables (but that wasn't your question) Apr 06 21:01:19 zecke: so all modifiable data is instantiated for each process? Apr 06 21:01:26 likewise: right Apr 06 21:02:17 If I want to have my library share data across processes, it should use posix shared memory or such? Apr 06 21:02:40 likewise: you cayes Apr 06 21:03:00 zecke: I was just wondering if there was a common 'best practice'. Apr 06 21:03:55 likewise: I have never done this in practive but you can use sysv5 shared memory or mmap Apr 06 21:13:14 NIght Apr 06 21:25:35 hmm Apr 06 21:25:42 mickey|meeting: hi :) Apr 06 21:25:52 hi Apr 06 21:26:01 good night all Apr 06 21:26:37 mickey|meeting: yes? Apr 06 21:27:01 nothing, just checked whether i'm still connected :) Apr 06 21:27:16 ibot: hmm Apr 06 21:27:27 somebody said hmm was hidden markov model Apr 06 21:27:50 heh Apr 06 21:28:12 * zecke looks indirectly to mickeyl's wife and asks if such a thing exists? Apr 06 21:29:08 mickey|meeting: do we make everyone rebuild tmp from scratch to go back to a frozen OE version? Apr 06 21:29:48 (for openmoko) Apr 06 21:32:36 psokolovsky_: most stuff works - but bitbake meta-opie dies with qa errors in siĀ«ome stuff (opie-sh, opie-theme) and fails when building lirc-modules. Apr 06 21:34:44 Bernardo: that means someone should fix them ;-) Apr 06 21:35:45 yep -I'll try... Apr 06 21:36:16 rwhitby: yeah, I'm afraid that's the only sane way Apr 06 21:37:08 ok, I'll work out how to get MokoMakefile to detect if someone is not running the frozen version, and print out a message which tells them to either clobber everything, or unfreeze if they want to continue with unofficial builds. Apr 06 21:37:43 mickey|meeting: did the frozen rev change? Apr 06 21:37:55 no Apr 06 21:38:06 not yet, at least Apr 06 21:44:38 zecke: ask koen next semester :) I think it's one of those wierd controls things Apr 06 21:46:52 'night Apr 06 22:04:29 03philippe 07org.oe.dev * rc6346a6d... 10/ (1 packages/gtk-webcore/osb-jscore_svn.bb): Apr 06 22:04:29 packages/gtk-webcore/osb-jscore_svn.bb Apr 06 22:04:29 * fix issue with missing create_hash_table script Apr 06 22:10:35 mickey|meeting: frozen OE version implemented in MokoMakefile Apr 06 23:12:44 Quick Question: Has anyone used OE on FreeBSD (as a compile env)? Apr 06 23:13:44 HistoryOnTheRoad: it's listed in http://www.openembedded.org/wiki/OEandYourDistro, so there's a chance :-) Apr 06 23:14:11 (although it says "bitkeeper" so that shows that entry is over 2 years old) Apr 06 23:15:03 Well, if it's 2+ years old, then it was supported _at one time_ ... which is encouraging, at least :) Apr 06 23:17:49 anyone have a recommendation for a development board or maybe even a cheap device to get started with embedded linux? seems to be thousands of them out there. Apr 06 23:19:02 rwhitby: are there any min. requirements for OE then? Apr 06 23:19:35 (gcc 4 or 3.2 ... or a specific version of them)? Apr 06 23:32:18 nite Apr 06 23:36:30 concept10, define cheap? Apr 06 23:37:27 * T0mW is still looking for the vendor that is giving boards away Apr 06 23:37:50 Crofton__, under $350... I say that because if I go over, I may as well get the openmoko based phone Apr 06 23:38:03 heh Apr 06 23:38:10 Omap starter kit Apr 06 23:38:27 I think there are a number of options under 350 Apr 06 23:38:35 OSK does not have display Apr 06 23:40:17 I have this firewall box with a AMD/Geode 233 with a full motherboard with ide/usb/serial/paraport but it seems too limited Apr 06 23:40:50 what is limiting you? Apr 06 23:41:20 Crofton__, the integrated graphics (cant get X running, yet) Apr 06 23:42:01 Crofton__, I really dont need X, I would just like to get it to run Apr 06 23:42:15 hmm, we need a web page linking machines OE supports Apr 06 23:42:21 yeah, I understand Apr 06 23:43:31 I just saw a guy boot a board with LinuxBIOS with kdrive/tinyx and matchbox in 6 secs. That is so great. Apr 06 23:45:24 I must run now, I'll add thinking about adding a page to the wiki linking to HW we suport Apr 06 23:48:20 Crofton__, okay, thanks and later Apr 06 23:50:55 concept10: The cheapest device I'm aware of that OE runs on is the Linksys NSLU2 Apr 06 23:53:15 NAiL, yeah, looks to be under $100 on ebay. thanks Apr 06 23:53:46 There's a lot of info over at nslu2-linux.org about the device Apr 06 23:55:21 NAiL, yeah, ive been reading stuff about it all day. **** ENDING LOGGING AT Sat Apr 07 02:59:57 2007