**** BEGIN LOGGING AT Wed Oct 17 02:59:56 2007 Oct 17 05:40:52 03henryk 07org.oe.dev * r129b44b2... 10/ (1 packages/openmoko2/openmoko-appmanager2_svn.bb): openmoko-appmanager2: DEPENDS on libipkg from ipkg Oct 17 06:01:03 good morning all Oct 17 08:16:46 is kernel-vmlinux not meant to be in blacklist (in kernel.bbclass) to avoid having it added to rdepends? Oct 17 08:23:44 morning Oct 17 08:25:26 good morning all Oct 17 08:25:35 Esben: it should be Oct 17 08:25:52 Esben: by I noticed it got dragged in when I did 'ipkg install kernel-modules' Oct 17 08:26:11 03koen 07org.oe.dev * ra84c8b1a... 10/ (1 packages/gsm/libgsmd_svn.bb): libgsmd: add maxrev to patches that have been applied upstream (finally!) Oct 17 08:26:33 koen: it is not in my kernel.bbclass (a few days since last update) Oct 17 08:27:10 and I don't use modules for these targets currently. Oct 17 08:51:41 ~lart network scripts and resolv.conf Oct 17 08:51:41 * ibot urinates on network scripts and resolv.conf Oct 17 09:14:42 hrw: task-base guru, where should I add openssh-sftp-server? Oct 17 09:16:33 distro-extra-rrecommends in angstrom-2007.1.conf? Oct 17 09:18:08 XorA: like koen told Oct 17 09:18:13 DISTRO_EXTRA_RDEPENDS probably Oct 17 09:18:24 rrecommends Oct 17 09:18:32 erm DEPENDS Oct 17 09:18:50 my devices do not need sftp support so I can consider removing it Oct 17 09:19:07 hrw: but you DEPEND on avahi which advertises sftp Oct 17 09:20:01 right.. Oct 17 09:23:30 so move avahi to RRECOMMENDS as well :) Oct 17 09:23:37 hey lrg Oct 17 09:24:29 koen: Im happy with that if you are Oct 17 09:24:43 I'm happy with that Oct 17 09:24:56 it's like that section of angstrom.conf has had lots of though Oct 17 09:24:57 t Oct 17 09:25:33 2.6.23 is working pretty neat on my pvr Oct 17 09:25:47 hrw: how does .23 compare to .22 on c7x0? Oct 17 09:28:13 hard to tell as I did only simple tests Oct 17 09:28:23 suspend/resume is maybe a bit faster Oct 17 09:29:20 should be add it to OE and make it the default for angstrom-2008? Oct 17 09:30:28 I want to make few changes in them Oct 17 09:30:59 ok, I'll hold of merging till you're done with it Oct 17 09:33:06 heh Oct 17 09:33:09 "we all know that sound support in the 2.6 kernel has a long way to catch up to the quality of the 2.4 kernel..." Oct 17 09:33:22 clueless pdaX people Oct 17 09:39:05 koen_: http://www.oesf.org/forum/index.php?showtopic=24895&st=0&gopid=169279&#entry169279 Oct 17 09:40:52 XorA: ;DD Oct 17 09:42:02 XorA: how much of wm8731/8750 driver was written by you? Oct 17 09:42:13 hrw: zero Oct 17 09:42:58 ok, this is outrageous, highs for the next few days Oct 17 09:43:15 25, 22, 26, 23, 24 C Oct 17 09:44:45 drat Oct 17 09:44:46 Crofton|home: 8C here Oct 17 09:44:57 253 days since I rebooted my server Oct 17 09:45:00 its to warm to sleep Oct 17 09:45:02 * koen_ watches fsck crawl Oct 17 09:45:31 hey koen, XorA, hrw Oct 17 09:45:34 hey lrg Oct 17 09:45:36 hi liam Oct 17 09:45:48 your ASoC drivers are getting abused :-D Oct 17 09:47:50 XorA: any bugs ? Oct 17 09:50:28 lrg: nah, just morons Oct 17 10:04:12 cbrake_away: do you think one would be able to flash the jffs2 image from ARMmon if the image had the partition size ? Oct 17 10:17:04 * * OE Bug 3165 has been created by eha(AT)doredevelopment.dk Oct 17 10:17:06 * * Add kernel-vmlinux to blacklist Oct 17 10:17:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3165 Oct 17 10:19:33 koen: r2897 in poky can be interesting for you Oct 17 10:20:12 hrw: that's holger's thing :) Oct 17 10:21:24 koen: I wrote generator of checksums.ini into base.bbclass - need some testing and will land in Poky and OE Oct 17 10:28:19 03xora 07org.oe.dev * r706e9716... 10/ (1 packages/gtk-webcore/midori_0.0.10.bb): Oct 17 10:28:19 packages/gtk-webcore/midori_0.0.10.bb : version bump midori, this one Oct 17 10:28:19 actually builds but doesnt run. Oct 17 10:28:23 03henryk 07org.oe.dev * r3e2a1f68... 10/ (1 classes/mono.bbclass): mono.bbclass: Stage all .dll files that have been packaged so that other packages can compile against them Oct 17 10:28:32 03henryk 07org.oe.dev * reb0ba139... 10/ (1 packages/gtk-sharp/gtk-sharp_2.10.2.bb): gtk-sharp: Inherit pkconfig to stage .pc files, so other packages can compile against gtk-sharp Oct 17 10:28:39 03henryk 07org.oe.dev * raa3f03af... 10/ (1 packages/gsm/libgsmd_svn.bb): libgsmd: fix typo in last commit Oct 17 10:29:04 good morninf Oct 17 10:29:11 eh morning Oct 17 10:29:12 hey florian Oct 17 10:30:27 hey everyone ... I have an error at building libgpg: http://pastebin.ca/739620 Is this known ? Oct 17 10:30:29 koen: Looks like that strange SRCREV behaviour happens on my build boxes (running Debian) only. I copied the configuration to my workstation (Ubuntu) and it works. Oct 17 10:31:29 florian: sounds like a too-old svn installation Oct 17 10:31:40 Longfield: looks strange, maybe a full harddisk? Oct 17 10:32:13 koen: something like that, do you know offhand which one we need? Oct 17 10:32:29 afaik 1.5, but I'm not sure Oct 17 10:33:14 I have svn 1.4.4 Oct 17 10:33:26 koen: here it is 1.4.2 vs. 1.4.3 Oct 17 10:36:42 hi mickeyl Oct 17 10:37:47 good morning hrw Oct 17 10:37:53 hi guys Oct 17 10:39:47 florian: hum, no full hard disk ... Oct 17 10:40:35 Longfield: can you reproduce it if you rebuild this package? (--c rebuild) Oct 17 10:40:42 s/--/- Oct 17 10:41:00 florian: going to do a rebuild Oct 17 10:44:15 florian: strange, the rebuild fixed it ... Oct 17 10:44:23 thx Oct 17 10:44:48 I had had this error twice, but without a rebuild Oct 17 10:45:28 I remember I have seen something like this before: Doing a rebuild it works... but I wasn't able to find out what causes this. Oct 17 10:51:27 03hrw 07org.oe.dev * r821ed938... 10/ (1 classes/base.bbclass): Oct 17 10:51:27 base.bbclass: add checksums.ini generator (from Poky) Oct 17 10:51:27 If file is fetched via HTTP or FTP and we do not have its checksum in Oct 17 10:51:27 metadata (conf/checksums.ini) then we generate checksums into Oct 17 10:51:27 TMPDIR/checksums.ini file. Content of that file can be then added into Oct 17 10:51:28 metadata one. Oct 17 10:51:32 03hrw 07org.oe.dev * r2b27ec64... 10/ (1 conf/checksums.ini): checksums.ini: add some new entries created by generator built-in to base.bbclass (from Poky) Oct 17 10:52:33 who will run 'bitbake -cfetch -k world' to populate checksums.ini? Oct 17 10:53:29 hrw : Is this a good idea ? i mean the "self-generated" checksums ? Oct 17 10:54:36 steliosk: checksums.ini present in OE were created by similiar way Oct 17 10:56:41 hrw : yes but this was done once and i was under the impression that they were checked against some "known" source of valid checksums Oct 17 10:58:30 steliosk: they were checked on big amount of fetched OE sources Oct 17 10:59:10 steliosk: as there is nearly no one who adds checksums of new sources this way will maybe make more people contributing them Oct 17 11:03:00 steliosk: If any of us have soruces that don't match these checksums, we'll soon find out and be able to investigate Oct 17 11:05:22 ok i understand that, but i think that blind insertion of checksums defeats the original idea Oct 17 11:06:18 I'm assuming it does not do a blind update when the checksum changes though? Oct 17 11:06:19 mickeyl: I just fixed python-2.5-manifest, could you fix your script that generates it as well? Oct 17 11:06:45 Crofton|home: no - it generate another file Oct 17 11:06:47 koen: well, usually it should go the other way round, but ok Oct 17 11:07:00 mickeyl: manifest is easier to fix Oct 17 11:07:15 but leaves me with additional work Oct 17 11:07:31 what fix btw.? Oct 17 11:08:14 mickeyl: Switch to speaker if pressing the speaker button in the dialer Oct 17 11:08:38 stefan_schmidt: ah, good. Oct 17 11:08:46 03koen 07org.oe.dev * rede6b342... 10/ (1 packages/python/python-2.5-manifest.inc): Oct 17 11:08:46 python-2.5-manifest.inc: fix hardcodes Oct 17 11:08:46 /usr/bin -> ${bindir} Oct 17 11:08:46 /usr/lib -> ${libdir} Oct 17 11:08:46 /usr/include -> ${includedir} Oct 17 11:08:47 (my question was related to python manifest fix thoug ;)) Oct 17 11:08:56 http://feeds.feedburner.com/~r/boingboing/iBag/~3/170980949/woman-jailed-for-50.html - USA and their 'freedom' Oct 17 11:09:04 mickeyl: that fix Oct 17 11:09:06 grr, wrong channel and wrong question Oct 17 11:09:59 ah Oct 17 11:10:44 hrw: some people seem to be more "free" than others :-/ Oct 17 11:11:11 hrw: If I could read Polish, I'd look for Polish police stories :) Oct 17 11:11:32 * Crofton|home makes a note that meth bears some resemblance to cat pee .... Oct 17 11:14:28 rootfs_deb.bbclass | 32 ++++++++++++++++---------------- Oct 17 11:14:28 1 files changed, 16 insertions(+), 16 deletions(-) Oct 17 11:14:38 koen: reindented? Oct 17 11:15:26 no, removing hardcodes Oct 17 11:15:58 I mentioned at OEDEM our classes have tons of stinking hardcodes Oct 17 11:16:16 I fixed up a few the past weeks and now I finally commit them :) Oct 17 11:17:04 03koen 07org.oe.dev * r66fb6062... 10/ (1 classes/rootfs_ipk.bbclass): Oct 17 11:17:04 rootfs_ipkbbclass: fix hardcodes Oct 17 11:17:04 /usr/bin -> ${bindir} Oct 17 11:17:04 /usr/lib -> ${libdir} Oct 17 11:17:09 03koen 07org.oe.dev * r6e6a6414... 10/ (1 classes/rootfs_deb.bbclass): Oct 17 11:17:09 rootfs_deb.bbclass: fix hardcodes Oct 17 11:17:09 /usr/bin -> ${bindir} Oct 17 11:17:09 /usr/lib -> ${libdir} Oct 17 11:30:32 koen: Be slightly careful with those changes since ${bindir} is not always == /usr/bin (e.g. its not for native packages) Oct 17 11:33:26 03mickeyl 07org.oe.dev * r920beec0... 10/ (3 files in 3 dirs): alsa-state: (fic-gta01) limit PCM a bit and crank up Amplifier, so that we are less likely to distort in the PCM section Oct 17 11:53:36 linked in dudes, there is a question "What are the top ten things you would like to see Trolltech do next?" Oct 17 11:58:16 "make koen filthy rich" Oct 17 11:58:39 I think you need to explain that in terms that are not so obvious :) Oct 17 11:58:40 no idea Oct 17 11:59:11 continure supporting open source develoeprs :) Oct 17 11:59:35 praise the recent GPL'ing of something Oct 17 12:00:16 * Crofton|home needs a picture of him that does not look like he has been drinking for days .... Oct 17 12:16:11 ~curse linux sata support Oct 17 12:16:12 May the fleas of a thousand camels infest your most sensitive regions, linux sata support ! Oct 17 12:16:35 someone know how to re-enable sata cdrom without poweroff/poweron? Oct 17 12:16:50 hdparm should be able to do that Oct 17 12:17:25 oxo: hdparm require /dev/ entry Oct 17 12:17:40 [53646.193870] ata6: SATA link down (SStatus 1 SControl 310) Oct 17 12:17:47 [53649.792333] ata6.00: disabled Oct 17 12:17:51 hmm, nasty Oct 17 12:17:57 and /dev/cdrom7 disappeared Oct 17 12:18:16 you could try re-scanning the scsi bus, but I fear that it won't work Oct 17 12:25:00 oxo: how? Oct 17 12:30:21 we need to figure out the openmoko sdk so we can have a press release b4 this one: Oct 17 12:30:26 http://www.businessweek.com/technology/content/oct2007/tc20071015_258257.htm?chan=top+news_top+news+index_businessweek+exclusives Oct 17 12:42:37 hmm, what is the general feeling towards https SRC_URIs? Oct 17 12:46:21 Henryk: avoid if possible? avoid if use self-signed or cacert certs? Oct 17 12:46:53 http://blog.haerwu.biz/2007/10/17/someone-know-how-to-re-enable-serial-ata-port/ Oct 17 12:47:13 f.e. packages/fontconfig/fontconfig_2.4.1.bb uses https://stage.maemo.org/svn/maemo/projects/haf/trunk/fontconfig/device_symbols.h Oct 17 12:47:25 which then gets somehow mapped to the infamous https://www.angstrom-distribution.org/unstable/sources/device_symbols.h Oct 17 12:48:02 this is bug in source stash mirror handling Oct 17 12:49:15 yes, one bug and one policy issue. (I can't fetch https://stage.maemo.org/... either because wget doesn't know the cert Oct 17 12:52:04 Henryk: I think that device_symbols.h file can be just put in metadata Oct 17 12:58:15 03xora 07org.oe.dev * rb33bb5b6... 10/ (1 packages/uboot/uboot-openmoko_svn.bb): Oct 17 12:58:15 packages/uboot/uboot-openmoko_svn.bb : same idirafter fix as rest of uboot Oct 17 12:58:15 recipes Oct 17 12:59:54 hmm, bitake lib/bb/fetch/__init__.py uri_replace() uses re.match to match the parts of the uri. this way "http" matches on "https". i think we want to change the PREMIRRORS and MIRRORS patterns to only match http. Oct 17 13:00:36 E.g. in angstrom-mirrors.bbclass change "http://.*/.*/ http://www.angstrom-distribution.org/unstable/sources/" to "http$://.*/.*/ http://www.angstrom-distribution.org/unstable/sources/" Oct 17 13:00:44 similarly for most other patterns Oct 17 13:03:10 feel free to fix the angstrom bbclass Oct 17 13:05:21 can someone tell if SRCDATE is in UTC? Oct 17 13:05:22 ok Oct 17 13:06:20 mranostay_work: it default to midnight UTC afaik Oct 17 13:08:14 koen: hmm, still, there is a policy decision: should it read http$://.*/.*/ (e.g. don't use the mirror for https urls) or should it read https?$://.*/.*/ (e.g. use the mirror for https urls; this time really working) Oct 17 13:08:41 hmm, I'll just use the latter Oct 17 13:08:55 it should use the mirror for all sources, ending up with http://angstrom... Oct 17 13:09:10 ok Oct 17 13:09:56 koen: kinda figured that, thanks Oct 17 13:13:10 gnarg. monotone is stupid. even if using ssh-agent it really wants to have a file in ~/.monotone/keys/ Oct 17 13:13:42 Henryk: Avoid where possible Oct 17 13:14:01 RP: what? Oct 17 13:14:05 Henryk: sorry, ignore me, the window stopped scrolling :/ Oct 17 13:14:14 ah, ok :) Oct 17 13:14:34 does OE actually use the SVN tarballs it creates on a rebuild? Oct 17 13:14:56 mranostay_work: it does Oct 17 13:14:59 mranostay_work: it should, yes Oct 17 13:23:52 koen: pushed Oct 17 13:56:31 03henryk 07org.oe.dev * r5da7fc50... 10/ (1 classes/angstrom-mirrors.bbclass): Oct 17 13:56:31 classes/angstrom-mirrors.bbclass: fix mirror regex Oct 17 13:56:31 * fixes the case where original https urls would end up as Oct 17 13:56:31 https://www.angstrom-distribution... for mirror url Oct 17 13:56:39 03henryk 07org.oe.dev * r0c8907e0... 10/ (1 conf/distro/openmoko.conf): conf/distro/openmoko.conf: Fix regex for the https case Oct 17 13:56:47 03koen 07org.oe.dev * r908d37bd... 10/ (1 packages/gpe-conf/gpe-conf_0.2.6.bb): gpe-conf: fix RDEPENDS Oct 17 13:56:58 and gtk theme gurus around? Oct 17 13:57:23 why does moko-gtk-theme look like this: http://scap.linuxtogo.org/files/8165d82ed2abfcaf8f92ac52f09eaa61.png ? Oct 17 13:59:15 morning all - what's the first X server I should try, just for proof of concept (i.e. easiest :-) Oct 17 13:59:29 HopsNBarley_: kdrive Oct 17 14:00:03 koen, as in xserver-common-kdrive? Oct 17 14:00:24 as in xserver-kdrive_.bb Oct 17 14:01:04 morning Oct 17 14:01:04 koen, danke! Oct 17 14:26:12 ok, meta-toolchain now starts to love me. Oct 17 14:26:17 ~praise RP Oct 17 14:26:18 All hail RP! Oct 17 14:27:32 mickeyl: OM bugzilla lacks 2007.2 entries Oct 17 14:28:16 latest moko-gtk-engine on a780: http://scap.linuxtogo.org/files/c2eb66c1d4e64363c9350a40b8cf196f.png Oct 17 14:28:18 in particular? Oct 17 14:28:49 hmm that doesn't look right Oct 17 14:28:51 libmokoui2, theme-2, etc Oct 17 14:28:54 where are the toolbar button gradients Oct 17 14:29:01 ah, i see Oct 17 14:29:14 about it not looking right: http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=945 Oct 17 14:29:23 well, since OM 2007.0 is completely abandoned we chose to just reuse the existing categories. Oct 17 14:30:19 so, openmoko-libs should still be used for libmoko*2 Oct 17 14:30:24 and theming for the theme Oct 17 14:35:40 mickey|packing: going to .tw? Oct 17 14:35:51 koen: ya, tomorrow Oct 17 14:37:03 my presence is required... Oct 17 14:38:24 mickey|packing, sounds "exciting" Oct 17 14:38:35 mmh Oct 17 14:39:00 i'm afraid it's more about internal problems than about further developments Oct 17 14:39:08 but lets hope... Oct 17 14:39:35 * Crofton avoided a meeting today for that sort of reason Oct 17 14:40:02 * Crofton curses TI docs on setting DDR2 controller for davinci Oct 17 14:50:37 * hrw curse gstreamer Oct 17 14:52:09 does anyone here have experience with blob? Oct 17 14:52:28 i'm trying to make it boot automatically from a tftp server Oct 17 14:52:38 and having trouble with the "automatically" bit Oct 17 14:52:49 or finding any useful documentation, for that matter Oct 17 14:54:46 iirc many people replace blob with redboot or uboot then use it Oct 17 14:55:07 joern: look into lakml archives - rmk was asking about blob about month ago Oct 17 14:55:12 iirc Oct 17 14:56:44 03mickeyl 07org.oe.dev * refa6e468... 10/ (1 packages/meta/meta-toolchain.bb): meta-toolchain: softlink libgcc_s.so until libgcc_s-dev gets built Oct 17 14:56:57 hrw: what (or where) are the lakml archives? Oct 17 14:57:03 linux arm kernel mailing list? Oct 17 14:57:06 yes Oct 17 14:57:09 ~lakml Oct 17 14:57:10 rumour has it, lakml is Linux-ARM-Kernel Mailing List: http://marc.theaimsgroup.com/?l=linux-arm-kernel Oct 17 14:57:23 thanks Oct 17 14:58:13 so general wisdom about how to use blob is: don't :) Oct 17 15:00:05 03mickeyl 07org.oe.dev * rbe08cf05... 10/ (3 files in 3 dirs): Oct 17 15:00:05 add {meta|task}-toolchain-openmoko{-sdk} based on meta-toolchain Oct 17 15:00:05 This builds a nice standalone OpenMoko toolchain now. Kudos to RP. Oct 17 15:00:49 mickey|packing, write a howto on using an external tool chain on the plane :) Oct 17 15:04:52 ;) Oct 17 15:04:56 will try Oct 17 15:15:05 anybody using directfb have a moment? Oct 17 15:17:44 Crofton , mickey|packing : Just do it (tm) :) Oct 17 15:32:20 hmm, compiling from scratch with BB_NUMBER_THREADS="4" finds a nice number of missing DEPENDS Oct 17 15:33:32 such as? Oct 17 15:34:20 just in this moment it's "Task failed: /oe/build/tmp/work/x86_64-linux/uicmoc4-native-4.3.1-r0/temp/log.do_compile.18090" cannot find -lz Oct 17 15:35:19 oh, hmm, that's a -native package. so it would need -lz in the host system, right? Oct 17 15:36:56 so it need zlib-native Oct 17 15:37:43 Henryk : When did you update the metadata ? XorA pushed some changes to this packages Oct 17 15:38:28 steliosk: about 4 hours ago Oct 17 15:38:48 as in: this host was completely formatted and set up 6 hours ago Oct 17 15:55:22 hrw: no, zlib-native apparently was already built and staged. Instead it seems to run the host g++ and require an -lz in the host. Oct 17 15:56:47 cute, the new nokia tablet has a keyboard Oct 17 15:57:47 koen: I am curious how many internals changed.. Oct 17 15:59:01 idem here Oct 17 15:59:12 http://pastebin.ca/739931 i didn't notice the call to "/usr/bin/ld" at first Oct 17 15:59:45 is this supposed to be so, e.g. simply add libz to the list of packages that need to be installed on the host? Oct 17 16:00:07 no idea about the internals Oct 17 16:00:13 * koen will not buy one Oct 17 16:00:20 (unless I get it for free again) Oct 17 16:00:21 HopsNBarley: check commits two weeks ago, some guy was actively pushing directfb changes (packages/cairo) etc. Oct 17 16:00:28 HopsNBarley: Would like to use it someday Oct 17 16:00:47 hrw, keesj : just download the schematics :-) Oct 17 16:01:17 I am a bit borred with the nokia, but I was really not impressed by openmoko Oct 17 16:01:42 likewise_: I do not have to. Oct 17 16:01:43 I love the jtag and openness of openmoko Oct 17 16:01:52 likewise_: but i do not have them neither Oct 17 16:03:55 I was at t-dose on saterday, I hoped to meet you guys Oct 17 16:04:46 gcc-cross depends on glibc - right? Oct 17 16:06:40 hrw: glibc-intermediate ? Oct 17 16:07:40 * florian votes for glibc-intermediate Oct 17 16:08:40 ok Oct 17 16:09:34 NOTE: Couldn't find shared library provider for libc.so.6 Oct 17 16:09:34 NOTE: package gcc-cross-4.1.2-r4: task do_package: completed Oct 17 16:09:40 thats why I ask Oct 17 16:10:24 Has it always done that or has it just started? Oct 17 16:11:28 I just noticed it in my new build Oct 17 16:11:56 hrw: I'd check and see if the poky autobuilder shows it Oct 17 16:13:32 * florian shoots usbtouchscreen.ko Oct 17 16:16:33 RP: it did it in yesterday build Oct 17 16:30:11 hrw: what is your friend's name that works with the sarge? can you email me his address as I wanted to ask him about i2s and the 9200 since we are having issues right now (creder@digitalcpt.com) Oct 17 16:30:50 christopher: do you use jabber? Oct 17 16:31:02 hrw: I don't but I can get it. Oct 17 16:32:15 hrw: it looked like he was going to put in some i2s dac support. I have some in my build but if the clocks aren't just right, I don't get any sound out. and I am not sure what i am doing wrong Oct 17 16:32:33 christopher: contact: mcgregor at blackmesaeast.com.pl Oct 17 16:36:16 hrw: ok, thanks. I'll email him and see if he has worked on it. Oct 17 16:36:41 hrw: I get some sound, but it is garbly and makes me think the clocks are off but I don't know that they can ever be perfect Oct 17 16:36:52 hrw: due to the divide down from the plla clock. Oct 17 16:37:13 RP: SDK build fails on x86/Angstrom -> http://rafb.net/p/jSn9MP28.html Oct 17 16:37:48 RP: There is a chance that icecream might be also playing some role in the breakage Oct 17 16:41:37 hrw: I just added a em-x270 patch from compulab. Currently, the kernel is a separate package. Is it pretty much OE policy that kernel recipes be merged to linux_XX if possible? Oct 17 16:42:35 cbrake: if possible use them. it is non-written-yet policy Oct 17 16:43:16 hrw: ok, I'll merge the recipes at some point then. Its a pretty simple recipe. Oct 17 16:44:08 exactly Oct 17 16:44:21 if it's seperate it should use linux.inc to avoid duplication Oct 17 16:44:45 koen: it already uses linux.inc, so they got that right Oct 17 16:44:52 maybe we should move linux.inc to kernel.bbclass Oct 17 16:45:08 hmmm Oct 17 16:45:18 there are patches in linux,inc Oct 17 16:45:35 I think ... Oct 17 16:45:50 * Crofton need to stop doing two things at once Oct 17 16:46:04 there aren't any patches in it Oct 17 16:46:21 koen: that would defintely get the point accross a little more ... Oct 17 16:46:51 its nice to see CPU module mfgrs starting to get OE -- at least one anyway ... Oct 17 16:46:52 yes, I figured it out, patches in version specific bb file Oct 17 16:47:10 on arm linux.inc unbreaks a lot of defconfigs Oct 17 16:47:10 koen: many parts of linux.inc and linux-rp.inc should end in kernel.bbclass Oct 17 16:47:31 hrw: RP and I planned that long ago, but we both didn't implement it :) Oct 17 16:48:14 ~curse java from sun Oct 17 16:48:15 May the fleas of a thousand camels infest your most sensitive regions, java from sun ! Oct 17 16:48:40 ah - the always popular Max Klinger curse Oct 17 16:49:04 * * OE Bug 3167 has been created by mike(AT)compulab.co.il Oct 17 16:49:06 * * Add machine configuration for EM-X270 Oct 17 16:49:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3167 Oct 17 16:49:34 cute, by the kernel hacker himself Oct 17 16:49:52 * koen ponders about compulab donations Oct 17 16:50:49 hrw: MACHINE_FEATURES = "thumb", yes or no? Oct 17 16:50:56 ehm Oct 17 16:51:01 koen: arm-thumb you mean? Oct 17 16:51:06 COMBINED_FEATURES Oct 17 16:51:16 afaik there is only one thumb Oct 17 16:51:22 like iwmmxt Oct 17 16:51:35 koen: and all arms newer then arm4 supports it... Oct 17 16:51:53 armv4 Oct 17 16:52:45 my goal is to be able to flag certain package (e.g. samba) to build with thumb like we do with iwmmxt now Oct 17 16:53:12 time for COMPILE_FEATURES maybe? Oct 17 16:53:23 COMPILE_FEATURES = "thumb iwmmxt" Oct 17 16:53:31 or ARCH_FEATURES Oct 17 16:53:50 that starts to look an awfull lot like use flags, though Oct 17 16:55:33 * CosmicPenguin coughs Oct 17 16:56:23 03cbrake 07org.oe.dev * r6d41cbbc... 10/ (6 files in 4 dirs): Oct 17 16:56:23 em-x270: add machine/kernel support (from compulab) Oct 17 16:56:23 em-x270 kernel recipe will likely be merged with linux_2.6.23 shortly ... Oct 17 16:56:24 koen: but thumb and iwmmxt are not part of machine but their (sub)architecture rather Oct 17 16:56:32 right Oct 17 16:56:39 I emailed the list Oct 17 16:56:54 http://lists.linuxtogo.org/pipermail/openembedded-devel/2007-October/003275.html Oct 17 16:57:12 bbl Oct 17 16:57:59 holy crap Oct 17 16:58:01 it boots Oct 17 16:58:06 I am a guru Oct 17 16:58:21 need to disable ide .... Oct 17 16:58:33 Crofton: don't be so elitist Oct 17 16:58:36 RP: Without icecc it builds but fails Q&A -> http://rafb.net/p/Rfw2JQ35.html Oct 17 16:58:47 :) Oct 17 16:58:54 allow me this moment in time Oct 17 16:59:16 I started from 0 Linux support for the board Oct 17 16:59:17 gah, one of the 'mickey|packing builds without proper QA' things again Oct 17 17:00:15 Crofton: good job Oct 17 17:00:36 now to play with kernel config and see how far i can get Oct 17 17:00:48 * steliosk thinks its time to add bugs to bugzilla Oct 17 17:05:03 likewise, yeah, i saw the upgrade to 1.1 - thanks for the tip. I'll let you know what i learn (-; Oct 17 17:19:04 * * OE Bug 3168 has been created by autobuild(AT)openembedded.org Oct 17 17:19:06 * * dbench-3.04-r1-do_install Oct 17 17:19:07 bye Oct 17 17:19:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3168 Oct 17 17:28:45 03pH5 07org.oe.dev * r1065bf1f... 10/ (1 conf/machine/magician.conf): magician.conf: use xserver-kdrive as virtual/xserver, add display info Oct 17 17:50:41 Crofton: Board Bring Up Phase 1 Succeeded :-) Oct 17 17:50:58 Crofton: Good Job. this is for the davinci dvm? Oct 17 17:51:32 Crofton: we probably won't build products from that board, but if I can lift along with any developments, I have a nice toy :-) Oct 17 17:53:05 * * OE Bug 3169 has been created by skoroneos(AT)digital-opsis.com Oct 17 17:53:06 * * meta-toolchain fails to build when icecc is used Oct 17 17:53:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3169 Oct 17 17:53:16 * * OE Bug 3170 has been created by skoroneos(AT)digital-opsis.com Oct 17 17:53:18 * * meta-toolchain fails at binutils-cross-sdk fails on x86/ Angstrom with QA errors Oct 17 17:53:20 * * http://bugs.openembedded.org/show_bug.cgi?id=3170 Oct 17 18:06:48 likewise, lyrtech SFFSDR Oct 17 18:07:07 basically, cribbing from the big block NAND DV board in u-boot works Oct 17 18:07:07 Crofton: even better :-) Oct 17 18:19:00 ok disabling ide and see if that get mes further Oct 17 18:26:22 Crofton: the autobuilder is churning on meta-toolchain now, so if that works out you can hand that to your student Oct 17 18:27:47 mythical student at this point Oct 17 18:27:49 but Oct 17 18:28:00 I can give it a try Oct 17 18:28:05 crap Oct 17 18:28:30 trying to rebuild dv kernel atm Oct 17 18:29:44 do we have a way for linux.inc to die gracefilly if you try to build it with a machine it does not support? Oct 17 18:39:40 http://rafb.net/p/2FTh6a75.html Oct 17 18:39:52 anyway , I really want to see how openmoko will progress! Oct 17 18:39:52 does this mean anything to anyone, kernel build failling Oct 17 18:41:06 03koen 07org.oe.dev * r994d1b5f... 10/ (1 classes/sdk.bbclass): sdk.bbclass: package more debug files Oct 17 18:41:10 03mickeyl 07org.oe.dev * r4bcc58df... 10/ (1 packages/uboot/uboot-openmoko_svn.bb): uboot-openmoko svn build variants for gta02v3 and gta02v4 Oct 17 18:46:16 re Oct 17 19:11:04 * * OE Bug 3170 has been RESOLVED (FIXED) by Oct 17 19:11:06 * *  meta-toolchain fails at binutils-cross-sdk fails on x86/ Angstrom with QA errors Oct 17 19:11:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3170 Oct 17 19:16:03 how do I let bitbake put (the files of) a package in ... such that other bitbake recipes can find it? Oct 17 19:20:45 do_staging Oct 17 19:25:41 for the cryptic error that I get... "staging" not defined? Do I need to inherit something? Oct 17 19:27:35 there's no default staging rules Oct 17 19:27:51 you copy out what you need to STAGING_INCDIR/LIBDIR Oct 17 19:34:04 seems primitive. I want (cairo) bindings for ruby find ruby itself. Oct 17 19:34:45 I agree, maybe someone else knows a better way Oct 17 19:35:06 So I want the ruby1.8 and ruby1.8-dev ipkgs (I think, or hope, or I will make them such) in those STAGING_*DIRs Oct 17 19:35:25 Well, you want the contents of them copied Oct 17 19:35:31 check glib-2.0 as an example Oct 17 19:35:38 koen : NOTE: package gcc-cross-sdk-4.1.2: completed :) Oct 17 19:39:03 that's likely a well-written bbfile Oct 17 19:39:13 what's the diff between oe_install and install? Oct 17 19:44:13 do_stage { Oct 17 19:44:18 autotools_stage_all Oct 17 19:44:18 } Oct 17 19:46:22 Crofton: http://ewi546.ewi.utwente.nl/tmp/angstrom-2007.9-test-20071017-arm-glibc-toolchain.tar.bz2 Oct 17 19:49:32 I was trying to add powerpc toolchain support for the e300c3 core to OE this week, but I think I pushed my luck. Oct 17 19:51:04 libmudflag issues Oct 17 19:51:12 flap even Oct 17 19:52:07 03koen 07org.oe.dev * rdd977588... 10/ (1 packages/meta/meta-toolchain.bb): meta-toolchain: include target os to distinguish between (e)glibc and uclibc Oct 17 19:58:03 for some reason, it does not see 'stage'. I'm probably doing something stupid... Oct 17 20:03:19 running bitbake -c stage ruby; it complains listid = taskData.tasks_lookup[fnid][target[1]]\n KeyError: 'do_staging' Oct 17 20:04:09 likewise: you can disable libmudflap Oct 17 20:05:18 staging is the name, iirc Oct 17 20:05:39 khem: hi. thanks, but I assume it's my fault because we already have a powerpc toolchain targetting 603e. I only half know what I'm doing, so I need to dig deeper first. Oct 17 20:05:58 (do_)stage from glib 2 Oct 17 20:06:02 khem: e300c3 a successor of 603e, minor changes. Oct 17 20:06:24 likewise,ok --disable-libmudflap to gcc configure should get rid of it Oct 17 20:06:32 either way, if I define do_stage() {} like I do with do_install() {} in other bbfiles... shouldn't really matter Oct 17 20:08:01 likewise, then I would suggest to keep 6003e toolchain even for e300c3 if there are no specific differences in instruction set Oct 17 20:09:14 khem: there are some changes, one register extra (needs a small binutils patch, not sure if that's upstream yet), and a GCC machine description to add new mul/div assembly instructions. Not sure if I *need* them... Oct 17 20:10:12 khem: Which I took from the Freescale-created toolchain, which is at http://www.bitshrine.org/gpp/tc-fsl-x86lnx-e300c3-nptl-4.0.2-2.src.rpm Oct 17 20:10:14 likewise: ok Oct 17 20:10:38 khem: I'll retry tomorrow more step-by-step wise. Oct 17 20:10:44 alrighty Oct 17 20:14:32 lsittasks does not list 'stage', either. Oct 17 20:17:08 Kero: use the do_stage method I posted Oct 17 20:23:46 koen, how can I use toolchain meta package ? Oct 17 20:24:09 to get it build Oct 17 20:25:45 bitbake -b ? Oct 17 20:28:30 'bitbake meta-toolchain' Oct 17 20:28:32 03cbrake 07org.oe.dev * r39f14a79... 10/ (1 conf/machine/em-x270.conf): em-x270.conf: remove redundant "arm" in PACKAGE_EXTRA_ARCHS Oct 17 20:28:37 03koen 07org.oe.dev * rff99dd15... 10/ (1 packages/tasks/task-sdk.bb): task-sdk: fix uclibc Oct 17 20:28:39 but it doesn't work for eglibc yet Oct 17 20:28:46 ah that's why Oct 17 20:28:55 I was hitting error Oct 17 20:31:44 physmap platform flash device: 01000000 at 02000000 Oct 17 20:31:44 physmap-flash physmap-flash.0: map_probe failed Oct 17 20:31:49 any suggestions? Oct 17 20:31:56 NAND flash Oct 17 20:32:52 koen: I do. added to the ruby_1.8.6-p110.bb file, which listtasks explicitely mentions as the file it is using. So I'm still doing something stupid. Oct 17 20:33:22 liststasks should report populate_staging Oct 17 20:33:31 it does Oct 17 20:33:45 right Oct 17 20:33:51 and do_stage is a wrapper around that Oct 17 20:34:02 khem: http://lists.linuxtogo.org/pipermail/openembedded-devel/2007-October/003281.html Oct 17 20:35:05 I see so eglibc-dev is needed Oct 17 20:35:25 a wrapper? now you got me confused. if it contains autotools_stage_all, does that wrap populate_staging ? Oct 17 20:36:01 no, do_stage() wraps do_populate_staging() Oct 17 20:36:13 so stage doesn't appear in listtasks Oct 17 20:36:54 one say, someone needs to explain what USE_FLAGS are and why they are a bad thing to us know genttooists Oct 17 20:37:17 so I do not run do `bitbake -c stage ruby` Oct 17 20:37:20 not right now though Oct 17 20:37:45 Crofton: use flags affect the build (-OMG -funrollloops), but aren't visible when packaging Oct 17 20:38:07 foo_1.2.3.ipk and foo_1.2.3.ipk are something completely different Oct 17 20:38:14 * Crofton is very weak at packaing concpets Oct 17 20:38:18 one might include ssl support, the other dbus support Oct 17 20:38:27 but I see the point Oct 17 20:38:31 but you can't see that till you acutally run the app Oct 17 20:38:38 * Crofton needs to do nfs root atm Oct 17 20:39:08 Kero: right you run -c populate_staging -f Oct 17 20:39:41 "md5sum libX11-1.1.2.tar.bz2" - can someone? Oct 17 20:39:43 and rebuild runs populate_staging, too Oct 17 20:40:14 rebuild runs all tasks in the depchain Oct 17 20:40:43 710bf38a9477a5a1b235bc94f1d0593c /home/koen/OE/downloads/libX11-1.1.2.tar.bz2 Oct 17 20:40:59 thx Oct 17 20:41:41 hrw|gone: your change to base.bbclass has a flaw: Oct 17 20:41:43 koen@bitbake:/data/build/koen/OE/build/tmp/angstrom$ grep binutils checksums.ini Oct 17 20:41:43 [ftp://ftp.gnu.org/gnu/binutils/binutils-2.18.tar.bz2] Oct 17 20:41:44 [ftp://ftp.gnu.org/gnu/binutils/binutils-2.18.tar.bz2] Oct 17 20:42:09 it doesn't check if the checksum is already in the temp checksums Oct 17 20:42:13 I still do not understand what you mean with "stage wraps populate_staging", I interpret that as, when running populate_staging, it does not see stage. Oct 17 20:43:23 Kero: see base.bbclass Oct 17 20:43:53 koen: I know Oct 17 20:44:26 koen: it is left to user how to manage it Oct 17 20:44:54 koen: I'd phrase that as "populate_staging wraps stage" Oct 17 20:46:36 and now I do see the proper temp/log.do_stage*, too (so I know what autotools_stage_all does, too) Oct 17 20:46:49 wow: http://www.cato.org/raidmap/ Oct 17 20:46:52 bye Oct 17 20:49:08 koen: urgs Oct 17 20:53:04 * * OE Bug 3167 has been RESOLVED (FIXED) by cliff.brake(AT)gmail.com Oct 17 20:53:06 * *  Add machine configuration for EM-X270 Oct 17 20:53:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3167 Oct 17 20:53:16 * * OE Bug 3171 has been created by autobuild(AT)openembedded.org Oct 17 20:53:18 * * gcc-cross-sdk-4.1.2-r6-do_unpack Oct 17 20:53:20 * * http://bugs.openembedded.org/show_bug.cgi?id=3171 Oct 17 20:53:28 * * OE Bug 3172 has been created by autobuild(AT)openembedded.org Oct 17 20:53:30 * * uclibc-0.9.29-r8-do_configure Oct 17 20:53:32 * * http://bugs.openembedded.org/show_bug.cgi?id=3172 Oct 17 20:57:27 Now we get to the hard part where staging misses the "usr" part in its directory structure (which ruby extconf.rb expects) Oct 17 21:02:45 koen, it does not work for eglibc because it shared same triplet as glibc Oct 17 21:03:13 LIBC_linux-gnueabi = "glibc" could be LIBC_linux-gnueabi = "eglibc" too Oct 17 22:30:05 my oe refuses to make any images (image bitbake tasks fail on do_rootfs). checking the logs it says no space left Oct 17 22:30:50 does that mean that there is no space on the target machine according to its machine.conf or does that reffer to the diskspace of the build machine? Oct 17 22:34:48 Well, was that disk full / nearly full? Oct 17 22:44:50 Tartarus: not at all Oct 17 22:45:21 197G free Oct 17 22:46:29 I think that should suffice for an angstrom image Oct 17 22:58:02 dcordes: quota enabled? Oct 17 23:08:05 likewise: sorry, what's that? Oct 17 23:08:58 dcordes: another reason for no space left *might* be over quota. check with quota -a if I'm not mistaken. Oct 17 23:09:17 dcordes: but it sounds as if you don't use quota. Oct 17 23:10:04 likewise: no, that is not present on my system. Could you take a look at my logs if I will paste them? Oct 17 23:11:31 dcordes: sure Oct 17 23:11:38 ok Oct 17 23:13:42 likewise: ah could it be that the reason is I removed the deploy../images directory? Oct 17 23:14:00 dcordes: no should not be an issue Oct 17 23:14:44 because the do_rootfs log says something about cannot stat blabla.jffs2 although I'm not building jffs2 Oct 17 23:15:19 dcordes: what does df tell you? Oct 17 23:15:20 likewise: http://www.pastebin.ca/740487 Oct 17 23:15:59 that there is hundrets of gigabytes space on each partition Oct 17 23:16:21 /dev/sda4 415G 219G 197G 53% /media/archiv Oct 17 23:16:25 that's the OE one Oct 17 23:16:54 dcordes: I do not see the error you mentioned. However, it looks like the jffs2 copy action is just a POSTPROCESS command. Oct 17 23:17:12 that is from the error log Oct 17 23:17:47 http://www.pastebin.ca/740490 Oct 17 23:21:24 dcordes: this is what you hit: Oct 17 23:21:26 conf/machine/include/zaurus-2.6.inc:44:MACHINE_POSTPROCESS_COMMAND = "zaurus_make_installkit" Oct 17 23:21:27 conf/machine/include/zaurus-2.6.inc:46:zaurus_make_installkit () { Oct 17 23:21:29 ... Oct 17 23:22:56 likewise: hm I know what the installkits are but what went wrong with it? Oct 17 23:23:27 dcordes: could you do a "find build/tmp/staging | grep jffs2"? Oct 17 23:23:54 dcordes: and see if there is a mkfs.jffs2? Oct 17 23:26:59 gerwinin: hee hallo Oct 17 23:28:28 likewise: hey Oct 17 23:28:42 likewise missed you on t-dose Oct 17 23:28:44 likewise: build/tmp/openmoko/staging/i686-linux/bin/mkfs.jffs2 Oct 17 23:28:58 likewise: all kind of people asking difficult questions about oe Oct 17 23:29:03 likewise Oct 17 23:29:30 gerwinin: My weekend was *way* too busy after a full week. I'm really sorry I missed it though. Oct 17 23:29:35 likewise: pff working on some imx27 software Oct 17 23:29:52 gerwinin: cool Oct 17 23:30:05 likewise: no problem Oct 17 23:30:16 likewise: I want to make a cheap arm board Oct 17 23:30:23 gerwinin: lpc based? Oct 17 23:31:08 likewise: didn't make a choice yet Oct 17 23:31:22 likewise: selected a touchscreen now (really cheap one ) Oct 17 23:31:50 likewise: need to find somebody with experience with bga's Oct 17 23:33:36 gerwinin: hmm, no sorry, we have populated our first BGA 10 years back. The first two fell off (3000 EUR each), and the third one was ok. After that, we have the boards populated at the manufacturer :-) Oct 17 23:33:57 likewise: I ment more the design of the pcb Oct 17 23:34:33 gerwinin: ah ok, you mean the design rules, or just drawing the pcb? Oct 17 23:34:44 likewise: sorry if I'm disturbing but is that mkfs.jffs2 thing ok? it is there.. Oct 17 23:35:27 likewise: the design of the pcb Oct 17 23:35:30 dcordes: yes, it's ok. I suspected that it would be missing... the zaurus_make_installkit seems to fail for you though Oct 17 23:36:06 likewise: I can design smaller pcb's myself but bga is new to me Oct 17 23:36:59 likewise: thing is that last week both, openmoko and angstrom images built fine for my zaurus. Is there a way to fall back to the "normal" images w/o installkit? Oct 17 23:37:19 Our PCB engineers are fluent with BGA designs, but they are kinda busy. BGA is simplest if you put via's under each pin, at least for your first design. Oct 17 23:37:40 likewise: if I make it can one of your engineers check it ? Oct 17 23:38:08 gerwinin: I can try - no promises though. In what format will you deliver? Oct 17 23:38:15 eagle Oct 17 23:38:25 hmm, we don't have that. Oct 17 23:38:36 Do you have orcad ? Oct 17 23:39:21 we use Altium Designer (or its predecessor). I'm not sure what formats we can accept. I haven't drawn PCB's for years now. Oct 17 23:40:26 gerwinin: If you want to have it manufactured, ask for the "pcb design rules" for bga's. Solder flow masks layers, correct footprint etc are important. Oct 17 23:42:11 likewise: Yes, i know, I want to make a cheap armboard which is available Oct 17 23:44:06 likewise: any idea how to get around this? Oct 17 23:45:18 dcordes: no. revert to an old checkout and see what is done there, or debug the piece that fails. Oct 17 23:45:47 dcordes: Sorry, I cannot help you further, I plan to go to sleep soon :-/ Oct 17 23:47:38 ok thanks and good night Oct 17 23:50:59 dcordes: thanks, good luck! Oct 17 23:51:03 cya all Oct 18 00:28:06 mickeyl's 'PACKAGE_ARCH_openmoko = "openmoko"' in revision 91b1890c469c39d7efcba63967f106bf6efd0c4f broke openmoko-devel-image for the openmoko distribution. what's that line supposed to do? Oct 18 02:02:16 03henryk 07org.oe.dev * rbd67f1f8... 10/ (1 packages/python/python-pygtk_2.10.4.bb): python-pygtk-2.10.4: Fix duplicate ${PN}-dev in PACKAGES **** ENDING LOGGING AT Thu Oct 18 02:59:57 2007