**** BEGIN LOGGING AT Wed Aug 26 02:59:57 2009 Aug 26 03:02:11 03Chris Larson  07org.openembedded.dev * r4dc61b382a 10openembedded.git/classes/srctree.bbclass: (log message trimmed) Aug 26 03:02:11 srctree.bbclass: add initial version. Aug 26 03:02:11 This enables operation inside of an existing source tree for a project, rather Aug 26 03:02:11 than using the fetch/unpack/patch idiom. Aug 26 03:02:11 By default, it expects that you're keeping the recipe(s) inside the Aug 26 03:02:13 aforementioned source tree, but you could override S to point at an external Aug 26 03:02:17 directory and place the recipes in a normal collection/overlay, if you so Aug 26 03:02:19 03Chris Larson  07org.openembedded.dev * r482e70f47f 10openembedded.git/classes/package_ipk.bbclass: Aug 26 03:02:22 package_ipk.bbclass: don't choke on empty SRC_URI. Aug 26 03:02:24 Signed-off-by: Chris Larson Aug 26 03:02:26 03Chris Larson  07org.openembedded.dev * re5b751653c 10openembedded.git/classes/gitver.bbclass: Aug 26 03:02:29 gitver.bbclass: add initial version. Aug 26 03:02:31 This provides a GITVER variable which is a (fairly) sane version, for use in Aug 26 03:02:33 ${PV}, extracted from the ${S} git checkout, assuming it is one. This is most Aug 26 03:02:35 useful in concert with srctree.bbclass. Aug 26 03:02:37 Signed-off-by: Chris Larson Aug 26 03:04:04 whee Aug 26 03:05:31 hmm, now that thats at least minimally functional, i think i should go back and finish re-implementing the private staging changes, though I'm still not really sure if that's the ideal approach to the problem.. Aug 26 04:02:20 goddamn, i forgot how much work the private staging crap was. hrmp Aug 26 05:54:09 I'm getting weird C++ linking poblems with libusb 0.1.12 Aug 26 06:03:48 hmm, using libusb-compat seems to fix the problem Aug 26 07:11:30 03Sebastian Krzyszkowiak  07shr/import * r119db4f8b8 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: add paroli Aug 26 07:11:31 03Sebastian Krzyszkowiak  07shr/import * r0b394839d1 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: add abiword Aug 26 07:11:32 03Sebastian Krzyszkowiak  07shr/import * r1ac28449b4 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: add cacao Aug 26 07:17:59 gm Aug 26 07:18:55 gm likewise_ Aug 26 07:26:38 hi khem, you up already?? Aug 26 07:27:27 ah wait, never mind, I'm sleepy myself still. :-/ Aug 26 07:31:37 likewise_: I havent slept yet :) Aug 26 07:31:58 its half past midnight here Aug 26 07:35:07 khem: I have always wondered; you use OE for work? Aug 26 07:35:49 likewise_: no :) Aug 26 07:36:29 its for fun Aug 26 07:38:00 although I would have preferred to do it at work then I dont have to work extra hours :) Aug 26 07:39:41 03Khem Raj  07org.openembedded.dev * rf73c64e729 10openembedded.git/classes/base.bbclass: (log message trimmed) Aug 26 07:39:41 base.bbclass: Replace os.system with subprocess call. Aug 26 07:39:41 Often gzip is reporting broken pipe errors with do_unpack of Aug 26 07:39:41 tar.gz files. Aug 26 07:39:41 If you use the commands described above to extract a tar.gz file, gzip Aug 26 07:39:43 sometimes emits a Broken pipe error message. This can safely be ignored Aug 26 07:39:47 if tar extracted all files without any other error message. Aug 26 07:39:49 03Khem Raj  07org.openembedded.dev * rddc8e9752d 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 26 07:39:52 03Khem Raj  07org.openembedded.dev * r224b698257 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 26 07:39:55 03Khem Raj  07org.openembedded.dev * raa34eaf616 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 26 07:39:58 03Khem Raj  07org.openembedded.dev * rc0be71ee07 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 26 07:40:01 03Khem Raj  07org.openembedded.dev * raf56cd467b 10openembedded.git/classes/base.bbclass: Aug 26 07:40:18 base.bbclass: Remove redundant import of subprocess and signal. Aug 26 07:40:18 Signed-off-by: Khem Raj Aug 26 08:32:05 how do i get tslib-calibrate in my image? i just can add tslib but this wont copy the calibrate stuff Aug 26 08:35:04 de_manuel_pi: this is in a separate package... Aug 26 08:35:38 de_manuel_pi: 'tslib-calibrate' iirc Aug 26 08:35:59 so i have to add it to IMAGE_INSTALL? Aug 26 08:37:58 de_manuel_pi: yes Aug 26 08:38:43 ok thx will try Aug 26 08:38:47 de_manuel_pi: if you define your own set of packages to install its a good idea to create your own task metapackage Aug 26 08:39:59 until now its just very little of packages, qt embedded and tslib Aug 26 08:40:18 my touchscreen drive me mad :) Aug 26 08:42:26 cool /usr/bin/ts_calibrate exists now Aug 26 08:46:26 :-) Aug 26 09:22:15 good morning Aug 26 09:24:21 hi mckoan Aug 26 09:29:57 florian: good morning Aug 26 09:31:16 hi pb__ Aug 26 09:32:48 hi lrg Aug 26 09:33:22 heh, wolfsonmicro.com?! Aug 26 09:33:46 oh Aug 26 09:37:42 any of you is going to CELF Genoble in October ? Aug 26 09:38:19 03Koen Kooi  07org.openembedded.dev * rbac5eef9c8 10openembedded.git/ (41 files in 5 dirs): dvsdk: merge in new versions from arago and clean them up a bit Aug 26 09:38:30 03Koen Kooi  07org.openembedded.dev * r172cb63ce5 10openembedded.git/recipes/dvsdk/ (68 files in 4 dirs): dvsdk: rename to ti now that other things like bitblit live in that dir now Aug 26 09:39:15 morgen (as I was in .de 1.5h ago) Aug 26 09:39:29 Kero: I have 1.8.6 somewhere Aug 26 09:39:33 03Koen Kooi  07org.openembedded.dev * r59ddf7ef9b 10openembedded.git/contrib/angstrom/sort.sh: angstrom feed sorter: skip base feeds if there aren't any packages to sort Aug 26 09:39:47 hrw: moin Aug 26 09:41:49 http://www.embeddedlinuxconference.com/elc_europe09/sessions.html Aug 26 09:46:09 mckoan: get mail Aug 26 09:51:53 hrw: yep thx Aug 26 09:53:27 mckoan: I do not know price of rooms there Aug 26 09:53:48 hrw: I'll take a look at Ibis (as usual) Aug 26 09:54:05 hrw: I have it, too, plus gtk2 bindings, as SHR uses gtk (well, and enlightenment, but I have no idea whether there is a ruby binding for that at all) Aug 26 09:54:52 wow..where is hosted linuxtogo? on a single i486sx? Aug 26 09:55:03 hrw: I changed a couple of things, biggest one being a split in more packages; and I have not dug into what you changed in the openssl extension, do you remember? Aug 26 10:00:18 Kero: no idea - it was years ago and I never used ruby Aug 26 10:01:02 hrw: OK, I'l figure it out some time. Aug 26 10:01:12 Kero: I have 1.8.6-p286 recipes Aug 26 10:01:48 hrw: are you OK if I submit 1.8.6-p383 recipes? Aug 26 10:02:08 Kero: can you check http://marcin.juszkiewicz.com.pl/download/oe/ first? Aug 26 10:02:19 and even so, I'd like to see what you updated for 1.8.6 and incorporate what you've done. Aug 26 10:02:30 I am ok with it - just would like to get my update compared Aug 26 10:02:39 certainly I'll check that. Aug 26 10:03:24 timeline-wise, I'm leaving for holiday on Friday, so end-of-Sep I hope to have things smoothed out for these packages. Aug 26 10:04:39 cool Aug 26 10:04:53 that'd also include finding out how I can run Ruby's own testsuite on the result :) Aug 26 10:22:20 03Florian Boor  07org.openembedded.dev * r3ecef33ae9 10openembedded.git/conf/machine/kixrp435.conf: kixrp435.conf: Machine description for the Intel KIXRP435 reference board. Aug 26 10:22:21 03Florian Boor  07org.openembedded.dev * racce2710fe 10openembedded.git/recipes/linux/ (linux-2.6.30/kixrp435/defconfig linux_2.6.30.bb): linux: Add support for the Intel KIXRP435 reference board to 2.6.30 Aug 26 10:22:22 03Florian Boor  07org.openembedded.dev * r7d2336c2e4 10openembedded.git/conf/machine/topas910.conf: topas910.conf: Some defines to build valid uImages. Aug 26 10:22:23 03Florian Boor  07org.openembedded.dev * rd7159b354e 10openembedded.git/conf/machine/topas910.conf: topas910.conf: No need for the lowlevel loader any more. But mtdutils are useful to deal with NAND. Aug 26 10:24:14 03Florian Boor  07org.openembedded.dev * r356c86ab3c 10openembedded.git/recipes/ixp4xx/ixp4xx-npe_2.4.bb: ixp4xx-npe: The KIXRP435 uses NPE too. Aug 26 10:25:32 Another funky piece of hardware for OE :) Aug 26 10:28:41 heh, very good Aug 26 10:29:38 Timesys seems to have quite some kernel patches for it... I should try to get these. Aug 26 10:30:37 I wonder will ST give permission to add nhk15 to OE Aug 26 10:30:49 Poky runs nice on it Aug 26 10:37:04 * * OE Bug 5305 has been created by chrisfriedt(AT)gmail.com Aug 26 10:37:06 * * x11-gpe-image do_rootfs opkg permission denied Aug 26 10:37:08 * * http://bugs.openembedded.org/show_bug.cgi?id=5305 Aug 26 10:43:18 lrg: wb Aug 26 11:38:05 * * OE Bug 5306 has been created by chrisfriedt(AT)gmail.com Aug 26 11:38:07 * * x11-gpe-image do_rootfs failed Aug 26 11:38:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5306 Aug 26 12:00:46 anyone played with the Karo triton pxa boads? Aug 26 12:14:30 hi Aug 26 12:14:49 I have put in my local.conf AND my image bake: Aug 26 12:14:51 PREFERRED_PROVIDER_virtual/xserver = "xserver-xorg" PREFERRED_PROVIDER_xserver = "xserver-xorg" Aug 26 12:15:08 but when I make my image I have this warning Aug 26 12:15:18 ERROR: Multiple .bb files are due to be built which each provide virtual/xserver (/home/koan/devel/openembedded/recipes/xorg-xserver/xserver-xorg_1.5.3.bb /home/koan/devel/openembedded/recipes/xorg-xserver/xserver-kdrive_1.3.0.0.bb). This usually means one provides something the other doesn't and should. Aug 26 12:15:24 ????????????????????? Aug 26 12:15:45 I put BBMASK = "xserver-kdrive" in my local.conf Aug 26 12:16:20 mmm Aug 26 12:17:31 I also have this in my machine config: http://pastebin.ca/1543105 Aug 26 12:17:44 in addition to the PREFERRED_PROVIDER_virtual/xserver like you Aug 26 12:17:59 I'm not sure which one did the trick, or both together Aug 26 12:21:29 I'll check first with BBMASK Aug 26 12:24:10 th1_: I have http://pastebin.ca/1543112 Aug 26 12:27:55 task-xterminal RDEPENDS/RRECOMMENDS kdrive! ugh! :-) Aug 26 12:52:02 sushisan, just use xterm instead of task-xterminal Aug 26 12:57:58 how can i set SRCREF to the latest revision of git archiv? Aug 26 13:04:49 de_manuel_pi, just don't set it Aug 26 13:05:29 set TAG = master instead and use SCR_URI="git://gitserver/repo.git;protocol=git;tag=${TAG}" Aug 26 13:05:52 then i get a: fatal: Not a valid object name 1 Aug 26 13:06:14 not if you set the SRC_URI like I said Aug 26 13:06:25 i want the latest version of the branch dev Aug 26 13:06:34 ok i try Aug 26 13:06:38 set TAG=dev and use that SRC_URI Aug 26 13:07:38 re Aug 26 13:07:50 morning hrw Aug 26 13:08:00 rkirti: how goes maemo/oe? Aug 26 13:08:30 hmm same problem it doesnt fetch the latest revision :( Aug 26 13:08:48 de_manuel_pi: SRCREV = "${AUTOREV}" Aug 26 13:09:33 hrw: maemo/oe is done as far as GSoC goes, right now lots of recipe cleaning needed. Aug 26 13:09:49 rkirti: where are they available to look? Aug 26 13:10:23 hrw: http://github.com/rkirti/maemo_angstrom/tree/master but it needs some updation Aug 26 13:11:15 hrw: towards the last phase, I built lots of things with my local OE copy, adding recipes here and there rather than having a clean overlay - so those things wont show up on github. I have to clean up my tree and add those to the overlay here Aug 26 13:11:54 so is tag the same as branch in the SRC_URI Aug 26 13:11:55 ? Aug 26 13:12:03 hrw: also -> http://www.cse.iitm.ac.in/~rkirti/maemo-oe/status.html Aug 26 13:14:03 * * OE Bug 5307 has been created by chrisfriedt(AT)gmail.com Aug 26 13:14:05 * * mesa 7.2 fails on do_install, patch for mesa-common.inc Aug 26 13:14:07 * * http://bugs.openembedded.org/show_bug.cgi?id=5307 Aug 26 13:16:53 rkirti: you used fremantle components? Aug 26 13:22:57 hrw: somewhat Aug 26 13:23:39 hrw: the exact fremantle hildon-desktop has closed source deps - so I used pre-alpha versions. I used whatever latest version was there which was open when I wrote those recipes ;) Aug 26 13:24:14 though fremantle has moved from alpha to beta since then Aug 26 13:31:37 03Michael Smith  07org.openembedded.dev * r22d0437dde 10openembedded.git/recipes/mesa/mesa-common.inc: Aug 26 13:31:37 mesa-common.inc: avoid bashism Aug 26 13:31:37 Closes bug #5307. Aug 26 13:31:37 Signed-off-by: Michael Smith Aug 26 14:06:23 where does this mysterious ti-codec.inc live? Aug 26 14:06:35 some files in the new recipes/ti dir are looking for it Aug 26 14:06:57 parsing fails Aug 26 14:06:57 hmm Aug 26 14:07:01 is it missing? Aug 26 14:07:22 yes, but looking back before koen's commits last night, it wasn't there before, either Aug 26 14:07:39 maybe he had a git foul up Aug 26 14:08:16 ------ Aug 26 14:08:40 When I try to compile xf86-video-cirrus I have this error Aug 26 14:08:48 ../config.h:4:25: error: xorg-server.h: No such file or directory Aug 26 14:09:14 I have others xf86-video direvers compiled without problem! Aug 26 14:09:15 --- Aug 26 14:09:26 Oh ok I thought it was a straight rename dvsdk -> ti but i guess there were other cleanups. Yeah, probably just missing file. Aug 26 14:11:38 sushisan: maybe it is too old driver Aug 26 14:12:38 msmith_, I believe the problem will resolve itself "soon" Aug 26 14:12:51 03Koen Kooi  07org.openembedded.dev * r73b0cf99f9 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 26 14:13:00 03Koen Kooi  07org.openembedded.dev * rb6235b4c21 10openembedded.git/recipes/ti/ (bitblit/recent-linux.diff bitblit_svn.bb ti-codec.inc): Aug 26 14:13:00 ti-codec: add missing.inc Aug 26 14:13:00 bitblit: add svn recipe Aug 26 14:13:05 Crofton|work: it's almost like your git sense was tingling Aug 26 14:13:10 heh Aug 26 14:13:22 I am going to ask a dumb git question in githib Aug 26 14:13:27 #github Aug 26 14:13:39 there are so many ways to screw up with git Aug 26 14:14:37 hrw: maybe , but I have xorg-server.h Aug 26 14:14:38 : Aug 26 14:14:40 ./staging/i586-kaeilos-linux/usr/include/xorg/xorg-server.h ./work/i586-kaeilos-linux/xserver-xorg-2_1.5.3-r4/install/xserver-xorg-dev/usr/include/xorg/xorg-server.h ./work/i586-kaeilos-linux/xserver-xorg-2_1.5.3-r4/image/usr/include/xorg/xorg-server.h ./work/i586-kaeilos-linux/xserver-xorg-2_1.5.3-r4/xorg-server-1.5.3/include/xorg-server.h ./work/i586-kaeilos-linux/xserver-xorg-2_1.5.3-r4/staging-pkg/staging/i586-kaeilos-linux/usr/ Aug 26 14:16:11 sushisan: add 1.3.2 one and build it Aug 26 14:16:32 hrw: add what where? Aug 26 14:16:56 sushisan: cp 1.2.1 recipe as 1.3.2 one and build it Aug 26 14:42:31 03Koen Kooi  07org.openembedded.dev * r70ddea99ee 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3.inc libgles-omap3_3.00.00.09.bb): Aug 26 14:42:31 libgles-omap3: catch up with dir rename Aug 26 14:42:31 * I really suck at git, sorry Aug 26 14:47:39 quit Aug 26 14:55:27 hi Aug 26 14:55:38 when I try to compile xf86-video-cirrus_1.3.2 I have this error: Aug 26 14:55:44 checking if xorg-macros used to generate configure is at least 1.2... configure: error: configure built with too old of a version of xorg-macros.m4 - requires version 1.1.0 or newer Aug 26 14:55:49 ? Aug 26 14:56:10 Hrm.. who mentioned to build eglibc toolchains here lately? Aug 26 14:58:03 sushisan: do you know what version of xorg-xserver you are building? Aug 26 14:58:09 (or maybe it's called xserver-xorg.. I can'tremember :) Aug 26 14:59:50 msmith_: mmmI think the lastest in OE (1.59 but now I'no not too sure! Aug 26 15:00:11 if you run bitbake -s, it shoudl tell you. Aug 26 15:00:59 each xorg module has its own independent versioning, so coming up with the required versions of each to work together can be a bit... challenging :) Aug 26 15:01:23 msmith_: in /tmp/work/i586-kaeilos-linux I have xserver-xorg-2_1.5.3-r4 Aug 26 15:02:32 however there are some things that can help you, like preferred-xorg-versions-X11R7.4.inc, or kaeilos-2009-preferred-versions.inc or angstrom-2008-preferred-versions.inc Aug 26 15:03:58 morning Aug 26 15:04:18 g'day kergoth Aug 26 15:04:22 morning Aug 26 15:04:38 msmith_: Aug 26 15:04:41 but I have: Aug 26 15:05:00 ./tmp/work/x86_64-linux/util-macros-native-1_1.1.6-r0/util-macros-1.1.6/xorg-macros.m4.in ./tmp/work/x86_64-linux/util-macros-native-1_1.1.6-r0/util-macros-1.1.6/xorg-macros.m4 ./tmp/work/i586-kaeilos-linux/util-macros-1_1.1.6-r0/install/util-macros-dev/usr/share/aclocal/xorg-macros.m4 ./tmp/work/i586-kaeilos-linux/util-macros-1_1.1.6-r0/image/usr/share/aclocal/xorg-macros.m4 ./tmp/work/i586-kaeilos-linux/util-macros-1_1.1.6-r0/stag Aug 26 15:05:11 msmith_: this is 1.1.6 Aug 26 15:05:15 sushisan: if you want to use recent xorg, you can do as I did and copy from angstrom: http://pastebin.ca/1543349 Aug 26 15:05:50 msmith_: trying... Aug 26 15:05:51 Yeah i think you might have a util-macros-native version that is behind your cirrus or xserver-xorg version Aug 26 15:06:08 msmith_: I can put this in local.conf? or in my image? Aug 26 15:06:40 In your local.conf, probably. Aug 26 15:06:42 what DISTRO are you using? Aug 26 15:07:23 you have kaeilos in your pathnames, but it looks like kaeilos-2009-preferred-versions.inc already takes care of this Aug 26 15:09:47 msmith_: kaeilos Aug 26 15:15:31 yeah that's bizarre... kaeilos is using xserver-xorg 1.6.1, not 1.5.3... can you put the non-comment parts of your local.conf into a pastebin? Aug 26 15:15:40 hmmmm Aug 26 15:16:40 itd be nice to have bitdoc pick up file comment headers from individual bitbake .conf/.bb/.inc/.bbclass files, in addition to using the doc flags on the variables Aug 26 15:24:42 re mickey Aug 26 15:32:35 msmith_: I'm back but I must to go out for now. in 2 hour I'm back again! Aug 26 15:44:55 ~curse Debian git/gitosis maintainers Aug 26 15:44:57 May you be reincarnated as a Windows XP administrator, Debian git/gitosis maintainers ! Aug 26 15:50:19 gitosis creates repositories as gitosis:gitosis, git-daemon runs as gitlog:gitlog... and let make them cooperate.... Aug 26 15:58:45 heh Aug 26 15:59:04 that's silly. why not just make a single git user/group? Aug 26 16:00:08 kergoth: yep Aug 26 16:00:30 guess that would make too much sense Aug 26 16:02:56 yep² Aug 26 16:03:25 hrw: File a bug. Aug 26 16:03:38 broonie: thats what I plan to Aug 26 16:07:12 hi Aug 26 16:07:29 ~seen hrw Aug 26 16:07:30 hrw is currently on #oe. Has said a total of 94 messages. Is idling for 3m 52s, last said: 'broonie: thats what I plan to'. Aug 26 16:09:42 ho woglinde Aug 26 16:11:37 okay I asked my wife Aug 26 16:11:42 so its okay to fly Aug 26 16:12:17 to london Aug 26 16:12:22 re mickeyl Aug 26 16:12:41 hi Aug 26 16:13:06 woglinde: cool Aug 26 16:13:17 mickey|sofa: idling from sofa today? Aug 26 16:15:39 hi woglinde, mickey|* Aug 26 16:16:12 hi pb Aug 26 16:16:26 pb ask my wife and so I can go to oedem Aug 26 16:16:38 jolly good Aug 26 16:16:52 woglinde: Phil also has to ask your wife? Aug 26 16:16:55 hrw: pretty much, ya Aug 26 16:16:59 * hrw starts to fear woglinde's wife Aug 26 16:17:00 lol Aug 26 16:17:03 no Aug 26 16:18:46 ;d Aug 26 16:21:35 mickey|sofa: apart from the long-awaited tsc, do you know of any specific agenda items for the oedem? Aug 26 16:22:20 the same as on first oedem: using OE logo/name on businesscards/promo materials, 'blessed companies' Aug 26 16:22:31 :-) Aug 26 16:22:51 were we able to sort out all hosting issues? Aug 26 16:23:00 of no we have another topic... Aug 26 16:23:10 no, hosting still an issue Aug 26 16:23:18 still services running on amethyst Aug 26 16:23:44 mickey|lunch: web? Aug 26 16:23:46 hrm Aug 26 16:23:50 web Aug 26 16:23:52 narcissus Aug 26 16:23:58 ah Aug 26 16:24:24 jo khem Aug 26 16:24:43 gm woglinde Aug 26 16:24:44 mickey|sofa: I guess we can host these at ltg after moving to the new server. Aug 26 16:24:46 pb__: there's a bunch of technical things I'd love to discuss, like, bitbake 1.9 / 2.0, splitting the recipes tree into subtrees, minimizing distroisms, etc. Aug 26 16:25:15 florian: i think the missing factor is someone taking care of moving, less the hosting itself ,since we can use melo on OSUSL Aug 26 16:25:43 will someone be taking notes, or summarizing the decisions made, for those of us who can't attend? Aug 26 16:25:51 mickey|sofa: okay, cool. if you would like to try to write a short summary (just one or two sentences) for each item, I will start trying to assemble an agenda. Aug 26 16:26:09 pb__: yes, i can do that. probably not today, but tomorrow Aug 26 16:26:18 mickey|sofa: great. no rush. Aug 26 16:26:32 kergoth: for sure. last time we used gobby (concurrent editor) for the notes, that was quite impressive and complete Aug 26 16:26:40 nice Aug 26 16:27:16 mickey|sofa: uh ok... oh well. ok, if noone volunteers I can take a look. shouldn't be too hard. But it will take some time. I have too much work here and moving LTG is quite time consuming. Aug 26 16:28:08 i guess cbrake and Laibsch are still kind of working on it...are you? Aug 26 16:28:24 I guess if and when OE e.V. has some funds, one good use for those would be funding "travel bursaries" for folks who couldn't otherwise afford to attend these meetings. Aug 26 16:28:44 but yeah, there should definitely be a record of what is discussed and resolved. Aug 26 16:28:50 *nod* Aug 26 16:29:10 membership fees is also a point for the agenda Aug 26 16:29:48 righto Aug 26 16:29:53 uh right, now that I think the bank account is fully functional we shoudl be able to creata a paypal account too. Aug 26 16:30:16 ah so they accepted the faxed id card? :) Aug 26 16:30:27 if anyone's bored, I'd be curious to hear opinions on my digusting hack(s) in http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=4dc61b382a9e13b3271c69ff6fcbf69432273172 Aug 26 16:30:53 mickey|sofa: I guess so... they didn't complain ;) but i'll test it soon Aug 26 16:32:10 kergoth: you ask for bored people in #oe? ;) Aug 26 16:32:16 hehe Aug 26 16:32:27 kergoth: add some documentation about how to use it :) Aug 26 16:32:38 s/bored/people who need something to distract them from the work they're supposed to be doing/ Aug 26 16:32:40 how's that? Aug 26 16:32:41 ;) Aug 26 16:32:45 and some documentation how to get bored ;) Aug 26 16:33:01 way better :-) Aug 26 16:33:58 hmm, documentation.. Aug 26 16:35:38 i can't think of anything to document Aug 26 16:35:50 make a recipe, put it in a source tree, inherit srctree in the recipe along with the rest of its inherits Aug 26 16:35:52 thats it Aug 26 16:37:10 anyone else think base.bbclass should really be broken up.. a lot, just to organize that mess? Aug 26 16:38:20 * rkirti raises hand Aug 26 16:38:23 +1 Aug 26 16:38:34 wv :) Aug 26 16:39:35 why do i keep adding things to my list like this? :) Aug 26 16:39:46 kergoth dont know Aug 26 16:39:55 glutton for punishment, i guess Aug 26 16:42:49 * florian runs home, bbiab Aug 26 16:42:58 bye florian Aug 26 16:43:44 mickey|sofa: the other night i had a idea, and I'm not entirely sure whether it's disgusting or lovely. consider: individual classes as individual python eggs. versioned, easy to hook into, .. Aug 26 16:44:06 express a requirement on autotools class > 1.0, what have you.. Aug 26 16:45:14 hmm, interesting. Aug 26 16:45:31 besides versioning, what would we get though? Aug 26 16:46:13 self contained, distribution independent of OE, the classes would be entirely python, so no hacks to declare classes and stuff Aug 26 16:46:25 s/classes/python classes/ Aug 26 16:46:42 * kergoth shrugs, can't decide whether he likes the idea or not, just a possibility Aug 26 16:48:21 what I'd like to see is some refactoring first. we never really did this with the classes. some of these are now 6 years old and there's a ton Aug 26 16:49:00 yeah, that's true. there are lots of remnant bits that aren't even used or needed anymore Aug 26 16:49:14 and lots of non-pythonic stuff, or stuff that doesn't take advantage of the current version we require Aug 26 16:49:39 i think the whole way image/rootfs_* is done needs to be scrapped and rearchitected Aug 26 16:50:04 need some nice convenience functions for aspects of the root filesystem construction Aug 26 16:50:12 correct. that's disgusting codfe Aug 26 16:50:15 code, even Aug 26 16:50:15 some of this preceded the 'def ..' syntax even existing Aug 26 16:50:45 agreed. there's a lot of old, bad code in there. Aug 26 16:51:27 i don't know about you guys, but when i first hack together a prototype of something, i find my naming conventions and stuff suck ass. and half the time i never go back and fix it. i think we have a fair bit of that sort of thing floating around too Aug 26 16:51:33 heh Aug 26 16:53:42 okay, so make it suck less first... gotcha ;) Aug 26 16:53:44 baby steps.. Aug 26 16:53:51 righto Aug 26 16:54:08 (prototyp) same thing here Aug 26 16:54:39 for me, often the right style just 'emerges' during the implementation Aug 26 16:54:52 then when you have the first version finished, you can right start with the cleanup Aug 26 16:54:53 i need to get better at the agile/refactoring mindset. thinking in terms of how to get to longer term goals via small incremental steps Aug 26 16:55:06 *g* Aug 26 16:55:08 yeah, that makes sense.. as long as you actually do the cleanup Aug 26 16:55:10 :) Aug 26 16:55:17 *nod* Aug 26 16:56:31 i think I'm still not used to how easy it is to use different programming styles with python. its nice, but I'm not sure if i have an intuitive grasp of when to use which yet. guess thatll come with experience Aug 26 16:58:40 it does Aug 26 16:59:00 at the moment, most of the time i seem to do sort of a bottom up method, kind of functional. get test code working, break it up into functions, then if the functions start to feel like they're all manipulating common data, rearchitect as a class Aug 26 17:10:12 does anyone know of a tool like 'script' that's a bit smarter about escape codes and such? Aug 26 17:10:45 hm only know script Aug 26 17:11:24 hi chouimat Aug 26 17:12:06 hi Aug 26 17:26:21 woglinde, mickey|lunch, pb__: http://kergoth.pastey.net/123427 Aug 26 17:32:35 re Aug 26 17:33:54 hm Aug 26 17:34:07 yeah that is cool Aug 26 17:35:25 03Evan Doiron  07org.openembedded.dev * rf6c302378f 10openembedded.git/ (11 files in 3 dirs): Aug 26 17:35:25 ppp: add v2.4.4 Aug 26 17:35:25 Signed-off-by: Evan Doiron Aug 26 17:35:25 Signed-off-by: Michael Smith Aug 26 17:35:25 Acked-by: Khem Raj Aug 26 17:35:35 03Evan Doiron  07org.openembedded.dev * r2c9d61f33d 10openembedded.git/ (conf/checksums.ini recipes/openvpn/openvpn_2.0.9+2.1rc19.bb): Aug 26 17:35:35 openvpn: add v2.0.9+2.1rc19 Aug 26 17:35:35 Signed-off-by: Evan Doiron Aug 26 17:35:35 Signed-off-by: Michael Smith Aug 26 17:35:37 Acked-by: Khem Raj Aug 26 17:57:04 kergoth: you could automate the process like we do for our assignments here - write a vim script to get rid of all possible ctrl chars the script tool writes :P Aug 26 18:13:34 argh Aug 26 18:15:46 ? Aug 26 18:16:28 bbl Aug 26 18:28:28 anyone run into issues with anonymous functions? Aug 26 18:28:47 specifically, i have a case where one class's anonymous function seems to not be executed whenever i also inherit this other class Aug 26 18:29:29 so somehow this class is fucking up other recipes anonymousf unctions.. but the first guess i had, that it was returning from the anonymous function, is not the case... Aug 26 18:29:35 * cbrake tries Qt Creator and is fairly impressed Aug 26 18:29:56 what a proliferation of IDEs ... Aug 26 18:38:03 hi Aug 26 18:38:28 hi Aug 26 18:38:42 Im trying to compile a kaeilos image Aug 26 18:39:02 but in openembedded main tree I dont have kaeilos-2009-preferred-versions.inc Aug 26 18:39:08 where is this file? Aug 26 18:39:21 hmm Aug 26 18:39:26 what oe branch are you on? Aug 26 18:41:10 I have made: Aug 26 18:41:13 git clone git://git.openembedded.org/openembedded Aug 26 18:41:15 git pull Aug 26 18:42:17 you can use "git branch" to tell you what branch you are on Aug 26 18:43:36 its there on .dev AFAICK Aug 26 18:43:47 AFAICT Aug 26 18:44:05 yeah same here Aug 26 18:44:07 git branch org.openembedded.dev * stable/2009 Aug 26 18:44:45 Yeah so you're on the stable branch for some reason... Try the dev one... Aug 26 18:45:17 where is the list of the devs branches? Aug 26 18:45:27 re Aug 26 18:45:31 (and in my case kailos branch) Aug 26 18:46:29 i don't know kaeilos really but it probably uses org.openembedded.dev as the main branch Aug 26 18:47:36 sushisan: that should have got you .dev Aug 26 18:47:58 sushisan: what does git branch tell you Aug 26 18:48:15 the one with * is your current branch Aug 26 18:48:15 mmmm Aug 26 18:49:00 at http://www.kaeilos.com/?q=download said: Aug 26 18:49:02 Get OpenEmbedded metadata cd /home/koan/devel git clone git://git.openembedded.org/openembedded Update OE metadata $ cd /home/koan/devel/openembedded $ git pull Aug 26 18:49:29 my * is stable/2009 Aug 26 18:50:02 then do this git checkout org.openembedded.dev Aug 26 18:50:10 that should get you onto .dev branch Aug 26 18:51:09 http://juliank.wordpress.com/2009/08/24/the-apt2-project/ Aug 26 18:51:12 omg Aug 26 18:51:32 khem: Aug 26 18:51:40 git checkout org.openembedded.dev Aug 26 18:51:41 error: You have local changes to 'conf/distro/kaeilos.conf'; cannot switch branches. Aug 26 18:51:56 but I dont have touched kaeilos.conf Aug 26 18:52:46 hmm then do git checkout conf/distro/kaeilos.conf Aug 26 18:52:52 this will revert it to pristine state Aug 26 18:52:58 and then do the branch command again Aug 26 18:53:40 git checkout org.openembedded.dev Aug 26 18:53:41 error: pathspec 'recipes/autoconf/autoconf_2.63.bb' did not match any file(s) known to git. Aug 26 18:53:47 mmm Aug 26 18:54:00 I dont remember if I've touched or not this file Aug 26 18:54:53 well somehow your tree is busted Aug 26 18:54:54 yes Aug 26 18:55:13 I delete it and now the chechout works! Aug 26 18:55:26 Your branch is behind 'origin/org.openembedded.dev' by 184 commits, and can be fast-forwarded Aug 26 18:55:44 * org.openembedded.dev Aug 26 18:55:47 ok! Aug 26 18:55:51 now the inc Aug 26 18:56:14 ./conf/distro/include/kaeilos-2009-preferred-versions.inc Aug 26 18:56:16 yes! Aug 26 18:56:22 TXS!!! Aug 26 18:57:58 np Aug 26 19:03:55 khem: Aug 26 19:03:56 Error, DISTRO_PR has changed (.1 to .3) which means all packages need to rebuild. Please remove your TMPDIR so this can happen. For autobuilder setups you can avoid this by using a TMPDIR that include DISTRO_PR in the path. Aug 26 19:04:08 I need to rebuild ALL packages? Aug 26 19:04:13 yes Aug 26 19:04:24 that's what it says, doesn't it? Aug 26 19:04:30 noooooooooooooooooooooooooooooooooooo Aug 26 19:04:33 Error message is self explanatory Aug 26 19:04:51 haha Aug 26 19:04:57 looks like somebody's going to be working late tonight :) Aug 26 19:05:14 yeah...the computer... alone! Aug 26 19:05:21 Now would be a good time to take up smoking... Aug 26 19:05:22 hehe Aug 26 19:07:20 msmith_: in my notebook a fresh rebuild take 3 days, if I'll wait awak I should take something...illegal! :-) Aug 26 19:08:04 heh. i think after 72 hours awake, you are legally insane in some jurisdictions Aug 26 19:08:33 You, sir, need a build server :) laptop hard drives are too slow for this much compiling... Aug 26 19:08:41 not in Argentina Aug 26 19:08:47 :-) Aug 26 19:09:29 well I have a T61 and it takes about 4 hrs Aug 26 19:09:37 for a console-image from scratch Aug 26 19:09:59 yeah, until you get into the x11 stuff, it's not so bad Aug 26 19:14:05 msmith_: Yes, my next investment Aug 26 19:16:57 khem: Lenovo ThinkPad? Aug 26 19:17:07 yes Aug 26 19:17:20 ugly black but is damn fast Aug 26 19:18:38 I have a Intel(R) Pentium(R) Dual CPU T2310 @ 1.46GHz Aug 26 19:19:01 1Gb ram Aug 26 19:23:01 khem: I have x61 tablet now :) Aug 26 19:23:22 * florian likes the design of his thinkpads Aug 26 19:24:57 trackpoint is the best :) Aug 26 19:25:00 florian: I liked it upto T43 Aug 26 19:27:09 * kergoth sighs and re-implements his srctree hacks from scratch Aug 26 19:27:25 khem: is the t61 that bad? Aug 26 19:38:13 03Stanislav Brabec  07org.openembedded.dev * r8ba4ce9074 10openembedded.git/recipes/linux-libc-headers/ (3 files in 2 dirs): linux-libc-headers: Added headers for version 2.6.30. Aug 26 19:40:20 florian: I dont like its looks but perf wise its really good Aug 26 19:43:14 re pb Aug 26 19:49:23 goddamn, I'm having to do some evil shit in some of these classes Aug 26 19:50:01 pb__: did you get the eglibc/micro issue resolved ? I could not reproduce it here Aug 26 20:04:40 khem: no, but I didn't really try; I am happy enough just using regular glibc. Aug 26 20:08:16 pb___: give it one more shot with build from scratch I would say Aug 26 20:09:09 I can try that later, though my previous build was from scratch so I doubt it will be much different. Aug 26 20:09:33 right now, the main thing I need is just to have a working image. Aug 26 20:10:03 pb___: ok Aug 26 20:10:59 I must admit I've never been entirely clear on the advantage of using eglibc compared to normal glibc. Aug 26 20:11:22 pb___: it embedded focussed Aug 26 20:11:27 thats the only difference Aug 26 20:11:53 you can customize it also like uclibc but thats not yet done in OE Aug 26 20:12:03 yeah, I gathered that much, but I don't quite understand what "embedded focussed" means in concrete terms. Aug 26 20:12:14 cross testing Aug 26 20:12:18 cross building Aug 26 20:12:35 size optimization Aug 26 20:12:52 letting non desktop arches to work better Aug 26 20:13:25 although the current model is that changes are pushed in to glibc as much as possible and then pulled back Aug 26 20:14:01 cross generation of locales Aug 26 20:15:01 I think these are quite a bit but once we start using it in OE we can improve on what it offers and also steer it to our needs thats why I am interested in making it default because then it gets more focus Aug 26 20:16:23 okay, fair enough. the only one of those that sounds like a significant advantage for oe right now is size optimization; cross building does already work fine with normal glibc, and cross generation of locales via qemu seems to work well enough for most folk. Aug 26 20:17:04 cross testing might be interesting, though I doubt too many people will want to actually use it; you could do that previously under normal glibc with a little bit of fiddling and I don't think many/any users have bothered. Aug 26 20:27:19 yes motive is to to make it more cross friendly so more and more devs can easy run tests Aug 26 20:28:07 | configure: error: C preprocessor "arm-angstrom-linux-gnueabi-gcc -E" fails sanity check Aug 26 20:28:08 moreover I keep it uptodate in terms of releases in OE :) Aug 26 20:28:21 Jay7: show the config.log Aug 26 20:28:55 ah.. it may be yesterday's hangup Aug 26 20:30:11 /data/openembedded/tmp/cross/armv5te/lib/gcc/arm-angstrom-linux-gnueabi/4.3.3/include-fixed/limits.h:122:61: error: limits.h: No such file or directory Aug 26 20:31:01 it is glibc-2.9-r34.1 Aug 26 20:59:51 morning all Aug 26 21:00:19 hey RP Aug 26 21:00:22 hey RP Aug 26 21:00:43 hi rp Aug 26 21:03:29 hey RP Aug 26 21:04:56 kergoth: RP: I feel so sorry for being so insistent..but actual state of packaged-staging in OE is half-broken to be polite :) Aug 26 21:05:25 I can't believe I'm the only one guinea-pig here... Aug 26 21:05:46 my backlog of commits i need to get upstream to OpenEmbedded is like... 150 commits, or something Aug 26 21:06:11 ok, I don't talk about improvements now, just fixes Aug 26 21:07:35 I'm not talking about just improvements either. there are recipe and class fixes in mvl6 that aren't upstream yet. perhaps some of those may help, i don't know offhand Aug 26 21:07:44 but I'm sure you can learn enough to fix the problems .. ;) Aug 26 21:08:04 ;) Aug 26 21:08:04 03Chris Larson  07org.openembedded.dev * r6a5e3eadec 10openembedded.git/classes/srctree.bbclass: Aug 26 21:08:04 srctree.bbclass: Don't automatically use git clean. Aug 26 21:08:04 This should be an opt-in feature. Aug 26 21:08:04 Signed-off-by: Chris Larson Aug 26 21:08:09 03Chris Larson  07org.openembedded.dev * rc45f041e4a 10openembedded.git/classes/srctree.bbclass: Aug 26 21:08:09 srctree: completely reimplement the task merging into do_populate_staging. Aug 26 21:08:11 This fixes issues seen with the use of kernel.bbclass, so in existing source Aug 26 21:08:13 tree kernel builds work happily now. Aug 26 21:08:15 Signed-off-by: Chris Larson Aug 26 21:08:45 RP: http://kergoth.pastey.net/123427 Aug 26 21:10:17 * kergoth sighs and reads the Moblin compliance guide Aug 26 21:16:44 03Stanislav Brabec  07org.openembedded.dev * r317b2b076c 10openembedded.git/recipes/update-rc.d/update-rc.d_0.7.bb: update-rc.d: INHIBIT_DEFAULT_DEPS to allow glibc to inherit update-rc.d. Aug 26 21:16:45 03Stanislav Brabec  07org.openembedded.dev * r056a67f607 10openembedded.git/recipes/glibc/glibc-package.bbclass: glibc: Install ld.so.conf, so ldconfig does not complain. Aug 26 21:16:48 03Stanislav Brabec  07org.openembedded.dev * rd007a914d5 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 26 21:16:49 03Stanislav Brabec  07org.openembedded.dev * r70031b2c7f 10openembedded.git/recipes/glibc/ (5 files in 2 dirs): glibc: nscd: Make it working. Aug 26 21:17:00 ant__: So don't use packaged-staging :/ Aug 26 21:17:16 :) Aug 26 21:17:19 saves time Aug 26 21:17:27 ant__: My workload is not getting any easier or reducing sadly Aug 26 21:17:38 * kergoth doesn't quite have the entirety of packaged-staging.bbclass modeled in his head, so isn't quite ready to clean it up for fix difficult bugs in it at the moment.. working on it Aug 26 21:17:41 btw I have a list of packages failing during rebuild from pstage... Aug 26 21:17:48 IOError: [Errno 2] No such file or directory: '/data/openembedded/tmp/oestats.id' Aug 26 21:17:50 hm.. Aug 26 21:17:58 who should create that file? Aug 26 21:18:10 ant__: Can you email a summary of the problems to the OE list and cc me please Aug 26 21:18:31 already in ML and patchwrk Aug 26 21:18:35 kergoth: Very nice :) Aug 26 21:18:40 21-august (Andrea) Aug 26 21:19:22 ah.. Aug 26 21:19:31 basic problem is bitbake doesn't find the compiler after rebuild from pstage Aug 26 21:19:38 bitbate does not creating that file when in interactive mode Aug 26 21:19:49 second one ishappens if you delete the /ipk dir Aug 26 21:19:53 is it bug or feature? Aug 26 21:19:55 RP: check out merge_tasks in http://bit.ly/4p4Mc8 ... i think i may have committed some sin in creating this monstrosity ;) Aug 26 21:20:53 kergoth: How about just using locking? Aug 26 21:20:59 not sufficient. Aug 26 21:21:21 one thread could release the lock after its configure completes, then the native version runs its configure and releases the lock, then the first thread runs the compile for non-native Aug 26 21:21:22 boom Aug 26 21:21:24 interleaving is the problem Aug 26 21:21:31 the lock needs to be across the whole process Aug 26 21:21:34 hence the task merging Aug 26 21:21:38 kergoth: Invent a new lock Aug 26 21:22:13 perhaps.. would be annoying to implement without modifying bitbake, unless i do the footwork in an event handler Aug 26 21:22:17 i'll look into that, thanks Aug 26 21:22:19 kergoth: That function is rather clever. Insane too mind Aug 26 21:22:55 kergoth: but if modifying bitbake removes the need for functions like that? ;-) Aug 26 21:23:34 aye. also assumes that no other recipes will express a dependency on any of the tasks from configure through populate_staging, if they do, things could get... unpleasant Aug 26 21:25:50 hi Aug 26 21:27:08 hmmm Aug 26 21:27:36 * kergoth pictures in his head how the "threads" would interact using a cross-task lock, to try to spy out any problems Aug 26 21:27:47 (fond memories of diagnosing locking issues in the kernel..) Aug 26 21:28:20 kergoth: I'd just make configure take the lock and populate_staging release it or something Aug 26 21:28:33 and then debian.bbclass will totally break things Aug 26 21:28:38 yeah, thats what i was thinking too Aug 26 21:29:11 debian.bbclass is both a blessing and a curse Aug 26 21:29:40 taskstarted even for configure -> acquire, taskcompleted/failed for staging -> release Aug 26 21:29:43 s/even/event/ Aug 26 21:30:42 do_configure[takelock] = "${WORKDIR}/foo" do_stage[releaselock] = "$..." Aug 26 21:30:52 but the events could do it too Aug 26 21:31:19 bitbake/oe is that complicated already there is something to be said for keeping simple :/ Aug 26 21:31:32 the downside i see to a lock based solution is that that bitbake thread is tied up, blocking on the lock, doing nothing, and the other thread isn't guaranteed to be finishing off that first task immediately, it could be off doing other things.. which means the thread is blocked longer than it has to be.. in the merged task solution, the max wait time is the length of the configure->install operation Aug 26 21:31:42 s/first task/first recipe/ Aug 26 21:32:11 Well, we could teach bitbake to not run the tasks until the lock is available Aug 26 21:32:27 means teaching the scheduler about locks but that isn't totally insane Aug 26 21:32:28 yeah, that's true. do a try-acquire or whatever Aug 26 21:32:50 as an alternative, could teach it to know how to run sets of tasks in series if they're declared as such Aug 26 21:32:58 yes Aug 26 21:33:07 our current scheduler is totally thick Aug 26 21:33:14 well, kind of Aug 26 21:33:24 its just very simple Aug 26 21:33:30 if we tought it how to handle series tasks, the acquire/release model could be done entirely in events, and would be trivial Aug 26 21:33:32 * kergoth nods Aug 26 21:33:48 someone asked for a parallel-build of images, being that each image has own rootfs Aug 26 21:33:59 ant__: put IMAGE_ROOTFS in WORKDIR. works great Aug 26 21:33:59 ant__: Poky does that Aug 26 21:34:33 ok, nasty is I have a lot of recipes killing parallel build... Aug 26 21:34:34 so, I'm pretty sure mvl6 is now available for sale, just hasn't had a press release yet Aug 26 21:34:34 kergoth: Have you hit the opkg issues yet? Aug 26 21:34:36 not entirely sure why.. Aug 26 21:35:12 kergoth: I'd love to take a look but I can't see that happening any time soon... Aug 26 21:35:25 RP: not yet, what issues? the opkg metadata for rootfs_ipk is already in workdir.. haven't tried multiple parallell image builds in a single image recipe, no real point Aug 26 21:36:01 one cool thing we do, which i was telling someone about yesterday, is our image.bbclass automatically produces a -dbg incarnation of the filesystem with the dbg packages installed, useful for gdb/gdbserver. its in the queue to be pushed Aug 26 21:36:18 kergoth: Image A completes its depenencies whilst recipe B is still building. Its a new version of B hence it uninstalls from staging and removes its packages from deploy/ipk as opkg-make-index runs Aug 26 21:36:21 boom :) Aug 26 21:36:50 if you want a sure failing test, just rebuild from pstage for Z Aug 26 21:36:51 ugh, didn't think about the interaction with pstage's binary package handling.. Aug 26 21:36:59 I've just made some patches to opkg-make-index which probably will make people wonder about my sanity Aug 26 21:37:10 poky has the fixes for that? Aug 26 21:37:27 kergoth: Yes. For certain definitions of fix Aug 26 21:37:31 hehe Aug 26 21:37:34 kergoth: I'm busy testing them Aug 26 21:37:42 kergoth: Just ignore all OSErrors Aug 26 21:37:50 if the package disappears, so what? :) Aug 26 21:38:00 mvl6 will need it too, I'm sure its just a matter of time till a customer reports it Aug 26 21:38:12 My autobuilder was hitting it every other run Aug 26 21:38:16 yow Aug 26 21:38:32 it is set to rebuild every time one of several git repos gets a commit Aug 26 21:38:48 right now our qa folks build one image per bitbake run, other than initramfs stuff. gets built nightly, not continuous Aug 26 21:39:18 kergoth: This is one image per run with an initramfs Aug 26 21:39:28 guess we've just gotten lucky thus far :) Aug 26 21:39:38 we build for ... like 8-9 machines at the moment Aug 26 21:39:45 I just build for one Aug 26 21:39:55 but its building new revisions every 10 mins Aug 26 21:39:58 * kergoth nods Aug 26 21:40:22 RP: any idea why this would happen? http://tinderbox.openembedded.net/public/logs/task/2600034.tx Aug 26 21:40:32 * kergoth finds it amusing how different third party collections end up hitting similar issues.. this is why we need to get stuff upstream faster :) Aug 26 21:41:05 ant__: 404 errors? They usually mean incorrect urls ;-) Aug 26 21:41:13 add a t...this is rebuild from bare /pstage, after having Aug 26 21:41:20 kergoth: Right Aug 26 21:41:29 fixed zaurus-updater and a nother couple Aug 26 21:41:33 kergoth: The trouble is I don't dare touch OE.dev these days, too out of touch :( Aug 26 21:42:44 I'm .. almost happy with my workflow for the upstream stuff. created a fake merge commit where the collection in question was created from OE recipes, then created a shell script to dump a log but hiding blacklisted commits, and i blacklist the commits that 'git cherry' finds as already cherry picked between them, and ones i add explicitly Aug 26 21:43:25 so its git-cherry-list foundation-sync..foundation-for-oe, and browse for stuff that should go up, cherry pick them over and add to the blacklist Aug 26 21:47:36 03Chris Larson  07org.openembedded.dev * r2259a81628 10openembedded.git/classes/ (clean.bbclass srctree.bbclass): Aug 26 21:47:36 srctree, clean: cleanup & split out do_clean bits into clean.bbclass. Aug 26 21:47:36 Signed-off-by: Chris Larson Aug 26 21:47:47 gogo kergoth Aug 26 21:47:49 kergoth: Neat. Its a bit more tricky since Poky's history and OE's doesn't coincide :/ Aug 26 21:47:50 *g* Aug 26 21:48:57 RP: ours doesn't either, at least directly. the merge commit is really by date, when we started the repositories. some stuff got added after that, but its a starting point to filter out the majority of the history. course, poky has been around long enough that the manual work would be excessive, but you could automate more of it Aug 26 21:49:20 i was thinking about checking the blob hashes for added recipes and see when that blob was in a tree in OE, so we know where it came from & what version Aug 26 21:49:25 ;) Aug 26 21:59:56 hmm, i should make my pure python do_unpack work on python 2.4 and get it pushed Aug 26 22:00:02 stupid lack of decorators.. Aug 26 22:06:26 kergoth: :) Aug 26 22:06:43 kergoth: My main worry with merging things with OE.dev is lack of testing Aug 26 22:06:57 yeah.. its hard to test every combination Aug 26 22:07:06 I'm not in a position to test OE.dev at all these days and don't feel right pushing changes without testing them Aug 26 22:07:13 ideally we'd be able to tell an autobuilder to fire away at a new branch you create with the isolated changes Aug 26 22:07:21 since its difficult to test all the combinations on your own Aug 26 22:13:19 kergoth: That would certainly help Aug 26 22:14:54 but even then, we'd probably want to teach the autobuilder to run builds inside of virtual machines which are different Linux distros and such, in case of build environment issues Aug 26 22:15:01 heh Aug 26 22:17:11 i had an idea a while back.. i think I'm going to set up a class to monitor the datastore for changes that a task causes, with the goal of being able to pass the tasks a fresh createCopy() of the data instead of the original, so their changes are truly local, and won't affect the rest of the build. tasks touching the metadata can lead to problems with nondeterministic builds Aug 26 22:19:22 good nite Aug 26 22:19:27 night Aug 26 22:31:04 * * OE Bug 5308 has been created by javiroman(AT)kernel-labs.org Aug 26 22:31:06 * * conf/machine/include/tune-geodelx.inc is missing Aug 26 22:31:08 * * http://bugs.openembedded.org/show_bug.cgi?id=5308 Aug 26 22:34:21 03Chris Larson  07org.openembedded.dev * r4cd5e9101d 10openembedded.git/classes/base.bbclass: Aug 26 22:34:21 base.bbclass: add cleanall task. Aug 26 22:34:21 Can be particularly useful when using multiple srctree recipes. Aug 26 22:34:21 Signed-off-by: Chris Larson Aug 26 22:35:25 03Chris Larson  07org.openembedded.dev * r09b6a74a6f 10openembedded.git/classes/srctree.bbclass: Aug 26 22:35:25 srctree: prefix messages with ${PF} and kill do_package hacks. Aug 26 22:35:25 Signed-off-by: Chris Larson Aug 26 22:50:11 03Martin Jansa  07fso/milestone5.5 * ra0cb94509c 10openembedded.git/recipes/navit/ (4 files in 2 dirs): Fix navit library path. navit.xml was renamed to navit_shipped.xml in svn rev 2492. Aug 26 22:50:12 03Sebastian Krzyszkowiak  07fso/milestone5.5 * rf6d595aa19 10openembedded.git/recipes/navit/ (navit.inc navit_svn.bb): navit: don't recommend speech-dispatcher and fix config file to work without -dev package Aug 26 22:50:12 03Sebastian Krzyszkowiak  07fso/milestone5.5 * r89c3e606c1 10openembedded.git/recipes/navit/files/navit.xml-so.patch: navit: update configuration patch Aug 26 22:50:14 03Sebastian Krzyszkowiak  07fso/milestone5.5 * r5da01ede94 10openembedded.git/recipes/navit/files/navit.desktop: navit: fix QA issue in desktop file Aug 26 22:50:17 03Sebastian Krzyszkowiak  07fso/milestone5.5 * rbcd3e18db0 10openembedded.git/recipes/navit/ (navit.inc navit_svn.bb): navit: don't need -dev package to run anymore Aug 26 22:50:22 03Angus Ainslie  07fso/milestone5.5 * ra8bb4e2e62 10openembedded.git/recipes/openmoko-projects/paroli_git.bb: paroli : add paroli-calculator and remove uneeded shr overide Aug 26 22:50:25 03Sebastian Krzyszkowiak  07fso/milestone5.5 * r333126d30e 10openembedded.git/recipes/navit/navit_svn.bb: navit: remove .so.0 patch for navit.xml, not needed anymore Aug 26 23:23:31 03Chris Larson  07org.openembedded.dev * r8036a37856 10openembedded.git/recipes/nfs-utils/nfs-utils_1.1.2.bb: Aug 26 23:23:31 nfs-utils: fix the broken /sbin/mount.nfs4 symlink. Aug 26 23:23:31 Signed-off-by: Chris Larson Aug 27 00:22:56 03Rolf Leggewie  07org.openembedded.dev * r572bf3a153 10openembedded.git/conf/distro/include/preferred-xorg-versions-X11R7.4.inc: Aug 27 00:22:56 preferred-xorg-versions-X11R7.4.inc: pin xextproto-native and xtrans-native Aug 27 00:22:56 * xextproto-native: 7.0.3 Aug 27 00:22:56 * xtrans: 1.2.1 Aug 27 00:22:57 03Khem Raj  07org.openembedded.dev * re77453c96b 10openembedded.git/conf/machine/ (alix.conf xo.conf): Aug 27 00:23:01 alix.conf, xo.conf: Use tune-geode.inc instead of tune-geodelx.inc Aug 27 00:23:03 Signed-off-by: Khem Raj Aug 27 00:23:05 03Khem Raj  07org.openembedded.dev * r9ae59f45ac 10openembedded.git/: Merge branch 'origin/org.openembedded.dev' of git@git.openembedded.org:openembedded into develop Aug 27 00:28:05 * * OE Bug 5308 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 27 00:28:07 * * conf/machine/include/tune-geodelx.inc is missing Aug 27 00:28:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5308 Aug 27 01:54:34 anybody knows about USB programming Aug 27 02:01:58 mankash, you'll have to be more specific Aug 27 02:03:23 I have a board based on IXP425 processor, it has 2 pin and 5 pin headers which are not used, I am wondering if any one of them is USB port Aug 27 02:06:02 USB needs 4 pins Aug 27 02:07:58 oops, sorry. thats bullshit Aug 27 02:08:18 what is bullshit here? Aug 27 02:08:20 2 pins should be enough Aug 27 02:08:26 D+ and D- Aug 27 02:08:51 true Aug 27 02:10:02 other way may be to test it is write 0 and 1 on the usb config registers and thenmeasure the voltage Aug 27 02:10:21 mankash, does this help? http://www.intrasoft-spb.ru/images/vulcan.jpg Aug 27 02:10:30 but do now knw if it is possible Aug 27 02:11:27 no Aug 27 02:11:57 top right says 2x USB Aug 27 02:12:07 looks like two 5 pin headers to me Aug 27 02:12:24 hmm Aug 27 02:22:00 a build of tcpdump wants to build sqlite3, alsa-lib, gstreamer plugins... what the fuck is OpenEmbedded smoking? Aug 27 02:22:04 * kergoth grumbles Aug 27 02:25:18 heh Aug 27 02:25:23 bluetooth support Aug 27 02:25:33 * kergoth sighs Aug 27 02:25:47 friends don't let friends use bluetooth Aug 27 02:25:50 USE_FLAG please? pretty please? Aug 27 02:25:55 Tartarus: was just thinking that Aug 27 02:26:10 i wonder if we should do one for --with/--without, and a different one for --enable/--disable, or something Aug 27 02:26:12 hmm Aug 27 02:26:17 implementing it wont take long Aug 27 02:26:53 this is what keeps surprising me. i go and try to implement something I've wanted for ages in bitbake/oe, and bam, its functional in less than 24 hours of effort. makes you wonder why we never did them sooner :) Aug 27 02:27:08 i guess most people just -use- the thing, and it works well enough they dont want to fuck with it Aug 27 02:27:43 MontaVista will really help drive improvements to the system, since unlike most companies using OE, they ship it to their customers, so instead of just improving it for their own use, they're improving it to satisfy customers Aug 27 02:50:10 kergoth: what feature is it ? Aug 27 02:51:50 hmm, features added for mvl6.. collections.inc, amend.inc, srctree & gitver, .... what else.. also added fetching of pstage binaries and configuration signature generation and inclusion in the pstage filenames, but i don't know if that bit will go upstream.. Aug 27 02:51:51 * kergoth thinks Aug 27 02:52:04 i hate my memory Aug 27 02:53:21 kergoth: i was thinking about this: PV = "${@'${GITVER}'[1:]}" Aug 27 02:53:25 to strip the leading v off Aug 27 02:53:42 I'm thinking a lot of git projects probably use version tags that start with a v Aug 27 02:53:56 msmith_: i was thinking about a hook. if you 'def' a, say, gitver_mangle(), it would run it. alternatively, for 'v', can probably be smart and strip it out if its there Aug 27 02:53:59 yeah Aug 27 02:54:11 yeah it would be really cool if the class would do sometihng like check if the GITVER starts with v[0-9], and strip the v Aug 27 02:54:15 gnulib has a git-version-gen that does that kind of thing Aug 27 02:54:23 could check for busybox style tagging too Aug 27 02:54:27 you could always provide a GITVER_RAW if people wanted the unmangled ver Aug 27 02:54:28 1_15_0 for 1.15.0 Aug 27 02:54:35 good idea Aug 27 02:54:40 Huh, yeah, that would also be neat. Aug 27 02:54:58 i'm sure there are lots of corner cases where people have weird tag names. but those people are gonna run into trouble anyway Aug 27 02:55:10 they can always use a python function to manipulate it if they have to Aug 27 02:55:15 yeah. Aug 27 02:55:16 or just call out an external version extraction script Aug 27 02:55:22 quilt includes a git-desc script for it Aug 27 02:55:32 that's true, since the source tree is there, you could even parse it out of a header file if you had to Aug 27 02:56:16 indeed Aug 27 02:56:32 that would be cool for things like svn. pull the version from a file, then supplement with the current svn revision Aug 27 02:57:05 ooh, i know. *if* autoconf/m4 are available at parse time, which isn't necessarily the case, mind you, but if it is, you could pull the version number passed to AC_INIT Aug 27 02:57:15 hehe **** ENDING LOGGING AT Thu Aug 27 02:59:57 2009