**** BEGIN LOGGING AT Sat Nov 24 02:59:57 2007 Nov 24 06:13:38 03xjqian 07org.oe.dev * r1aa5367f... 10/ (3 files in 2 dirs): gcc-(cross)-package.inc: selective fortran packaging. Nov 24 07:44:50 i know this is off-topic, but i am wondering if someone can help me obtain an original rom image for the hx4700... i lost mine, and i need to restore it. Nov 24 09:54:04 * * OE Bug 3361 has been created by autobuild(AT)openembedded.org Nov 24 09:54:06 * * gcc-cross-4.1.2-r11-do_compile Nov 24 09:54:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3361 Nov 24 10:11:30 evenin Nov 24 10:14:21 BOO Nov 24 10:16:56 what's up Nov 24 11:23:09 03pH5 07org.oe.dev * re6a5433d... 10/ (1 packages/openmoko2/neod_svn.bb): neod: set platform to htc for htcuniversal to enable power/aux keys Nov 24 11:29:03 morning all Nov 24 11:29:10 hey RP Nov 24 11:29:12 koen : glibc builds ok on the dht Nov 24 11:29:47 koen : x11-image finished also Nov 24 11:29:54 RP: Hi ! Nov 24 11:30:08 Is OK to me to send a set of patches to the mailing list? Nov 24 11:30:24 RP: And i was about to ping you with a question :) Nov 24 11:30:34 otavio : better use the bugtracker Nov 24 11:30:36 otavio: Bugzilla is preferred for patches Nov 24 11:30:40 I've produced some and want to share them and get them merged Nov 24 11:31:07 RP: well ... I sent one 5 days ago and got no comments about it yet so I thought ml would be faster Nov 24 11:31:27 otavio: To the bugzilla? Nov 24 11:31:31 RP: yes Nov 24 11:31:43 http://bugs.openembedded.org/show_bug.cgi?id=3342 Nov 24 11:31:52 RP: I am using an older version of the image.bbclass and i encounter a problem where makedevs does not create the devices in the dev dir Nov 24 11:32:24 otavio: We're just all very busy :(. I'd mention it in here occasionally and someone will look at it sooner or later ;-) Nov 24 11:32:49 RP: i assume that since its running in a fakeroot "enviroment" there should not be any problem Nov 24 11:33:22 steliosk: I remember seeing that, it did get fixed. I wish I could remember what we did though... Nov 24 11:33:28 I'm not used to monotone but there's an easy way to keep a branch there and it be pulled by you all? Nov 24 11:33:36 steliosk: I think there were bugs in makedevs? Nov 24 11:33:40 I'm using the GIT mirror since I'm used to it Nov 24 11:35:30 RP: copied the latest from .dev and results where the same Nov 24 11:36:02 RP: if i run the same command from the bash command line with fakeroot the devs get created Nov 24 11:36:43 RP: if i run it without fakeroot it does not, so it looks like its a fakeroot issue Nov 24 11:36:51 steliosk: Which version of fakeroot are you using? Nov 24 11:37:00 RP: let me check Nov 24 11:38:36 otavio : what machine/cpu you had issues ? Nov 24 11:39:40 otavio: The patch looks good, its the same as the one we apply to 2.18? Nov 24 11:40:26 i386 Nov 24 11:40:35 RP: it has a file change Nov 24 11:40:40 RP: configure.in Nov 24 11:40:48 RP: but virtually the same Nov 24 11:40:59 RP: I have a firefox fix too Nov 24 11:42:14 RP: makedevs 1.0.0-r5 fakeroot 1.7.1 Nov 24 11:43:12 steliosk: Same versions as Poky which works :/ Nov 24 11:44:13 RP: i know its weird. I haven't noticed it till we noticed a "canot create initial console" message when booting a board Nov 24 11:44:49 steliosk: Have you compared you device table file with the one in OE? Nov 24 11:45:05 steliosk: It could be the entry for creating the /dev directory itself is missing Nov 24 11:45:34 RP: i tried even the .dev device_table.minimal.txt with same results Nov 24 11:46:51 steliosk: All I can suggest is trying to trace the process and seeing where/when it goes wrong Nov 24 11:49:38 otavio: I've pushed the binutils fix. Nov 24 11:49:58 RP: great. Nov 24 11:50:07 RP: please close the bug there then :-) Nov 24 11:51:33 otavio: I marked it as fixed Nov 24 11:58:24 RP: one nice and easy fix for my sourceforge downloading problems was to set SOURCEFORGE_MIRROR="http://qa.debian.org/watch/sf.php" and then I suggest to use it on bitbake.conf by default Nov 24 11:58:41 RP: it uses a wrapper done by Debian to set the referer and like Nov 24 12:00:44 03rpurdie 07org.oe.dev * rf403c01c... 10/ (5 files in 4 dirs): binutils: Add patch to work with recent makedevs versions (#3342) Nov 24 12:07:09 RP: not makedevs but makeinfo Nov 24 12:07:12 RP: ;-) Nov 24 12:19:40 otavio: gah ;-) Nov 24 12:20:21 RP: have time to take a look on a firefox buildfix? Nov 24 12:20:38 otavio: The sourceforge change would need discussion Nov 24 12:20:54 RP: OK .. I did it in my local.conf ;-) Nov 24 12:20:58 otavio: I'm afraid firefox isn't really something I know much about Nov 24 12:21:05 RP: just think it ought to be useful to others Nov 24 12:21:16 otavio: That is what local.conf is used for :) Nov 24 12:21:23 otavio: Mention it on the mailing list Nov 24 12:21:46 Will do it later Nov 24 12:22:00 First, I want to finish the changes I had to do to make the image Nov 24 12:22:04 otavio: The bugtracker is good for tracking patches, the mailing list for archiving info and discussing things Nov 24 12:22:28 RP: yeah but it changes from team to team Nov 24 12:22:53 RP: on Debian Boot (debian installer team) we use ml to discuss patches that changes things more deeply Nov 24 12:23:03 RP: on parted too Nov 24 12:23:14 RP: here, you prefer bugtracker Nov 24 12:23:26 otavio: I'm telling you how OE works :) Nov 24 12:23:39 otavio: Deep changes would be discussed on the mailing list Nov 24 12:23:44 RP: got it Nov 24 12:24:16 otavio: We used to have patches posted to the list but too many got lost Nov 24 12:25:18 RP: is mtn able to have a branch and you guys pull from it? As we ofthen do with GIT? Nov 24 12:25:34 RP: if it can, I could set it up and then use it to make the patches available Nov 24 12:25:58 RP: I'm not used to mtn and then I'd like to ask for someone used to it Nov 24 12:26:34 otavio: In theory it can be used that way but we prefer to use patches Nov 24 12:27:08 RP: using patches contributors do not receive the credit for the change Nov 24 12:27:23 RP: and this can help to get people interest on helping on OE too Nov 24 12:27:32 otavio: I should have used the --author swicth when I committed that, sorry :( Nov 24 12:27:46 * RP keeps forgetting to do that :( Nov 24 12:28:15 RP: I'm not talking about this specific change but how I see happen on other projects too Nov 24 12:28:24 RP: no problem ;-) Nov 24 12:28:27 I did reference the bug number as a pointer to where it came from... Nov 24 12:30:04 * * OE Bug 3362 has been created by autobuild(AT)openembedded.org Nov 24 12:30:06 * * linux-rp-2.6.21-r11-do_compile Nov 24 12:30:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3362 Nov 24 12:31:04 * * OE Bug 3342 has been RESOLVED (FIXED) by rpurdie(AT)rpsys.net Nov 24 12:31:06 * *  binutils 2.17.50.0.5 fails to build due makeinfo version problem Nov 24 12:31:09 * * http://bugs.openembedded.org/show_bug.cgi?id=3342 Nov 24 12:31:39 users Nov 24 13:56:02 koen: ping Nov 24 13:56:19 pong Nov 24 13:56:57 koen: does pty work for you? it SIGSEGVs here Nov 24 13:57:24 once we have all the dbus collaboration line, we can swap like hell Nov 24 13:57:27 above dbus and below Nov 24 13:57:35 until we come to a rock solid and fun environment Nov 24 13:57:42 pty works for me on the a780 Nov 24 13:57:48 "on" the a780? Nov 24 13:57:53 oops sorry, wrong channel Nov 24 13:58:12 isn't pty supposed to run on the PC? Nov 24 13:58:21 not on the a780 Nov 24 13:58:36 it's needed to give gsmd a char device instead of a socket that ezxd provides Nov 24 13:58:52 i wanted to use it like thomas describes Nov 24 13:58:58 but it just crashes here Nov 24 13:59:17 tried stracing it? Nov 24 14:00:13 ya, no clue Nov 24 14:00:19 gdb fails somewhere in libc Nov 24 14:00:26 s/fails/shows/ Nov 24 14:08:23 mickeyl : Your daily reminder about the pdf's :) Nov 24 14:08:42 ouch sorry. where's my head Nov 24 14:08:53 * steliosk looks for mickeyl's head Nov 24 14:09:00 * mickeyl notes as first-time todo after the league game he's going to play in 30 minutes Nov 24 14:10:05 koen : Does the autobuilder still have issues with the dht/glibc ? Nov 24 14:10:38 koen : i build x11-image from scratch without problems Nov 24 14:10:56 steliosk: the autobuilder can't :( Nov 24 14:11:50 * steliosk wonders if he had forgot to post any patches Nov 24 14:12:13 but i seriously dough it Nov 24 14:13:34 koen : -rw-r--r-- 1 stelios stelios 17563272 2007-11-23 21:16 Angstrom-x11-image-glibc-ipk-2007.9-test-20071123-dht-walnut.rootfs.jffs2 Nov 24 14:17:27 time to go home... Nov 24 14:17:35 bbl Nov 24 16:06:37 03koen 07org.oe.dev * r3a45ef9a... 10/ (1 conf/machine/fic-gta02.conf): fic-gta02: ship Xfbdev as well Nov 24 16:06:43 03koen 07org.oe.dev * ra3bde965... 10/ (3 files in 3 dirs): xserver-kdrive-common: use Xfbdev on gta02 for now Nov 24 18:01:38 03koen 07org.oe.dev * r8dbcfd96... 10/ (3 files in 3 dirs): xserver-kdrive-common: add missing break in the switch statement Nov 24 18:16:07 03rpurdie * r985 10/ (6 files in 6 dirs): Add bb.utils.lockfile() and bb.utils.unlockfile() from Poky. Use these functions in the fetcher code Nov 24 18:30:23 morning Nov 24 18:42:48 hi Nov 24 18:50:04 !oebug Nov 24 18:50:05 Syntax error: !*bug Nov 24 18:50:20 !oebug 3363 Nov 24 18:50:21 * * Bug 3363, Status: NEW, Created: 2007-11-24 10:32 Nov 24 18:50:22 * * mwester(AT)dls.net: Build fails on gcc-cross_4.1.2 - errors with gfortran Nov 24 18:50:23 * * http://bugs.openembedded.org/show_bug.cgi?id=3363 Nov 24 18:51:10 xjqian: ^^^^ Can you take a look at that? Is there any way to disable the fortran stuff (I don't think very many folks really need to have fortran in order to build their distros!) Nov 24 18:52:10 * CM ponders what a fortran centred distro would be like Nov 24 18:52:50 Sigh. Yeah. Next we'll build pascal, APL, COBOL, and RPG along with gcc for each distro. ;) :D Nov 24 18:53:02 Hehe Nov 24 19:09:12 well, a solution to the fortran thingy would be most welcome... Nov 24 19:12:04 * * OE Bug 3363 has been created by mwester(AT)dls.net Nov 24 19:12:06 * * Build fails on gcc-cross_4.1.2 - errors with gfortran Nov 24 19:12:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3363 Nov 24 19:13:24 slow bot. Nov 24 19:14:41 the bot is relatively fast, but the bughost is uber-slow sending the notification emails at times Nov 24 19:16:10 Sigh. Setting LD_LIBRARY_PATH so that the missing library can be found results in a failure from glibc. Nov 24 19:17:22 Why O why is such a disruptive change not done as a separate package, such as "gcc+gfortran-cross_4.1.2" instead? Nov 24 19:18:54 Well, the heck with this. I could spend all bloody day getting the toolchain to work, but I'm just going to revert this gfortran stuff locally, and get on with my life. :( Nov 24 19:42:36 mwester: got it working? Nov 24 19:46:38 CoreDump: I think I'm past that point. Nov 24 19:46:47 Basically, the following: Nov 24 19:47:05 bitbake mpfr-native to build the missing library and stage it. Nov 24 19:47:41 * CoreDump tries Nov 24 19:47:52 then do export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/path/to/staging/i686-linux/lib bitbake gcc-cross Nov 24 19:48:00 ugh Nov 24 19:48:33 I guess I'll revert my tree a few revisions :\ Nov 24 19:48:38 then make sure that the LD_LIBRARY_PATH is set BACK to the original value (does not contain the /path/to/staging/i686-linux/lib), and do bitbake Nov 24 19:48:45 It's going onward right now. Nov 24 19:49:16 mwester: thanks Nov 24 19:49:39 Hi, I always seem to have trouble with building gettext-native. This time I am getting http://pastebin.org/9241 Does anyone have an idea what might be wrong? Nov 24 19:51:34 redguy: is the "tmp" from /home/users/matik/oe/build-clean/tmp/ a symlink? Nov 24 19:53:27 if so..don't do that Nov 24 20:03:07 CoreDump|bbl: nah, no symlinks in that path Nov 24 20:34:54 re Nov 24 20:59:13 here is another use for the usd versus euro situation .... Nov 24 20:59:19 http://www.cnn.com/2007/TECH/11/24/town.sells.ap/index.html Nov 24 21:05:26 CoreDump|bbl: hrw|gone: PING Nov 24 22:02:54 03xjqian 07org.oe.dev * rdd018c5b... 10/ (5 files in 2 dirs): mpfr: unify, remove obsolete directory in 2.3.0 staging. Nov 24 22:04:49 mwester: seems --with-mpfr=${STAGING_DIR_NATIVE}${layout_exec_prefix} is the guity part Nov 24 22:07:04 mwester: the apparent dependence of gfortran on host is because gfortran brings in libmpfr to the host to make up for the above Nov 24 22:07:27 mwester: I have to clean up my tmp to verify that, though Nov 24 22:18:34 xjqian, I have a clean tmp if there is a diff you need testing .... Nov 24 22:20:04 Crofton|home: my tmp is small, I just cleaned it. running now... thanks anyway Nov 24 22:20:14 ok Nov 24 23:14:04 * * OE Bug 3364 has been created by autobuild(AT)openembedded.org Nov 24 23:14:06 * * mpfr-native-2.3.0-r1-do_populate_staging Nov 24 23:14:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3364 Nov 24 23:47:31 * RP replaced pkgmaps with pkgdata in poky :) Nov 25 00:08:33 hi Nov 25 00:08:53 something new : ERROR: Build of /oe/org.openembedded.dev/packages/binutils/binutils-cross_2.18.bb do_compile failed Nov 25 00:08:59 ?? Nov 25 00:09:58 .../gas/config/tc-arm.c:1786: note: 'firsttype$defined' was declared here Nov 25 00:10:41 .../gas/config/tc-arm.c:1786: error: 'firsttype$defined' may be used uninitialized in this function Nov 25 00:11:11 | cc1: warnings being treated as errors Nov 25 00:12:03 shame...I was building overnight Nov 25 00:12:08 'nite Nov 25 00:22:18 xjqian: so, hows the gcc thing coming? Nov 25 00:29:21 CoreDump: still running here ... Nov 25 00:34:23 xjqian: what distro? Fedora gfortan does not have the mpfr package/library available, so it can never resolve that via LD_LIBRARY_PATH to the host. Nov 25 00:36:07 Re: autobuilder error and oebug 3364: Not much detail there. I thought it was supposed to attach the log or something. :( Nov 25 00:38:30 Hmm. INHERIT is evil... Nov 25 00:40:17 * mwester thinks of all the OO enthusiasts who would take issue with that statement ;) Nov 25 00:40:57 mwester: I'm speaking in an OE context ;-) Nov 25 00:43:55 * RP is playing with packaged staging Nov 25 01:20:11 mwester: I'm using debian lenny Nov 25 01:20:51 mwest: which version of mpfr do you have problem with 2.1.1? Nov 25 01:23:12 xjqian: The problem is that there is no mpfr on the host. "bitbake mpfr-native" executes just fine. "bitbake gcc-cross" fails, because somewhere in there, something to do with gfortran attempts to link with libmpfr.so.1, and fails. The version is not under my control. Nov 25 01:24:16 The one select is mpfr-native-2.3.0-r0 Nov 25 01:24:54 s/select/selected/ Nov 25 01:29:09 xjqian: Have you committed a fix for this? Nov 25 01:30:46 mwester: not yet, I just reproduced your error without libmprf (pulled in from gfortran) on host Nov 25 01:31:19 mwester: mpfr-native-2.3.0 workes perfect for me, I don't think there is a problem there Nov 25 01:32:49 mwester: can you remove --with-mpfr=${STAGING_DIR_NATIVE}${layout_exec_prefix} from the EXTRA_OECONF in gcc-cross_4.1.2.bb and test? Nov 25 01:32:57 no the problem is almost certainly in what's been done to the gcc-cross recipe. Can we revert that back to a version that doesn't have the gfortran stuff? Nov 25 01:33:27 The entire Openmoko project will be running into this as they all sync and attempt to build; some have already encountered this failure. Nov 25 01:34:59 I have hacked up my local workspace to get around this, so I can't test right now. I've lost the day because last night's build didn't finish, and I finally now have a build nearly done -- I'm reluctant to mess it up and lose another day. :( Nov 25 01:37:43 sure, I will disapprove that. sorry for the greef Nov 25 01:59:54 03xjqian 07org.oe.dev * r57de0451... 10/ (1 packages/gcc/gcc_4.1.2.bb): gcc_4.1.2: revert enabling fortran. Nov 25 02:37:31 sheesh, that "getting started - makefile" thread sure got large **** ENDING LOGGING AT Sun Nov 25 02:59:56 2007