**** BEGIN LOGGING AT Tue Dec 10 02:59:58 2013 Dec 10 09:29:32 good morning Dec 10 09:30:47 otavio: thx, do you plan to submit any patch for the missing branch from git SRC_URI today? Dec 10 09:42:57 /go devi Dec 10 09:43:00 oops Dec 10 10:25:26 ant_work: why is the 'require klibc.inc' at the bottom of the klibc_2.0.2.bb? Dec 10 10:25:40 ant_work: it makes it impossible to conform to the OE style guide :( Dec 10 10:27:07 koen: because it is shared Dec 10 10:27:22 is used by klibc and by klibc utils Dec 10 10:28:04 I admit it is not the best example ;) Dec 10 10:28:05 koen: "guidelines" Dec 10 10:28:15 Dec 10 10:31:01 morning all Dec 10 10:32:17 ant_work: you're explaining why it's a .inc, not why the require is at the *bottom* of the file, instead of the top Dec 10 10:33:48 see, I remember I had to add some redefinitions *after* the require but that's all. Dec 10 10:46:06 koen: by updating to 2.0.3 I'll try to arrange it better (and remove the PR) Dec 10 10:47:07 koen: by the way, note that gcc-cross is atm not safe for multimachine-builds of the same arch Dec 10 10:47:16 err..klcc-cross ofc Dec 10 10:50:27 multimachine builds are not support by OE-core Dec 10 10:51:07 if they were, I wouldn't get millions of "OMG!!! openssl for arm tried to overwrite the license file for openssl for i586!!" warnings for each and every recipe Dec 10 10:53:05 apart klcc-cross, I don't see particular issues in my limited multim. builds Dec 10 10:54:37 multimachine issues have been generally easy to work around Dec 10 10:54:58 * koen remembers doing those in the shared-sysroot days Dec 10 10:55:11 first build armv4, then v4t, then v5te and finally v6 Dec 10 10:55:14 mine is more serious: the lack of an arch-sysroot Dec 10 10:55:20 otherwise the wrong libc would end up being used Dec 10 10:58:15 if you see, klcc-cross contains the paths to the sysroot of the last machine it was built for Dec 10 11:05:01 hey guys, silly question ... has "wifi" moved from image features to distro features in the past 9 months ? Dec 10 11:05:16 (haven't pulled my build box since then ...) Dec 10 11:06:58 (or to "packageconfig" maybe ?) Dec 10 11:47:05 mckoan: no; I am awaiting someone to do it. Dec 10 11:47:15 mckoan: if you have the time, please send. Dec 10 11:47:32 mckoan: John has send for Wandboard and U-Boot, which I will merge Dec 10 12:16:29 bencoh: I don't think it has... Dec 10 12:21:09 I can't specify "wifi" in image_features anymore, I guess I missed something at some point Dec 10 12:25:29 bencoh: I don't think that's ever been a valid image feature... we did add a check in the last release for invalid features Dec 10 12:26:50 oh ... I see ... :] Dec 10 12:39:27 koen: before updating klibc to 2.0.3 I wanted to see if it is possible (now) to avoid staging the mtd headers, once necessary for mtd-utils Dec 10 12:40:00 but this can be done later as well, so updates to 2.0.3 are welcome ;) Dec 10 12:41:39 * ant_work busy atm with collie's hostic NOR-flash pair Dec 10 12:47:56 otavio: I have no idea where is the new repo and branch for sabre boards Dec 10 12:48:30 otavio: I found a branch called ...wandboard Dec 10 12:52:42 mckoan: he sent it for wandboard Dec 10 12:53:05 mckoan: the repository does not change. You must find the branch used and put it in SRC_URI Dec 10 13:00:22 otavio, do you need updates for nitorgen6x for meta-fsl-extra? Dec 10 13:00:34 Crofton|work: yes Dec 10 13:00:38 k Dec 10 13:00:47 I need to do a build test, but having the branch issue Dec 10 13:00:52 no hw to run test though Dec 10 13:00:54 Crofton|work: you can do a fetch, I did push the wandboard there Dec 10 13:00:55 that OK? Dec 10 13:01:01 Crofton|work: sure Dec 10 13:01:03 k Dec 10 13:46:47 strange. sometimes when building from scratch (using poky master) m4 configure does not ever stop Dec 10 13:48:14 dv_: I was there (Gentoo 32b) Dec 10 13:48:17 So all was originated by perl on buildhost compiled without threads. Dec 10 13:49:22 ant_work: so m4 needs to be built with -j1 ? Dec 10 13:49:45 I've rebuilt gentooƬs perl with -ithreads ;) Dec 10 13:50:06 so the host's perl is the culprit Dec 10 13:51:42 http://lists.openembedded.org/pipermail/openembedded-core/2013-October/085721.html Dec 10 13:51:58 ^^ Dec 10 13:54:56 dv_: on another build I could see gettext-native failing for the same reasomn Dec 10 13:54:59 | DEBUG: Executing shell function do_configure Dec 10 13:54:59 | This Perl not built to support threads Dec 10 18:16:20 otavio, I sent some pacthes to the meta-fsl list Dec 10 18:16:39 I was going for minimum change to make it build again Dec 10 19:49:26 koen: ping? Dec 10 20:41:37 Crofton|work: Can you use a 'SRCBRANCH' to hold the branch for the SRC_URI use? Gary has asked for it and your patches are split upon each recipe ... so I prefer yours Dec 10 20:42:19 as is use a var for the branch? Dec 10 20:43:27 ;branch=${SRCBRACNH} \" and set SRCBRACNH? Dec 10 20:45:13 also, there is a commit that uses GITBRANCH as the name Dec 10 20:47:20 yes Dec 10 20:47:43 If you want, send one to rename GITBRANCH to SRCBRANCH Dec 10 20:47:53 this was before we agreed on this Dec 10 20:49:31 Crofton|work: you did a typo in your comment (but you got the idea) heh Dec 10 20:53:43 yeah Dec 10 20:53:50 ok Dec 10 20:53:53 SRCBRANCH Dec 10 20:56:12 :) Dec 10 21:50:06 otavio, they parse and seem to build Dec 10 23:22:09 how can I force OE to never download kernel tarballs from the mirrors? Dec 10 23:22:19 if git fetch fails, fetch in general shall fail Dec 10 23:23:31 well, one option would be to change your mirrors to not include git repositories. see mirrors.bbclass. but there might be a variable to disable it wholesale, not sure Dec 10 23:23:59 okay, Dec 10 23:24:05 also, I cloned the repo manually Dec 10 23:24:38 it should be placed in $DL_DIR/git2 , and .done files should be placed in $DL_DIR and $DL_DIR/git2 to let OE know it does not have to clone it, right? Dec 10 23:25:39 it has to be a bare cone, and named correctly Dec 10 23:25:41 but otherwise yes Dec 10 23:26:10 but it might be easier to just use PREMIRRORS to tell it to clone from your existing local clone, rather than trying ot poke into bitbake fetch internals :) Dec 10 23:29:56 "Failed to fetch URL git://git.freescale.com/imx/linux-2.6-imx.git, attempting MIRRORS if available" Dec 10 23:29:59 every time Dec 10 23:30:09 any way on getting more details about the failure? Dec 10 23:30:37 uh wait. forgot about log.do_fetch Dec 10 23:30:58 a-ha! the recipe has an invalid revision Dec 10 23:33:58 otavio: check out the revision in linux-imx_3.0.35.bb Dec 10 23:34:26 I cant find it in the fsl linux-imx repo Dec 10 23:53:54 PREMIRRORS_prepend = " git://git.freescale.com/imx/linux-2.6-imx.git file:///opt/mirror/git.freescale.com.imx.linux-2.6-imx.git " is this acceptable? I only want to redirect this one link Dec 10 23:56:13 should do, in theory. you forgot the \n separator at the end, though Dec 10 23:56:30 oh. right. thanks Dec 10 23:56:43 np Dec 10 23:58:37 ah I think I see what is going on. for unknown reasons, eventually fetch executes git branch --contains bdde708ebfde4a8c1d3829578d3f6481a343533a --list master . but that commit is not in master - it is in a different branch Dec 11 00:01:44 recent change to the fetch code, branch= must be specified if the srcrev isn't included in master Dec 11 00:01:51 branch has to line up with srcrev, now Dec 11 00:02:07 there've beeen multiple patches on teh list to fix some of those, i'm not sure if they've all been merrged Dec 11 00:02:11 s/list/lists/ Dec 11 00:02:22 ah Dec 11 00:02:27 that explains it Dec 11 00:02:51 I now modified the srcrev to explicitely contain the srcrev, and it seems to work. it is unpacking Dec 11 00:03:26 thanks! one more remark, though : I noticed in my yocto DL dir there are several tarballs and clones of kernel repos. each one of them occupying ~800MB Dec 11 00:03:48 what do you think of using --single-branch for the kernel? **** ENDING LOGGING AT Wed Dec 11 02:59:59 2013