**** BEGIN LOGGING AT Mon Mar 17 02:59:58 2014 Mar 17 08:10:11 good morning all Mar 17 08:16:35 good morning Mar 17 09:26:19 hi Mar 17 09:26:57 I saw (in ML) that it's now possible to use _%.bbappend to match any version Mar 17 09:27:20 but I'm unable to see the date of this mecanism in oe-core or bitbake Mar 17 09:27:46 commit 537f1f9bbe110acc9848ef95f43468c07d87af79 Mar 17 09:27:46 Author: Jacob Kroon Mar 17 09:27:46 Date: Sat Feb 22 17:52:29 2014 +0100 Mar 17 09:27:46 ast: Fix support for anonymous methods in wildcard .bbappend files Mar 17 09:28:08 this is last fix for that Mar 17 09:28:12 wow :) Mar 17 09:28:14 so it's safer after this point Mar 17 09:29:01 searching for wildcard in bitbake git log will reveal other 3 commits touching it :) Mar 17 09:30:24 I seached for "any version" and "%_.bbappend" Mar 17 09:30:32 Anyway this is quite new Mar 17 09:50:27 Compile error in libtheora 1.1.1 on Dora: http://pastebin.com/1eW1fD5Z. Host is Ubuntu 13.04. Anyone with the same issue? Mar 17 10:13:05 morning all Mar 17 10:13:20 bluelightning, morning Mar 17 10:13:29 hi kroon Mar 17 10:29:30 ouch. user error. without the linaro layer it chose gcc 4.6 Mar 17 10:31:40 tasslehoff: where does 4.6 come from? OE-Core dora has 4.8.1 and 4.7.2... Mar 17 10:33:14 bluelightning: meta-oe/toolchain-layer Mar 17 10:33:26 in my angstrom-setup-scripts based config Mar 17 10:36:59 ah, I see Mar 17 10:37:21 I wonder if that should really be there anymore in dora if it leads to failures... Mar 17 10:47:07 bluelightning: yes. I suspect it is also the reason why I have problems getting Qt 5.2.1 to work Mar 17 10:51:35 bluelightning: are there any automated tools to check layer (inter)dependencies? Mar 17 10:52:21 koen: there's bitbake-layers show-cross-depends Mar 17 10:54:01 thanks! Mar 17 11:21:11 bluelightning: how does that handle things like PACKAGECONFIG ? Mar 17 11:22:09 koen: I doubt that it has any specific handling for PACKAGECONFIG, AFAIK it just uses whatever your current configuration is Mar 17 11:22:20 that's what I thought Mar 17 11:23:36 koen: is it particularly important? given that "soft" dependencies via default-disabled PACKAGECONFIG options have to be permitted, we have quite a few of them now Mar 17 11:24:14 I'm trying to get a picture of what all the linaro layers depend on so I can restructure them Mar 17 11:24:21 so in this case, no, it's not important Mar 17 11:24:28 but for e.g. meta-oe it would be Mar 17 17:27:49 I was looking for a recipe that showcased how to handle git submodules in SRC_URI. Is there a canonical example? Thanks. Mar 17 17:31:41 frobware: pretty sure its just a matter of using the gitsm fetcher instead of git, which should just be a matter of using gitsm://, but I've never tried that fetcher personally Mar 17 17:32:03 kergoth, yeah, I took a look at that... let me try again. Mar 17 17:34:51 kergoth: switching git:// to gitsm:// gives me: Mar 17 17:34:52 Could not switch to '/scratch/oe-14.02/jenkins-setup/downloads/git2/github.com.facebook.hhvm.git.tmp/hphp/submodules': No such file or directory Mar 17 17:36:07 kergoth, followed by "Unable to fetch in submodule path 'hphp/submodules/folly'" Mar 17 17:37:53 interesting. does git clone --recursive on the url outside of bitbake work fine? if so, i'd say it's just a bug in the gitsm fetcher, and would suggest checking the yocto bugzilla, iopening a new issue if one doesn't eixst for the issue Mar 17 17:40:30 kergoth: yes the recursive clone works fine out of bitbake. just cloned the tree in /tmp + the expected submodule ("Submodule 'hphp/submodules/folly' (git://github.com/facebook/folly.git) registered for path 'hphp/submodules/folly'") Mar 17 17:44:04 definitely sounds like a gitsm bug to me, or at the very least a usability problem Mar 17 17:46:02 kergoth, could be this: https://bugzilla.yoctoproject.org/show_bug.cgi?id=5525. will try building a newer version of git than I get with 12.04. Mar 17 17:51:10 ah, interesting Mar 17 17:51:17 sorry i wasn't much help, never tried it Mar 17 17:51:38 (honestly, i think our default git fetcher should default to pulling submodules unless opt-ed out, the repo has them for a reason, it likely isn't buildable without them anyway) Mar 17 17:52:04 kergoth: quite. less magic too. Mar 17 17:52:06 but I'm not a big fan of how our fetchers work in general at the moment Mar 17 17:53:45 kergoth: and less churn should submodules come and go. Mar 17 17:54:00 indeed Mar 17 17:54:18 i get why it was done this way, to avoid impacting existing users, but that impact would likely be for the best :) Mar 17 17:54:19 kergoth: and git v1.9.0 worked for me. Mar 17 17:55:00 ah, good to know, and good to hear, at least you're unblocked :) Mar 17 17:55:17 kergoth: yep, many thanks... Mar 17 17:55:44 kergoth: just need to update various++ 12.04 instances... bah! Mar 17 17:56:21 fun fun Mar 17 18:43:54 khem: ping Mar 17 20:23:24 hi Mar 17 20:23:37 khem: ping Mar 17 20:24:08 I am running bit bake -c menuconfig virtual/kernel and for some unknown reason I only see menuconfig started/menuconfig succeeded, but not configuration window Mar 17 20:24:10 any idea? Mar 17 22:27:30 is it possible to have a FDO-type filesystem structure (bin -> /usr/bin, sbin -> /usr/sbin, lib -> /usr/lib)? I have attempted to set base_{bin,sbin,lib}dir in my distro config but I'm getting a lot of errors Mar 17 23:03:45 aport: I know folks have set prefix= to flatten it in the other direction, dropping /usr, but i don't know of anyone doing it the other way around. i'm sure patches would be welcome to fix any breakage Mar 17 23:22:32 kergoth: thanks for the info. I'll keep playing with it. Mar 18 00:02:14 So the problem I was having was related to chrpath, which can't acommodate larger paths (/lib -> /usr/lib). I changed to patchelf and this succeeded. The kicker is that in this setup, rpath does not need to point to /usr/lib, as /lib is symlinked. I just need to install the packages to /usr/lib... **** ENDING LOGGING AT Tue Mar 18 02:59:58 2014