**** BEGIN LOGGING AT Fri Nov 16 02:59:58 2012 Nov 16 08:35:24 morning all Nov 16 08:55:36 hi Nov 16 09:04:40 yes Nov 16 09:17:18 hi afournier Nov 16 09:17:31 hi silvio Nov 16 09:26:04 morning all Nov 16 10:06:31 hi bluelightning Nov 16 10:06:51 hi silvio Nov 16 11:19:53 bluelightning: do you use mysql in images? Nov 16 11:20:02 https://bugs.launchpad.net/linaro-oe/+bug/1068759 Nov 16 11:21:06 hrw: I did notice this was a problem on first run but subsequent runs were OK - did your startup fix not help with this? Nov 16 11:21:25 it got broken in october Nov 16 11:21:31 my change is older Nov 16 11:22:27 reverted 415c49a0c7ba008ed26f6639dd57a96fa9b986e1 locally to check Nov 16 11:23:01 hrw: are you saying that that commit broke it? Nov 16 11:23:25 if so I can't see how Nov 16 11:23:41 no idea - just blindly checking Nov 16 11:24:57 as I had to find working image 2 weeks ago but forgot which one used finally Nov 16 11:33:34 1017 build works - with r9 Nov 16 11:38:31 1018 fails - also r9 ;( Nov 16 11:55:15 mysqld got moved from /usr/libexec/mysqld to /usr/lib/mysql5/mysqld Nov 16 11:55:28 I rmember that change... Nov 16 11:57:05 hi hrw Nov 16 11:57:21 and /usr/bin/mysqld_safe checks /usr/libexec/ Nov 16 11:57:28 now I know where to patch Nov 16 12:00:10 oh, right... so the breaking change was not in the mysql recipe itself Nov 16 12:03:34 yes Nov 16 13:20:26 ok, mysql now know where mysqld is Nov 16 13:20:33 need to get tables working too Nov 16 13:49:55 I am having an issue building a recipe that I created myself Nov 16 13:50:01 http://pastebin.com/iY37HTQY Nov 16 13:50:07 the bb file is at the bottom Nov 16 13:50:27 is it possibly because my git address has some odd characters in it, e.g. @? Nov 16 13:51:34 ok ignore, I forgot a $ sign on the {SRCREV} Nov 16 13:51:37 doh! :( Nov 16 13:54:54 is there a way to pass a username and password to the git command? Nov 16 13:55:04 in a bb file? Nov 16 14:06:16 or failing that is the a way to use git over ssh? Nov 16 14:08:47 jackmitchell: for the latter you can just add ;protocol=ssh to the end of the URI Nov 16 14:10:00 bluelightning: I got it Nov 16 14:10:06 bluelightning: literally just now ;) Nov 16 14:10:18 bluelightning: shouldn't he be appending ;protocol=git if he want to get git repo Nov 16 14:10:31 bluelightning: it didn't like the : in my url Nov 16 14:10:58 it had to go from: git://git@codebasehq.com:dbproducts/r0005/r0005-serial.git;protocol=ssh Nov 16 14:11:00 to git://git@codebasehq.com/dbproducts/r0005/r0005-serial.git;protocol=ssh Nov 16 14:11:17 notice the slash instead of : after the com Nov 16 14:22:36 in my do_configure I'm getting the following error, does anyone recognise what the issue might be? Nov 16 14:22:37 configure: WARNING: unrecognized options: --disable-silent-rules, --with-libtool-sysroot Nov 16 14:25:34 presumably you are overriding the default do_configure from autotools.bbclass? Nov 16 14:27:51 bluelightning: no, I think it may have been a red herring and it was something else failing Nov 16 14:28:04 bluelightning: how do I run do_configure from a devshell? Nov 16 14:29:22 I'm not sure, I have to admit it's not something I've ever tried to do Nov 16 14:34:32 ok, I see what it is, it is compiling for my host rather than for the target Nov 16 14:34:47 does autotools not work this stuff out automatically? Nov 16 14:35:45 which stuff? Nov 16 14:36:29 well it seems to have compiled my application using host gcc Nov 16 14:36:38 as the resulting object files are 64 bit Nov 16 14:37:25 normally it does yes Nov 16 14:37:34 oh, ok Nov 16 14:38:12 this is the error: http://ix.io/3pw Nov 16 14:39:43 jackmitchell: the question is what was the command used to build r0005-serial.o? Nov 16 14:39:46 ah-hah! my bad, I think I have added the object files to the repo Nov 16 14:39:59 so it isn't re-building them as armv7a binaries... Nov 16 14:40:01 ah right, heh... that would do it Nov 16 14:40:30 always pays to set up .gitignore first :) Nov 16 14:54:58 where are the debug symbols for libstdc++? Most libs have a -dbg package, but not seeing how this works for libstdc++ yet Nov 16 14:57:48 ok, hopefully last question today... how do I get the ncurses forms and menu libraries into my sysroot? Nov 16 14:58:36 I am DEPENDing on ncurses Nov 16 14:58:43 but that doesn't seem to be enough Nov 16 14:59:24 hello all Nov 16 15:01:56 I can see ncurses-libform in packages-split of ncurses but it won't recognise ncurses-libform in DEPENDS Nov 16 15:04:36 it seems to be created in a do_packages_split auto function Nov 16 15:04:59 DEPENDS is for build-time dependencies, thats a package therefore it's runtime Nov 16 15:05:47 bluelightning: ah so RDEPENDS? Nov 16 15:06:17 jackmitchell: you're talking the sysroot though, so you mean build time surely? Nov 16 15:08:02 bluelightning: well I need libform.so.5 in my sysroot for the application to run Nov 16 15:08:35 bluelightning: so it currently builds fine, but won't run on target due to missing libraries Nov 16 15:14:47 bluelightning: ah, it's ok - all fixed! Nov 16 15:15:11 jackmitchell: we use the word sysroot to mean the sysroot for the build, I would recommend not using it when you mean the root filesystem for the target... Nov 16 15:15:55 jackmitchell: what did you do to fix it? shared library dependencies are normally automatically picked up... Nov 16 15:22:08 bluelightning: yes, that was the issue I didn't realise that bitbake automatically picked up the dependancy Nov 16 15:22:28 bluelightning: so I re-build the whole image and checked in the sysroot to find the libform.s0.5.9 Nov 16 15:23:47 bluelightning: re: sysroot - I will take that in account! Nov 16 15:24:30 jackmitchell: just so people don't misunderstand :) Nov 16 16:04:00 i've noticed that there are two udev_182 recipes, one in meta and one in meta-openembedded Nov 16 16:04:10 is there any particular reason for this? Nov 16 16:04:30 and it seems that the udev recipe in meta-openembedded is broken Nov 16 16:04:47 the one in OE-Core is fairly new Nov 16 16:04:52 as it fails to build because it doesn't include local.rules Nov 16 16:05:10 I suspect the one in meta-oe has only ever been used in conjunction with systemd Nov 16 16:05:22 you're right the one in meta-oe should be dropped Nov 16 16:05:43 well my problem is that the one in meta-oe gets picked by default, and it fails to build Nov 16 16:06:01 :( Nov 16 16:06:09 you could mask it out with BBMASK Nov 16 16:06:22 that's obviously a hack though Nov 16 16:09:09 how come the one in meta-oe gets picked over the one in meta? i thought the oe-core layer has the highest priority by default Nov 16 16:10:27 also, the recipe in meta has a higher pr value Nov 16 16:11:39 dromede_: see BBFILE_PRIORITY in the layer.conf files Nov 16 16:11:51 i really think meta-oe's priority should be dropped below oe-core Nov 16 16:12:00 if a recipe exists in both, it's almost guaranteed that oe-core's is of higher quality Nov 16 16:12:41 meta priority is 5, meta-oe is 6 Nov 16 16:12:50 we discussed that already Nov 16 16:12:52 so i reckon that meta has a higher priority? Nov 16 16:12:53 http://lists.linuxtogo.org/pipermail/openembedded-core/2011-August/007387.html Nov 16 16:13:01 dromede_: no. you got it backwards Nov 16 16:13:04 oh Nov 16 16:13:11 higher number is higher priority Nov 16 16:13:21 got it, thanks Nov 16 16:13:44 ant_work: i think that bitbake behavior was as designed, but you'd have to ask RP to confirm Nov 16 16:13:49 but yes, it does seem kinda counterintuitive that meta-oe has a higher value than oe-CORE Nov 16 16:13:55 but i think lowering meta-oe's priority is ideal for now.. Nov 16 16:13:57 dromede_: np Nov 16 16:14:17 dromede_: you can override its priority variable from local.conf, fyi, if you need to Nov 16 16:14:49 we have (maybe) an issue here but the real culprit are the two udev recipes... Nov 16 16:15:22 i saw khem talk about removing the meta-oe recipe, on the mailing list Nov 16 16:16:11 I usually removed xserver-nodm-init as well when I was playing with that layer Nov 16 16:16:23 i'll use BBMASK for now Nov 16 16:18:41 kergoth: I'd much rather see the junk removed from meta-oe rather than just hacking around it by lowering meta-oe's priority Nov 16 16:19:01 kergoth: more on that, don't you find confusing that a layer called meta-oe sits under the meta-openembedded 'umbrella'? Nov 16 16:21:05 heh Nov 16 16:21:18 and there is i.e. a meta-gnome in the parent and in the child .. :/ Nov 16 16:21:46 sry, meta-gnome vs. recipes-gnome Nov 16 16:22:46 dromede_: of that, I only use meta-initramfs atm, not the whole layer Nov 16 16:24:05 bluelightning: I agree, but I don't think it's a hack. meta-oe is essentially a pile of recipes that aren't in oe-core. we can still use those recipes with its priority lower. Nov 16 16:24:14 i used BBMASK to solve my udev problem, for now Nov 16 16:24:18 * kergoth shrugs Nov 16 16:24:36 yes, seems is up to a distro to mask those sort of conflicts Nov 16 16:25:36 kergoth: I firmly believe meta-oe shouldn't be overlaying or appending anything that's in OE-Core Nov 16 16:26:06 if it does so that's effectively imposing policy Nov 16 16:27:19 Agreed. if it never does, then the priority is irrelevent anyway, so lowering it hurts nothing in an ideal world :) Nov 16 16:30:48 ugh, headache, /me goes to get something Nov 16 16:33:59 kergoth ;( Nov 16 16:34:20 what would worry me is that might reduce the pressure to remove the things that shouldn't be there Nov 16 16:37:17 fair point **** ENDING LOGGING AT Sat Nov 17 03:00:00 2012