**** BEGIN LOGGING AT Wed Dec 18 02:59:59 2013 Dec 18 09:36:22 morning all Dec 18 09:45:26 hi Dec 18 09:45:42 good morning Dec 18 10:27:15 does anyone know an Enlightenment distro meta layer? meta-efl doesn't provide any... Dec 18 10:27:41 why should it have separate DISTRO? Dec 18 10:34:25 doesn't angstrom have a E image? Dec 18 10:42:43 it has meta-angstrom/recipes-images/angstrom/efl-nodm-image.bb, SHR also has Enlightenment based images Dec 18 11:41:58 hi Dec 18 11:42:08 asked on yocto but nobody replied so trying here;0 Dec 18 11:42:11 hi all, is possible to overload pkg_postinst_${PN} in bbappend file (I need to overload something in nsswitch.conf) Dec 18 11:42:21 it seems it's not called when building imag Dec 18 11:49:11 mbelisko: you can set it or _append to it in the bbappend yes Dec 18 11:49:47 mbelisko: but if the original part gets deferred (e.g. because it explicitly exits when running during do_rootfs) of course your part will be deferred as well Dec 18 11:51:28 bluelightning: ok thanks Dec 18 11:51:38 I'm trying to debug do_rootfs output Dec 18 13:04:43 HI, I got this error mit meta-efl : Enlightenment found Evas can't load SVG files. Check Evas has SVG loader support. But evas-generic-loader seems to include svg support. Why is it missing ? Dec 18 13:06:29 you need to install evas-generic-loader-svg if you don't have it already Dec 18 13:07:12 isn't it included by packagegroupe-x11-illume dependencies ? Dec 18 13:11:20 JaMa, btw, I had an error about evas_font_load.c freetype header include. Is this known and gonna be fixed in meta-efl ? Dec 18 13:14:59 it's included in e-wm dependencies Dec 18 13:15:16 oswin: see jansa/efl branch in meta-oe-contrib, I plan to upgrade efl to 1.8.2 soon Dec 18 13:16:01 is it yet stable ? Dec 18 13:16:45 no Dec 18 13:17:02 the efl should be stable, but the recipes aren't ready yet Dec 18 13:18:32 bluelightning: thanks for e-mail about parsing error it should be fixed already Dec 18 13:18:50 bluelightning: it was very unlucky timing, that problem was there only for few minutes :) Dec 18 13:19:05 JaMa: ah ok, np Dec 18 13:19:45 JaMa, any idea on what could be wrong to get SVG error even if evas-generic-loader-svg is included ? Dec 18 13:21:00 JaMa, cool for 1.8, will be nice :) Dec 18 13:21:21 oswin: nope Dec 18 13:40:33 could solve the svg problem by changing this line https://github.com/openembedded/meta-oe/blob/master/meta-efl/recipes-efl/efl/evas_1.7.9.bb#L49 to += so it don't override the evas-loader-generic Dec 18 13:50:18 oswin: looks correct, can you send patch? Dec 18 14:22:39 JaMa, yep Dec 18 14:24:31 JaMa, argh not allowed Dec 18 14:25:10 ? Dec 18 14:25:27 "You are not allowed to post to this mailing list, and your message has Dec 18 14:25:27 been automatically rejected" Dec 18 14:25:50 I sent to openembedded-devel@lists.openembedded.org Dec 18 14:26:09 you need to subscribe first Dec 18 14:28:34 yes, and subscribe at http://lists.openembedded.org/mailman/listinfo/openembedded-devel Dec 18 14:28:53 if the email response is still telling you to use linuxtogo.org, it's wrong Dec 18 14:29:04 (just noticed this the other day) Dec 18 14:29:23 I believe Michael will fix it soon if he hasn't already Dec 18 14:36:35 cut and paste error? Dec 18 14:43:28 Crofton|work: I guess so Dec 18 14:55:31 Is this known : Enlightenment found Evas can't load the 'Sans' font. Check Evas has fontconfig support and system fontconfig defines a 'Sans' font. ? Dec 18 14:58:15 #define HAVE_FONTCONFIG 1 is set in evas config, and I don't see any contradiction to this in evas recipe Dec 18 14:58:39 recipe event tells --enable-fontconfig Dec 18 14:59:48 even* Dec 18 15:06:26 you're probably missing fonts Dec 18 15:23:02 rburton, mmh you are right, which recipe could provide me fonts ? Dec 18 18:01:25 stupid rpath Dec 18 18:41:10 it looks like the db recipe does not recreate configure from configure.ac Dec 18 18:48:30 Crofton|work: right, looking at it I'm guessing doing that would trash something custom in the supplied configure Dec 18 18:48:56 which shouldn't really happen, but I guess there is some brokenness upstream in this case Dec 18 18:50:34 yeah Dec 18 18:50:51 which is making it hard to patch out the rpath crap Dec 18 18:51:08 not having any luck rebuilding locally Dec 18 18:55:24 Crofton|work: er, but db itself doesn't have rpath problems does it? Dec 18 18:55:40 opnly due to blind luck I think Dec 18 18:55:49 the c++ one does Dec 18 18:55:56 since it links against libstdc++ Dec 18 18:57:08 looks at the build log, it is hard t see how db avoided the problem Dec 18 18:59:43 I fear I need to patch configure Dec 18 19:32:01 * rburton passes Crofton|work a bottle of whisky Dec 18 22:16:50 -SOFLAGS= -rpath $(libdir) Dec 18 22:17:05 can I do that, or does the -rpath arg need replacing? Dec 18 22:22:32 how can I prepend only for the main recipe, not native or sdk? Dec 18 22:23:06 there's a 'target' class override now Dec 18 22:23:11 _class-target Dec 18 22:23:15 ok Dec 18 22:23:23 I wa going to guess that based on grep :) Dec 18 22:23:27 alternatively, you can always do it through an intermediate variable, too Dec 18 22:23:38 _prepend = "${FOO}" FOO = "bar" FOO_class-naitve = "baz" Dec 18 22:23:45 but that shoudlnt' be needed with rrecent yocto Dec 18 22:36:01 | libtool: install: cp -p .libs/libdb-5.3.lai /home/balister/src/oe-core/build/tmp-eglibc/work/armv7ahf-vfp-neon-oe-linux-gnueabi/db/5.3.21-r0/image/usr/lib/libdb-5.3.la Dec 18 22:36:01 | cp: cannot stat '.libs/libdb-5.3.lai': No such file or directory Dec 18 22:36:02 wtf Dec 18 22:38:11 don't you just love libtool? Dec 18 22:40:24 yeah Dec 18 22:40:43 all i want to do is get the bogus rpath out Dec 18 22:41:03 * mr_science stabs libtool in the left eye with a blunt pencil Dec 18 22:41:45 i'd use a blunt sliderule but it's behind the emergency glass... Dec 18 22:41:49 SOFLAGS= -rpath $(libdir) Dec 18 22:41:55 thisis the offensive line Dec 18 22:43:23 tried overriding with -rpath-link ? Dec 18 22:43:53 where would I pass that? Dec 18 22:44:07 that was getting so common it's in my standard makefile template... Dec 18 22:44:10 that woudln't work, -rpath and -rpath-link are separate Dec 18 22:44:36 that is, it won't prevent you from getting a qa failure due to the rpath still being in the binary Dec 18 22:44:41 yeah, but -rpath-link should be able to pass the sysroot and get a round hard-coded rpath Dec 18 22:44:58 ah, i don't hit that in oe-classic Dec 18 22:45:47 guess i'll lose some workarounds when i migrate to poky... Dec 18 22:46:11 you won't need that one, since rpath isn't allowed to get into the binaries in the first place Dec 18 22:46:16 kergoth: then the solution is to rebuild the offending lib/binary? Dec 18 22:55:16 how the fuck does libtool decide to instal a foo.lai file that does not exist Dec 18 22:57:07 because it's evil... Dec 18 22:57:17 what package are you building? Dec 18 22:57:23 db Dec 18 22:57:30 but trying to add c++ suppport Dec 18 22:57:38 trying to beat down a qa error Dec 18 23:08:03 which version of db? Dec 18 23:12:23 whatever is on oe-core Dec 18 23:13:11 looking at db-4.8-libtool.patch Dec 18 23:13:33 all it does is add quotes around the $LD var in a bunch of places Dec 18 23:13:58 it patches configure actually Dec 18 23:17:50 in oe-core? Dec 18 23:17:57 new? Dec 18 23:17:57 no, gentoo Dec 18 23:18:00 ah Dec 18 23:18:23 apparently it is painful to rebuild configure from onfigure.ac Dec 18 23:18:38 that's what the gentoo package does Dec 18 23:18:41 looks like when I drop the rpath, the lai file is not creted Dec 18 23:18:51 eautoreconf , libtoolize, etc Dec 18 23:19:10 I can't autoreconf Dec 18 23:19:57 if the recipe doesn't do that, then your options are pretty much get rid of the -rpath stuff Dec 18 23:20:10 I did reomve it from the MAkefile Dec 18 23:20:11 s/options/option/ Dec 18 23:20:28 you could try making it an -rpath-link instead Dec 18 23:20:46 depends on the context i guess... Dec 18 23:21:07 this is not something I am good at :) Dec 18 23:21:21 nobody is... Dec 18 23:21:54 i've just been fixing that kind of crap in gentoo for about 10 years Dec 18 23:22:20 little over a year in oe-classic Dec 18 23:29:02 Crofton|work: does it pass --with-sysroot in configure? Dec 18 23:30:19 some packages have a --disable-libtool option Dec 18 23:30:22 only if oe_runconf does so Dec 18 23:30:32 apparently this does not Dec 18 23:30:43 have a --disable-libtool option... Dec 18 23:33:23 i can't see why it needs to do that Dec 18 23:33:57 SOFLAGS="-rpath \$(libdir)" <= just not a good idea... Dec 18 23:34:56 try setting hardcode_automatic=no ? Dec 18 23:37:31 actually, hardcode_action=unsupported looks like what you want Dec 18 23:38:10 that configure script looks crazy... Dec 18 23:49:54 yes, something nutty is going on Dec 19 00:19:31 Crofton|work: still there? Dec 19 00:20:22 what about something like "export ac_cv_hardcode_action=unsupported" ? **** ENDING LOGGING AT Thu Dec 19 02:59:59 2013