On Fri, 13 Nov 2015 20:02:24 -0500 Conrad Knight <iestynap...@gmail.com> said:

> Sorry to bring this up again, but i ran into the same issue with the
> efl-1.16 release. During the elementary build process, the
> freshly-compiled edje_cc (from efl) crashes with a segfault. This time
> i wasn't able to finish building no matter what combination of CFLAGS
> i used.
> 
> I making sure there were absolutely no left-over efl-1.15 libraries
> left, i manually searched through /usr/local/lib/. Previously i had
> been relying on "make uninstall" in the previous version's source
> directory. I found an empty directory:
> /usr/local/lib/edje/modules/elm/v-1.15
> 
> After deleting this directory, the build completed without problems!
> Is it possible that edje_cc loads modules dynamically at build time,
> based only on the names of directories it finds? In this case, edje_cc
> is built first (with the efl libraries) and then used to build the
> elementary libraries, which apparently includes some of these modules.
> (After the build finished and i did a "make install" for elementary, i
> found a new edje/modules directory, elm/v-1.16.)
> 
> If this is indeed what caused the segfault, is this a bug in edje_cc?
> Trying to load a non-existent, or incompatible module? Or at least a
> problem with the logic in the build process/order?

edje_cc doesnt - but libedje can load modules.... but directory existence does
not mean a module loads. it needs a module.so inside to do anything... without
a backtrace we dont have enouhg info to go on - i have had only module dirs for
years. it will THINK there is an available module there if the dir is there...
but the load will fail... ie if the elm dir above exists AND the correct
version dir exists. in your case it would have had to be v-1.16 as that is what
edje will look for so as such the available module list should not have been
affected edje_available_modules_get() will do the above. otherwise the
actual .so has to be there to load ... in the RIGHT version dir.

> -Conrad.
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> enlightenment-users mailing list
> enlightenment-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/enlightenment-users
> 


-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    ras...@rasterman.com


------------------------------------------------------------------------------
_______________________________________________
enlightenment-users mailing list
enlightenment-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/enlightenment-users

Reply via email to