* configure.ac re-organisation
* don't compile edje binaries on windows ce
* add doc rule to build documentation (hence gendo is removed,
Doxyfile is moved to doc/ and edje.c.in is renamed doc/edje.dox.in)
Next commit will fix completely the documentation
* fix edje.pc.in
SVN revision: 37137
* reorganize configure.ac
* remove eina dependency, use the new ac_attribute m4 macro
* fix Libs.private field and add Requires field in embryo.pc
* rename embryo.c.in to embryo.dox.in
* add doc rule to build the documentation. The doc will completely
be fixed in the next commit
SVN revision: 37116
* reorganize configure.ac
* select default opptions in configure.ac (disable openssl
for windows ce)
* fix Libs.private field and add Requires field in eep.pc
* rename eet.c.in to eet.dox.in
* use css file from other packages. Now the doc is correctly
displayd (need a rename of eet.css to e.css, though. will be
next commit)
SVN revision: 37114
* src/lib/evil_unistd.h:
add execvp stub function for ecore_app. Does nothing,
only here to make compilation working with mingw32ce.
SVN revision: 37102
On the let's break 'em all bandwagon, add the two callbacks (or
virtuals if you like the name) so we can know when some object is
attached or detached as a member of our object, this will help
eliminate confusing and duplicate functions like
evas_object_smart_clipped_smart_add().
Please recompile *ALL* your libs and applications that depend on Evas
and Smart Objects, like ETK.
SVN revision: 37084
- Array layout would be broken or worse not possible to reload at all if you put NULL pointer in it.
- Array of string now work.
- Fix a double free issue with array on double load.
* Add the test to detect this case.
SVN revision: 37051
"The attached patch fixes the ecore_path_group_available by removing the
senseless check on shared object suffix.
It also renames the function to ecore_path_group_available_get, to match the
ecore API."
SVN revision: 37031
* Add a test case for that one.
Problem was simple type inlining in complex structure (like list/hash/array) is
just a hack. We are creating a subtype with the data chunk. That work with INT
or SHORT for example, but not with STRING because it's a pointer to a STRING and
not directly the STRING. This result in a double pointer dereferencing where it
shouldn't. In fact STRING is not really like other simple data type. So we
should handle it differently.
Still need to fix Array and Hash.
SVN revision: 37024
We usually want to create an Ecore_Evas and attach an object to it, be
it the background, your smart object that will manage the scene (ie:
edje) and this is replicated everywhere. Not anymore!
ecore_evas_new() and ecore_evas_object_associate() will behave much
like regular toolkits "window-new()" and "window-main-child-add()",
actually it was based on elm_win.c and hopefully we can remove that,
or most of that code and replace with this helper.
I'll add an Evas smart object to handle stacks of objects, that is, it
will be a clipped smart object that on resize it will resize every
child to the same size. This means we can associate this stack object
and add a background and then your stuff on top of it.
SVN revision: 37010
* src/lib/Makefile.am:
store cpp flags in a variable instead of using an
automake conditional
* src/bin/Makefile.am:
* src/bin/evil_suite.c:
* src/bin/evil_test_memcpy.c:
add test for getenv / setenv. More tests on it will
come later
* src/lib/evil_stdlib.c:
fix getenv code and minor cleanup
SVN revision: 37004