platform/upstream/efl.git
9 years agoEvas masking: Fix some garbage pixels with the SW engine
Jean-Philippe Andre [Mon, 9 Feb 2015 12:16:06 +0000 (21:16 +0900)]
Evas masking: Fix some garbage pixels with the SW engine

A rare case of garbage data would happen if smooth scaling
was called with a mask and 1:1 scaling. Use the proper
render_op to COPY for the first pass.

@fix

9 years agorelease: Update NEWS and bump version for 1.13.0 release v1.13.0
Stefan Schmidt [Mon, 9 Feb 2015 12:12:51 +0000 (13:12 +0100)]
release: Update NEWS and bump version for 1.13.0 release

9 years agoevas - gl engine - pipe renderer - disabling of unused arrays for lines
Carsten Haitzler (Rasterman) [Mon, 9 Feb 2015 17:28:36 +0000 (02:28 +0900)]
evas - gl engine - pipe renderer - disabling of unused arrays for lines

@fix - this fixes disabling of unused texa and texsam arays when line
drawing - spotted in code review and bug hunting of another bug.

9 years agoevas - gl enigne - mask logic fix with texa draw arrays
Carsten Haitzler (Rasterman) [Mon, 9 Feb 2015 17:27:00 +0000 (02:27 +0900)]
evas - gl enigne - mask logic fix with texa draw arrays

this fixes a logic oops in addition of masking to evas which caused
crashes with a null texa array causing a crash in gldrawarrays

9 years agoEvas GL: Fix GLES1.1 extension initialisation bug
mythri.venugopal [Mon, 9 Feb 2015 05:23:29 +0000 (14:23 +0900)]
Evas GL: Fix GLES1.1 extension initialisation bug

Summary:
Extension function pointer initialisation requires glGetString(GL_EXTENSIONS).
To get GLESv1 extension string, GLESv1 context has to be bound.
Change involves updating extensions after GLESv1 context has been bound.

Reviewers: jpeg

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1946

Signed-off-by: Jean-Philippe Andre <jp.andre@samsung.com>
9 years agoefl: interface are still clearly in beta.
Cedric BAIL [Sat, 7 Feb 2015 12:46:12 +0000 (13:46 +0100)]
efl: interface are still clearly in beta.

9 years agoevil: fix initialisation.
michelle legrand [Sat, 7 Feb 2015 12:45:05 +0000 (13:45 +0100)]
evil: fix initialisation.

Eina_Log is not available at this point of the initialisation.

9 years agoUpdate russian translation
Igor Murzov [Sat, 7 Feb 2015 12:38:49 +0000 (15:38 +0300)]
Update russian translation

9 years agoforgotten ee bit from previous commit
Mike Blumenkrantz [Fri, 6 Feb 2015 23:25:01 +0000 (18:25 -0500)]
forgotten ee bit from previous commit

9 years agoecore evas should not crash when unsetting pointer
Mike Blumenkrantz [Fri, 6 Feb 2015 22:54:08 +0000 (17:54 -0500)]
ecore evas should not crash when unsetting pointer

ref D812

disappointed.jpg

9 years agoEvas masking: Add forgotten file in previous commit
Jean-Philippe Andre [Fri, 6 Feb 2015 03:35:12 +0000 (12:35 +0900)]
Evas masking: Add forgotten file in previous commit

9 years agoEvas masking: Also fix map masking with non-bgra images
Jean-Philippe Andre [Fri, 6 Feb 2015 03:31:24 +0000 (12:31 +0900)]
Evas masking: Also fix map masking with non-bgra images

9 years agoecore-drm: Remove unused variable
Chris Michael [Thu, 5 Feb 2015 16:21:36 +0000 (11:21 -0500)]
ecore-drm: Remove unused variable

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-drm: Fix issue with ecore_drm_output_crtc_Find sending back an
Chris Michael [Thu, 5 Feb 2015 16:06:41 +0000 (11:06 -0500)]
ecore-drm: Fix issue with ecore_drm_output_crtc_Find sending back an
invalid id. Fix T2086

Summary: In order to avoid getting returned an invalid id, we should
not be checking the currently active encoder and crtc combination
because outputs may have changed, encoders/crtc combinations could
have changed, etc, etc.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore_cocoa: Handle more special characters and capital characters.
Michelle Legrand [Thu, 22 Jan 2015 10:24:42 +0000 (11:24 +0100)]
ecore_cocoa: Handle more special characters and capital characters.

This improve the situation and only affect Ecore_Cocoa code base by
detecting a useful charactere.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: add ecore event on mouse scroll for cocoa.
Michelle Legrand [Thu, 29 Jan 2015 12:56:49 +0000 (13:56 +0100)]
ecore_cocoa: add ecore event on mouse scroll for cocoa.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: replace "keyname" by "key".
Michelle Legrand [Wed, 28 Jan 2015 14:11:01 +0000 (15:11 +0100)]
ecore_cocoa: replace "keyname" by "key".

Ecore_Event_Key attribute "key" cannot be NULL. This
raise a segmentation fault in Enventor when hitting on
modifier keys.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: create a log domain and use Eina_Log.
Nicolas Aguirre [Thu, 29 Jan 2015 16:08:23 +0000 (17:08 +0100)]
ecore_cocoa: create a log domain and use Eina_Log.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: don't send mouse event with negative x or y values.
Nicolas Aguirre [Thu, 29 Jan 2015 16:04:28 +0000 (17:04 +0100)]
ecore_cocoa: don't send mouse event with negative x or y values.

With cocoa you may have negatives values when Mouse Down or Up. This
changes fix this behavior by sending mouse event only if x and y are
inside the ecore_evas space.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: add new events type for multiple window support.
Romain Perier [Thu, 15 Jan 2015 15:45:32 +0000 (16:45 +0100)]
ecore_cocoa: add new events type for multiple window support.

This commit adds a method in Ecore_Cocoa_Window to be able to retrieve an unique
identifier for each window. It adds new events type and modify existing ones to
pass these windows identifiers through the event call chain. Resize, GotFocus and
LotFocus events are also updated to pass these identifiers to event handlers.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: use the right Cocoa notification handlers for focus events
Romain Perier [Thu, 15 Jan 2015 16:20:16 +0000 (17:20 +0100)]
ecore_cocoa: use the right Cocoa notification handlers for focus events

Don't use NSAppKitDefined events subtype for focus events, which contain NULL
window object most of the time. Use the NSWindowDelegate method designed for that
purpose instead. It fixes random focus issues in windows which was caused by
incorrect window identifier not found in ecore_evas_cocoa.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: add on the fly resizing support.
Romain Perier [Thu, 15 Jan 2015 16:05:29 +0000 (17:05 +0100)]
ecore_cocoa: add on the fly resizing support.

Until now, video_resize events was received only when application returned control
to the event loop. When a window is resized dynamically a lot of video_resize
events are emitted from EcoreCocoaWindow::windowDidResize and not handled
immediatly, only when the main thread is back to the ecore main loop. This is why
there are not refreshed window areas. Call ecore_main_loop_iterate() from
windowDidResize solves the issue.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: fix _nsevent_window_is_type_of() type check
Romain Perier [Tue, 23 Dec 2014 16:51:51 +0000 (17:51 +0100)]
ecore_cocoa: fix _nsevent_window_is_type_of() type check

isKindOfClass should receive an instance of a class and a class as arguments and
not two class. It fixes Mouse events sent to a window on OSX.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_cocoa: handle more special characters and capital characters.
Michelle Legrand [Thu, 22 Jan 2015 10:24:42 +0000 (11:24 +0100)]
ecore_cocoa: handle more special characters and capital characters.

This patch is self contained for Mac OS X and improve its keyboard handling.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: fix eina_test_rwlock test on OSX.
pierre lamot [Wed, 21 Jan 2015 11:04:37 +0000 (12:04 +0100)]
eina: fix eina_test_rwlock test on OSX.

add a reimplementation of clock_gettime for OSX within the test in
order to be able to test eina_condition_timedwait

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore: check that a test doesn't enter in an infitie loop.
pierre lamot [Wed, 21 Jan 2015 10:27:56 +0000 (11:27 +0100)]
ecore: check that a test doesn't enter in an infitie loop.

check that ecore_test_ecore_thread_eina_thread_queue_t6 doesn't
enter in an infinite loop if something goes wrong, if the main thread
doesn't receive the message number 10000

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: fix semaphore initialisation
pierre lamot [Wed, 21 Jan 2015 10:05:41 +0000 (11:05 +0100)]
eina: fix semaphore initialisation

This patch fix the semaphore initialisation on posix plateform

there was a preconditon test which returned the semaphore
uninitialised when count_init was 1, this caused the semaphore to
be unusable on OSX plateform. Furthermore, it seems that we need to
unlink the semaphore before its initialisation on OSX as there seems
to have some kind of persistence of the semaphore accross execution.

warning, this patch change the signification of the parametter
count_init on linux plateform, this parametter is now consistent on
every plateform, with the meaning of setting the initial count value
of the semaphore.

This used to be on linux 1 -> the semaphore is shared and initialised
at 1 and 0 -> the semaphore is not initialised, thus, by side effect
not shared and initialised at 0.

This patch set on linux plateform the semaphore as systematically
shared

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevas: jpeg decoders alows a difference of 1 bit per component.
pierre lamot [Tue, 20 Jan 2015 14:39:08 +0000 (15:39 +0100)]
evas: jpeg decoders alows a difference of 1 bit per component.

The required precision of decompressed images allows a difference of
1 bit for each pixel compoment [1] . Such difference has been noticed
on OSX when using libjpeg9 from macports.

evas_suite images tests has been modified to compare jpeg images with
this tolerance. Other image formats are still compare with exact
precision

[1] http://en.wikipedia.org/wiki/JPEG#Required_precision

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore: fix path comparison in Ecore_File test suite.
pierre lamot [Thu, 11 Dec 2014 13:09:06 +0000 (14:09 +0100)]
ecore: fix path comparison in Ecore_File test suite.

The Ecore_File test suite was comparing path with strcmp,
which can lead to incorect comparison in some case (comparing
realpath and expected path for instance)

This patch adds a function to compare paths within this tests. Paths
are compared by getting the "realpath" of each operand

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: memcmp can't be used to compare timeval due to potential padding on 64bit archi...
pierre lamot [Thu, 11 Dec 2014 10:21:50 +0000 (11:21 +0100)]
eina: memcmp can't be used to compare timeval due to potential padding on 64bit architecture.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoedje: epp - fix build of edje file in cas of recursive includes.
Nicolas Aguirre [Tue, 3 Feb 2015 09:44:11 +0000 (10:44 +0100)]
edje: epp - fix build of edje file in cas of recursive includes.

Amitesh Singh fixed this issue with Commit
e66e132d9dabecb8273fe3962099d4523fef1055, but this commit has been reverted
recently, by a commit freeing ressources
(6d0b834a76c3301bf4c4b29752f902dbe8a8525e). It's clear that we are leaking here
if we remove the free, but there is also a valgrind error when this memory is
freed. After trying to debug and understand the code of epp, i can only
propose to remove this free, we will need to include a lot of files before this
memory leak could be visible.

It is also not a real problem, the memory leak only concern the file name passed
in the include directive and epp is started per file anyway. It is unlikely to
create any real life problem and we better spend our time on more pressing matter.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevas: Evas_3D - fix order of transformation while updating bounding object.
Oleksandr Shcherbina [Thu, 5 Feb 2015 14:25:29 +0000 (15:25 +0100)]
evas: Evas_3D - fix order of transformation while updating bounding object.

Summary:
The order of tranformation changed to scale, orientation, position as
in some cases it can lead to incorrect value for the bounding box.

@fix

Reviewers: cedric, Hermet, raster

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1942

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevas: Evas_3D - fix evas_3d_scene_pick when trying to get empty texcoords from mesh.
Oleksandr Shcherbina [Thu, 5 Feb 2015 14:17:43 +0000 (15:17 +0100)]
evas: Evas_3D - fix evas_3d_scene_pick when trying to get empty texcoords from mesh.

Summary:
During filling evas pick public data by API evas_3d_scene_pick
segfault can occur if mesh was created without texcoords.
See functions - _pick_data_mesh_add, _pick_data_texcoord_update

@fix

Reviewers: cedric, Hermet, raster

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1941

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeldbus: eina_value not allocated from mempool.
Guilherme Lepsch [Thu, 5 Feb 2015 14:15:37 +0000 (15:15 +0100)]
eldbus: eina_value not allocated from mempool.

Summary: Eina_Value must be allocated with eina_value_new() to be freed by eina_value_free().
@fix

Reviewers: felipealmeida, larryolj, cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1927

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevas: Evas_3D - fix when bounding sphere is large, situation with incorrect frustum...
Bogdan Devichev [Thu, 5 Feb 2015 14:06:10 +0000 (15:06 +0100)]
evas: Evas_3D - fix when bounding sphere is large, situation with incorrect frustum calculation was possible.

Summary: When center of bounding sphere was out of frustum and number of intersections between bs and planes of frustum was more then 3, objects disappeared.
@fix

Reviewers: Hermet, raster, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1938

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: Eina_File - add test case for path resolution.
Daniel Hirt [Thu, 5 Feb 2015 14:01:12 +0000 (15:01 +0100)]
eina: Eina_File - add test case for path resolution.

Summary:
This is a follow-up to the 7cbf22212f0c3292dbd1919f4cdd15802e9b9d84
revert commit. It's a bit late, as the issue has already been fixed with
the later 7143bd7fb5196f97003216528bfbfb8bec4826bd commit.

This test checks eina_file_sanitize "../" case with absolute paths (i.e.
that start with "/"), and should fail prior the revert mentioned above.

Should come up with more tests like this one.

Reviewers: stefan_schmidt, cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1937

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: Eina_Value - use eina_mempool_free to cleanup the memory
vivek [Thu, 5 Feb 2015 13:57:14 +0000 (14:57 +0100)]
eina: Eina_Value - use eina_mempool_free to cleanup the memory

Summary:
The new value is allocated using Eina_Mempool, it should be freed using
eina_mempool, not using libc free.

@fix

Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1940

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoEvas masking: Fix GL masking with maps and stuff
Jean-Philippe Andre [Thu, 5 Feb 2015 13:27:27 +0000 (22:27 +0900)]
Evas masking: Fix GL masking with maps and stuff

This fixes an issue spotted after the previous fix.
Passing the Y-invert flag is necessary because in the usual
case a map is rendered with Y-invert (OpenGL coords vs. Evas coords)
but in case a map is rendered in an FBO (another map's surface)
then Y-invert must be unset.

9 years agoEvas masking: Fix another issue with masking
Jean-Philippe Andre [Thu, 5 Feb 2015 07:11:45 +0000 (16:11 +0900)]
Evas masking: Fix another issue with masking

Fixes rendering in the following case:
- Object with a map has a mask
- Object is child of smart object which also has a map (eg. transit)
--> Masking did not apply to the children before this patch.

NOTE: This works fine in SW but still didn't work in GL, see the
      following commit...

9 years agoelua: correct eo inheritance behavior
Daniel Kolesa [Thu, 5 Feb 2015 11:10:29 +0000 (11:10 +0000)]
elua: correct eo inheritance behavior

Now we try to treat first inherit as a parent and everything else as a mixin.
If the first inherit is not a class (interface or mixin), every inherit is
treated as a mixin.

@fix

9 years agoeldbus: fix memory leak.
Guilherme Lepsch [Thu, 5 Feb 2015 10:28:36 +0000 (11:28 +0100)]
eldbus: fix memory leak.

Summary:

@fix

Reviewers: felipealmeida, larryolj, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1920

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoEcore_con: fix initialisation clean up and avoid memory leak.
Srivardhan Hebbar [Thu, 5 Feb 2015 10:26:58 +0000 (11:26 +0100)]
Ecore_con: fix initialisation clean up and avoid memory leak.

Summary: When ecore_init failed or when failed to initialize ecore_con_log_domain, then evil_shutdown was not called. So cleaned up code and added the fix.

@fix

Reviewers: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1921

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevil: prevent _evil_init_count from going below zero.
Srivardhan Hebbar [Thu, 5 Feb 2015 10:24:36 +0000 (11:24 +0100)]
evil: prevent _evil_init_count from going below zero.

Summary: _evil_init_count should never go below zero. This can occur, if a developer mistakenly calls evil_shutdown before calling evil_init. So fixing the code so that it never goes below zero.

Reviewers: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1922

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeldbus: production code inside Eina Safety Check assert.
Guilherme Lepsch [Thu, 5 Feb 2015 10:23:41 +0000 (11:23 +0100)]
eldbus: production code inside Eina Safety Check assert.

Summary: Eina Safety Checks assert macros don't do anything (no-op) if EINA_SAFETY_CHECKS is undefined.

@fix

Reviewers: felipealmeida, larryolj, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1926

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevas: Evas_3D - fix bug with uninit Evas_Mat4 flag.
Dmytro Dadyka [Thu, 5 Feb 2015 10:16:57 +0000 (11:16 +0100)]
evas: Evas_3D - fix bug with uninit Evas_Mat4 flag.

Reviewers: Hermet, raster, cedric

@fix

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1930

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoedje: Fix typo in a documentation while reading it.
Daniel Juyung Seo [Thu, 5 Feb 2015 01:50:03 +0000 (10:50 +0900)]
edje: Fix typo in a documentation while reading it.

9 years agoecore-x: Port deprecated API and new API function to xcb
Chris Michael [Wed, 4 Feb 2015 16:28:01 +0000 (11:28 -0500)]
ecore-x: Port deprecated API and new API function to xcb

Summary: This ports the deprecated screensaver_suspend and new API
screensaver_suspend to work with xcb.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-x: Deprecate old misspelled API function and add new one
Chris Michael [Wed, 4 Feb 2015 16:26:03 +0000 (11:26 -0500)]
ecore-x: Deprecate old misspelled API function and add new one

Summary: This deprecates the old misspelled screensaver_suspend API
function and adds a new one.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-x: Deprecate misspelled screensaver API function
Chris Michael [Wed, 4 Feb 2015 16:23:24 +0000 (11:23 -0500)]
ecore-x: Deprecate misspelled screensaver API function

Summary: This API function is misspelled, so after speaking with
Stefan, we've decided to deprecate this function add add a properly
spelled one.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoEcore_con: preventing _init_con_ssl_init_count to go below zero.
Srivardhan Hebbar [Wed, 4 Feb 2015 13:09:24 +0000 (14:09 +0100)]
Ecore_con: preventing _init_con_ssl_init_count to go below zero.

Summary: _init_con_ssl_init_count should not go below zero. This can occur if a developer mistakenly calls ssl shutdown before calling ssl init. So adding the check to prevent this.

Reviewers: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1925

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: fix valgrind invalid read of size in eina_file_path_sanitize.
Daniel Hirt [Wed, 4 Feb 2015 13:08:32 +0000 (14:08 +0100)]
eina: fix valgrind invalid read of size in eina_file_path_sanitize.

Summary:
Apparently eina_tmpstr_strlen counts the null character as well. This
doesn't follow how strlen works, as the latter excludes it from the count.
This resulted in mistreatment of the string in _eina_file_escape, with
tmp_str paths that had "../".

This fix will do for now, but it is advised that we avoid using
eina_tmpstr_strlen, to prevent such confusions in the future.

Test Plan:
The following lines will throw a valgrind 'invalid read of size 1' error
prior this fix:
  char *path = "home/mydir/../myfile";
  Eina_Tmpstr *tmp_str = eina_tmpstr_add(path);
  char *ret_path = eina_file_path_sanitize(path);

@fix

Reviewers: cedric, stefan_schmidt

Subscribers: tasn, cedric

Differential Revision: https://phab.enlightenment.org/D1929

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoedje: quick fix to guarantee that ed->path is always correctly set.
Cedric BAIL [Tue, 3 Feb 2015 17:55:33 +0000 (18:55 +0100)]
edje: quick fix to guarantee that ed->path is always correctly set.

NOTE: It will be better to completly remove ed->path in favor of using ed->file->f,
but that would be way to intrusive for this release cycle.

9 years agoecore-x - undo api/abi break with input selecting before release
Carsten Haitzler (Rasterman) [Wed, 4 Feb 2015 09:58:48 +0000 (18:58 +0900)]
ecore-x - undo api/abi break with input selecting before release

9 years agoupdate po's
Carsten Haitzler (Rasterman) [Wed, 4 Feb 2015 09:58:18 +0000 (18:58 +0900)]
update po's

9 years agoeet - fix endianess break added by cedric a year ago in image decoding
Carsten Haitzler (Rasterman) [Wed, 4 Feb 2015 09:03:20 +0000 (18:03 +0900)]
eet - fix endianess break added by cedric a year ago in image decoding

@fix

this fixes T1987 - the commit that broke it was:

commit 18d494489c5651aa152622a56579bbc1bda853cb
Author: Cedric BAIL <cedric.bail@samsung.com>
Date:   Fri Jan 3 16:20:53 2014 +0900

    eet: use eina_swap*() function instead of custom slower one.

9 years agoEvas masking: Fix rendering of restacked masks of masks in a map
Jean-Philippe Andre [Wed, 4 Feb 2015 05:43:14 +0000 (14:43 +0900)]
Evas masking: Fix rendering of restacked masks of masks in a map

I know. This title does not explain anything. Whatever.

This fixes the following issue:
- Mask a genlist (big mask)
- Each item has an icon masked (small mask)
- Apply a map to the genlist
- Scrolling the genlist
--> The big mask still works but totally screws up the
    small icons with masks.

Note: Once again this patch only affects code paths where an
      object is a mask.

9 years agoEvas masking: Fix some remaining issues with animations
Jean-Philippe Andre [Wed, 4 Feb 2015 03:08:50 +0000 (12:08 +0900)]
Evas masking: Fix some remaining issues with animations

Yeah, mixing maps and masks of masks in a genlist leads
to tons of amazing bugs :)

This commit removes x,y from the "mask" field in an object,
as they are duplicates of cur->geometry.{x,y} but were not
properly kept in sync.

This patch fixes a situation of:
- A genlist in a map
- Each item has an icon masked
- Scrolling the genlist
--> The masked items would not render properly before this
    patch.

Remaining known problem:
- Mask a genlist (big mask)
- Each item has an icon masked (small mask)
- Apply a map to the genlist
- Scrolling the genlist
--> The big mask still works but totally screws up the
    small icons with masks.

Note: These changes look scary just before the release
      but I would have to backport them to 1.13.x as they
      definitely are bug fixes. Also, they only concern
      code paths used exclusively by masking.

9 years agoEvas masking: Fix masking of multiple objects in a map
Jean-Philippe Andre [Tue, 3 Feb 2015 08:32:59 +0000 (17:32 +0900)]
Evas masking: Fix masking of multiple objects in a map

All those masking bug fixes become harder to explain. But here goes:

 - Take a genlist, apply a mask to it (for example put everything
   in an elm_layout). Also play with various objects in the genlist.
 - Also apply a map to it (for instance, elm_transit zoom).

 --> Now some elements will be masked, some others will not,
     and some may even not render at all.

This patch restores a mask in the current drawing context, instead
of just unsetting it.

9 years agoEvas masking: Forward previous mask where needed
Jean-Philippe Andre [Tue, 3 Feb 2015 06:51:01 +0000 (15:51 +0900)]
Evas masking: Forward previous mask where needed

In a situation where an object with mask of mask is in a map
(Yes! It can happen!) the masks would not get properly "multiplied".

Now the problem is that some objects still seem to bypass some
masks... Hmm...

9 years agoEvas masking: Prevent drawing masks in the wrong surface
Jean-Philippe Andre [Tue, 3 Feb 2015 06:38:50 +0000 (15:38 +0900)]
Evas masking: Prevent drawing masks in the wrong surface

Masks should be actually rendered only to their specific target
surface.

9 years agoecore-drm should probably send the expected (reversed) scroll z value
Mike Blumenkrantz [Wed, 4 Feb 2015 00:16:06 +0000 (19:16 -0500)]
ecore-drm should probably send the expected (reversed) scroll z value

9 years agoecore-wayland: Don't send key repeat events for keys which should not
Chris Michael [Tue, 3 Feb 2015 20:29:47 +0000 (15:29 -0500)]
ecore-wayland: Don't send key repeat events for keys which should not
get them

Summary: Some keys in certain keyboard layouts should not be repeated
so add a test which uses xkb to inform us if this key supports
repeating.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoUpdating serbian translation
maxerba [Tue, 3 Feb 2015 20:07:00 +0000 (21:07 +0100)]
Updating serbian translation

9 years agoecore_con: remove code in examples that triggered warnings.
Srivardhan Hebbar [Tue, 3 Feb 2015 13:30:21 +0000 (14:30 +0100)]
ecore_con: remove code in examples that triggered warnings.

Summary:
When the examples were compiled they used to through the below warning
$ gcc -o ecore_con_client_simple_example ecore_con_client_simple_example.c `pkg-config --libs --cflags ecore ecore-con eina`
ecore_con_client_simple_example.c:11:0: warning: "EINA_UNUSED" redefined
 # define EINA_UNUSED
 ^
In file included from /usr/include/eina-1/Eina.h:215:0,
                 from /usr/include/ecore-1/Ecore.h:304,
                 from ecore_con_client_simple_example.c:5:
/usr/include/eina-1/eina/eina_types.h:112:0: note: this is the location of the previous definition
 #  define EINA_UNUSED __attribute__ ((__unused__))
 ^
So removed these defines as they are no longer required now.

Reviewers: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1919

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agorelease: Update NEWS and bump version for 1.13.0-beta3 release v1.13.0-beta3
Stefan Schmidt [Tue, 3 Feb 2015 08:48:45 +0000 (09:48 +0100)]
release: Update NEWS and bump version for 1.13.0-beta3 release

9 years agoEvas masking: Check only the variable we want to set
Jean-Philippe Andre [Tue, 27 Jan 2015 06:02:27 +0000 (15:02 +0900)]
Evas masking: Check only the variable we want to set

This is a left-over from a previous fix a few weeks ago.
The point of this "if" is just to avoid writing the COW value
if not needed.

For reference:
  commit f876cf31f8d6fd48f7602bd593c24c1a07138d8f
  Author: Jean-Philippe Andre <jp.andre@samsung.com>
  Date:   Tue Dec 23 18:57:45 2014 +0900
    Evas masking: Fix invalid geometry after mask redraw

9 years agoevas callback: Fix typo in documentation of evas_object_event_callback_del_full
Amitesh Singh [Tue, 3 Feb 2015 02:02:20 +0000 (11:02 +0900)]
evas callback: Fix typo in documentation of evas_object_event_callback_del_full

Reviewers: cedric, seoz

Subscribers: cedric, seoz

Differential Revision: https://phab.enlightenment.org/D1904

9 years agoevas/examples: just change file name.
ChunEon Park [Tue, 3 Feb 2015 01:50:03 +0000 (10:50 +0900)]
evas/examples: just change file name.

9 years ago[Evas: Evas_3D] Improve shadows example.
Dmytro Dadyka [Tue, 3 Feb 2015 01:44:44 +0000 (10:44 +0900)]
[Evas: Evas_3D] Improve shadows example.

Reviewers: cedric, raster, Hermet

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1891

9 years agoecore-drm: Trap for failure to dup stdin
Chris Michael [Mon, 2 Feb 2015 16:59:27 +0000 (11:59 -0500)]
ecore-drm: Trap for failure to dup stdin

Summary: If we fail to duplicate stdin, we should write out an error
and exit appropriately.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-drm: Fix formatting
Chris Michael [Mon, 2 Feb 2015 16:46:50 +0000 (11:46 -0500)]
ecore-drm: Fix formatting

NB: No functional changes, just formatting

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoRevert "eina: eina_file - fix "invalid read of size 1" in valgrind"
Daniel Hirt [Mon, 2 Feb 2015 12:10:04 +0000 (14:10 +0200)]
Revert "eina: eina_file - fix "invalid read of size 1" in valgrind"

This reverts commit f52f5628916f44a6b8ce1fdd3b9cc40f6c997eda.

This is reverted because it breaks eina_file_path_sanitize when using
  "/../" in paths, for example:
eina_file_path_sanitize("/home/../mydir/myfile")
  returns: "/mydir/myfili"

What invalid read size does this fix? Why was no test case specified?
Anyway, this change affects too much code to leave it in like this.

9 years agoedje: edje_entry - remove dead code
WooHyun Jung [Mon, 2 Feb 2015 10:12:42 +0000 (19:12 +0900)]
edje: edje_entry - remove dead code

9 years agoEdje: edje_edit - abort Recursive Reference in edje_edit_part_source_set
Vitalii Vorobiov [Mon, 2 Feb 2015 09:14:45 +0000 (11:14 +0200)]
Edje: edje_edit - abort Recursive Reference in edje_edit_part_source_set

It is unable to do recursive reference such as:
>   Having group A with GROUP part that has group B as source.
>   Having group B with GROUP part that has group A as source.
Here we have a loop that is not allowed by edje_cc, so edje_edit also need
to check this case.

@fix

9 years agoecore-wl should maybe clear its repeat timer for keyboard on window free
Mike Blumenkrantz [Sun, 1 Feb 2015 23:32:04 +0000 (18:32 -0500)]
ecore-wl should maybe clear its repeat timer for keyboard on window free

9 years agoeina: eina_file - fix "invalid read of size 1" in valgrind
Romain Perier [Fri, 30 Jan 2015 09:49:58 +0000 (10:49 +0100)]
eina: eina_file - fix "invalid read of size 1" in valgrind

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: fix tests suites for Windows to have a predictable result.
michelle legrand [Fri, 30 Jan 2015 09:49:57 +0000 (10:49 +0100)]
eina: fix tests suites for Windows to have a predictable result.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_win32: don't allocate Ecore_Event_Key events with NULL keys
Romain Perier [Fri, 30 Jan 2015 09:49:56 +0000 (10:49 +0100)]
ecore_win32: don't allocate Ecore_Event_Key events with NULL keys

If "keyname" and "key" are NULL an Ecore_Event_Key is allocated
throught calloc. strlen() is called on these variables, which crashes
the program. As it is not correct to return a key event for NULL
keys to ecore, it is preferrable to return NULL and exit the function
before the key event is allocated, which fixes the crash in the same
time.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: use uintptr_t for Eina_Thread storage to be the same size for the different...
Nicolas Aguirre [Fri, 30 Jan 2015 09:49:55 +0000 (10:49 +0100)]
eina: use uintptr_t for Eina_Thread storage to be the same size for the different architectures.

This should not be an ABI regression on the platform where it work (Unix). It is on Windows,
but the API was unusable at all without this change, so this is a necessary breakage to
unbreak things.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agocheck: fix tests suites on Windows
michelle legrand [Fri, 30 Jan 2015 09:49:54 +0000 (10:49 +0100)]
check: fix tests suites on Windows

Disable timeout because we cannot fork on Windows.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeolian: fix to pass make check on windows
michelle legrand [Fri, 30 Jan 2015 09:49:53 +0000 (10:49 +0100)]
eolian: fix to pass make check on windows

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeo: fix to pass make check on windows
michelle legrand [Fri, 30 Jan 2015 09:49:52 +0000 (10:49 +0100)]
eo: fix to pass make check on windows

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoautotools: fix make check build error on Windows
michelle legrand [Fri, 30 Jan 2015 09:49:50 +0000 (10:49 +0100)]
autotools: fix make check build error on Windows

Windows does not use ECORE_X.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeio: Win32 make GetOverlappedResult function non-blocking
michelle legrand [Fri, 30 Jan 2015 09:49:49 +0000 (10:49 +0100)]
eio: Win32 make GetOverlappedResult function non-blocking

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevil: get right Windows environment variables for home path.
michelle legrand [Fri, 30 Jan 2015 09:49:48 +0000 (10:49 +0100)]
evil: get right Windows environment variables for home path.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevil: add windows compatible langinfo.
michelle legrand [Fri, 30 Jan 2015 09:49:45 +0000 (10:49 +0100)]
evil: add windows compatible langinfo.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoEdje: edje_cc - abort recursive Reference that is made by GROUP parts
Vitalii Vorobiov [Thu, 29 Jan 2015 20:21:05 +0000 (21:21 +0100)]
Edje: edje_cc - abort recursive Reference that is made by GROUP parts

Summary:
It is easy to create edj collections that aren't working at all and
edje_cc easily allows user to do that.
For example:
>   Having group A with GROUP part that has group B as source.
>   Having group B with GROUP part that has group A as source.
In this case edje_cc compile source code perfectly, but if user try to load
this edje as layout or use together with edje_edit it will cause unexpacted
and wrong behaviour.

@fix

Reviewers: seoz, Hermet, reutskiy.v.v, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1908

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevil: fix SEGFAULT in strcasestr
Andrii Kroitor [Thu, 29 Jan 2015 20:17:00 +0000 (21:17 +0100)]
evil: fix SEGFAULT in strcasestr

Summary:
there was an unsigned int underflow.

@fix

Test Plan: strcasestr("a", "bbb");

Reviewers: cedric, raster, Hermet, seoz

Subscribers: cedric, reutskiy.v.v

Differential Revision: https://phab.enlightenment.org/D1909

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeolian: fix unary expression parsing (correctly consume tokens)
Daniel Kolesa [Thu, 29 Jan 2015 11:12:04 +0000 (11:12 +0000)]
eolian: fix unary expression parsing (correctly consume tokens)

Also disable typed expression validation for now. The interfaces are not
finished yet so some things will fail validation because of unexistent
type definitions.

Fixes T2064.

@fix

9 years agoevil: fix initialisation of the socket library.
Vincent Torri [Wed, 28 Jan 2015 14:39:44 +0000 (15:39 +0100)]
evil: fix initialisation of the socket library.

@fix

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoSet the name for table, box items
Vyacheslav Reutskiy [Wed, 28 Jan 2015 16:47:54 +0000 (16:47 +0000)]
Set the name for table, box items

If box or table items haven't name in the edc we generate a new
name by mask "item_0xFFFFFFFFFFFFFFFF". Item name important for
Edje_Edit.

@fix

9 years agoembryo_cc_sc2.c: prevent memory corruption.
woochan lee [Mon, 26 Jan 2015 04:50:16 +0000 (13:50 +0900)]
embryo_cc_sc2.c: prevent memory corruption.

Summary:
Modified code to following coverity result.
(overflow the memory access.)

@fix

Reviewers: raster, Hermet

Reviewed By: Hermet

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1893

9 years agoedje: Edje_Edit - fix setting NULL into string pointer
Vitalii Vorobiov [Wed, 28 Jan 2015 14:40:09 +0000 (16:40 +0200)]
edje: Edje_Edit - fix setting NULL into string pointer

Some of edje_edit functions allows to set up NULL instead of string,
but by using _edje_if_string_free() function it should set variable into NULL,
so previous data won't appear anymore.

@fix

9 years agoevas: fix some clipping issues before/after Evas_Map animations
Jean-Philippe Andre [Wed, 28 Jan 2015 08:28:34 +0000 (09:28 +0100)]
evas: fix some clipping issues before/after Evas_Map animations

Summary:
This is an attempt at fixing:
 - T1767: The ultimate evil map & clip bug

Force recalculation and re-propagation of clipper geometry
after or just before a map is applied (only when transiting
between map enabled and map disabled).

I realized that doing clip_unset+clip_set in the E widget
code would fix the issue, but this is not a solution that
makes a lot of sense.

Unfortunately I have no idea about the side effects of this
patch, especially in terms of performance.

Fixes T1767 and maybe T1630.

Test Plan:
Open PackageKit popup in E, check the animations
and that clipping works fine both during, before and after
the animations.

Reviewers: raster, cedric

Reviewed By: cedric

Subscribers: cedric, Hermet

Maniphest Tasks: T1767

Differential Revision: https://phab.enlightenment.org/D1897

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoevas-image-loaders: Fix compiler warning of comparison between signed
Chris Michael [Tue, 27 Jan 2015 16:03:41 +0000 (11:03 -0500)]
evas-image-loaders: Fix compiler warning of comparison between signed
and unsigned values

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoevas-software: Fix compile errors when using xcb
Chris Michael [Tue, 27 Jan 2015 16:00:02 +0000 (11:00 -0500)]
evas-software: Fix compile errors when using xcb

Summary: This fixes compile issues when building evas software engine
using xcb. When calling xcb_outbuf_free, the Render_Engine has no 'ob'
field, so remove that call. Also fix call to
render_engine_software_generic_init using the proper function
parameters.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoevas-software: Fix software engine to compile using xcb
Chris Michael [Tue, 27 Jan 2015 15:57:07 +0000 (10:57 -0500)]
evas-software: Fix software engine to compile using xcb

Summary: Fix several compile issues with software engine when compiled
against xcb. This fixes an improper call to
evas_software_xcb_outbuf_flush with improper parameters, and adds a
missing EINA_UNUSED to that function.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoevas-software: Fix call to evas_software_xcb_outbuf_setup function
Chris Michael [Tue, 27 Jan 2015 15:52:53 +0000 (10:52 -0500)]
evas-software: Fix call to evas_software_xcb_outbuf_setup function

Summary: Seems we were missing a comma here and this was causing a
compile break.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-xcb: Add missing EINA_UNUSED for function paramater
Chris Michael [Tue, 27 Jan 2015 15:49:41 +0000 (10:49 -0500)]
ecore-xcb: Add missing EINA_UNUSED for function paramater

Summary: Depth is not used in the ecore_x_window_full_new function, so
add missing EINA_UNUSED

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-xcb: Fix ecore_x_window_full_new function to compile again
Chris Michael [Tue, 27 Jan 2015 15:48:10 +0000 (10:48 -0500)]
ecore-xcb: Fix ecore_x_window_full_new function to compile again

Summary: Let's use the proper variable names when assigning return
mask values.

NB: How did all this even get Into git with someone checking it and
compiling it ??

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>