platform/upstream/efl.git
8 years agoedje: add size_class test case
Jee-Yong Um [Wed, 16 Dec 2015 05:53:59 +0000 (14:53 +0900)]
edje: add size_class test case

Summary: add edje size_class APIs' test case

Reviewers: jpeg

Reviewed By: jpeg

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D3391

8 years agoemile: fix compilation with --enable-liblz4
Hein-Pieter van Braam [Wed, 16 Dec 2015 01:15:46 +0000 (17:15 -0800)]
emile: fix compilation with --enable-liblz4

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoevas: update/clarify RENDER_FLUSH callback docs
Mike Blumenkrantz [Tue, 15 Dec 2015 22:57:37 +0000 (17:57 -0500)]
evas: update/clarify RENDER_FLUSH callback docs

8 years agoevas: do not unset/delete master clip after every frame
Mike Blumenkrantz [Tue, 15 Dec 2015 22:21:42 +0000 (17:21 -0500)]
evas: do not unset/delete master clip after every frame

this is not ideal since it triggers a client-side rerender of every object
which was clipped to the master clip (double render) and then this ends up
forcing the server to rerender the same area twice as well

not only that, it causes all surface damages to to be the size of the entire
window - framespace for every frame

@fix

8 years agoevas: trigger FLUSH_PRE callback prior to kicking off an async render
Mike Blumenkrantz [Tue, 15 Dec 2015 22:18:36 +0000 (17:18 -0500)]
evas: trigger FLUSH_PRE callback prior to kicking off an async render

while not occurring immediately before flush as in sync rendering, this
is functionally close enough that it will serve the purpose for which the
callback was intended, namely receiving a callback that occurs after render
update calculations have occurred but before flush happens

@fix

ref cbb447c878ef779230ebf4470b615b8276c50fa2

8 years agoevas: resize master clip if framespace|viewport.changed is set
Mike Blumenkrantz [Tue, 15 Dec 2015 22:17:36 +0000 (17:17 -0500)]
evas: resize master clip if framespace|viewport.changed is set

8 years agoevas: do not add unclipped smart members to the master clip
Mike Blumenkrantz [Tue, 15 Dec 2015 22:16:26 +0000 (17:16 -0500)]
evas: do not add unclipped smart members to the master clip

this is probably a pretty unlikely scenario, but messing with smart
members in this way is probably not a great idea

@fix

8 years agoecore-evas wayland: move frame sending to pre-flush callback
Mike Blumenkrantz [Tue, 15 Dec 2015 22:14:43 +0000 (17:14 -0500)]
ecore-evas wayland: move frame sending to pre-flush callback

it was possible to deadlock rendering if a pre-render occurred but
the resulting render had no updated regions

@fix

8 years agoecore-drm: Fix failure of setting/closing evdev->fd which causes fd leak
Duna Oh [Tue, 15 Dec 2015 15:02:49 +0000 (10:02 -0500)]
ecore-drm: Fix failure of setting/closing evdev->fd which causes fd leak

Summary:
When a input device is plugged in, _cb_open_restricted() is called before creating evdev.
So setting fd value on evdev was failed in _cb_open_restricted() and also closing evdev->fd was invalid.
Using a eina_hash which has 'path-fd' pairs, we can find fd value after evdev is created.

@fix

Test Plan:
(1) Multiple input devices are connected. Their evdev->fd remains zero or initial value.
(2) When one of those devices are plugged out, fd leak would happen.

Reviewers: raster, zmike, gwanglim, stefan_schmidt, devilhorns, ManMower

Subscribers: cedric, jpeg, Jeon, input.hacker

Differential Revision: https://phab.enlightenment.org/D3428

8 years agoEina safety: Set error flag in case of safety check error
Jean-Philippe Andre [Tue, 15 Dec 2015 08:39:37 +0000 (17:39 +0900)]
Eina safety: Set error flag in case of safety check error

Eina_Error is almost not used at all. Meh.

8 years agoEina safety: Add internal function to log errors
Jean-Philippe Andre [Tue, 15 Dec 2015 07:56:13 +0000 (16:56 +0900)]
Eina safety: Add internal function to log errors

The main reason is convenience for debugging when using GDB,
this will give a simple breakpoint for all safety check failures.

Also, this creates a more visible log domain (red).

8 years agoEina log: Disable backtrace by default for dev builds
Jean-Philippe Andre [Tue, 15 Dec 2015 06:27:20 +0000 (15:27 +0900)]
Eina log: Disable backtrace by default for dev builds

The EINA_LOG_BACKTRACE thing is aimed at production environments,
so we can extract a backtrace from a log file post-mortem, but not
for continuous development of EFL itself.

I know this should make a few people happy.

8 years agoEina log: Fix meaning of EINA_LOG_BACKTRACE env
Jean-Philippe Andre [Tue, 15 Dec 2015 06:20:12 +0000 (15:20 +0900)]
Eina log: Fix meaning of EINA_LOG_BACKTRACE env

Most eina log env vars mean "if loglevel <= val then print log"
but eina_log_backtrace was "if loglevel < val" which I thought
was a bit confusing. The default behaviour is unchanged.

8 years agoExamples: Remove strange characters at beginning of file
Jean-Philippe Andre [Tue, 15 Dec 2015 05:55:20 +0000 (14:55 +0900)]
Examples: Remove strange characters at beginning of file

8 years agoExamples, Evas: remove MAX_PATH, which is already defined on Windows, and not used...
Vincent Torri [Tue, 15 Dec 2015 05:49:33 +0000 (14:49 +0900)]
Examples, Evas: remove MAX_PATH, which is already defined on Windows, and not used in this example

Reviewers: cedric, jpeg

Reviewed By: jpeg

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3442

8 years agoExample, Ecore: fix compilation on Windows
Vincent Torri [Tue, 15 Dec 2015 05:48:36 +0000 (14:48 +0900)]
Example, Ecore: fix compilation on Windows

Summary:
The socket API is different on Windows and on Linux.

This is the perfect example where we need to abstract the socket API in Eina :
1) to avoid all these includes specific to sockets
2) to avoid on Windows undefined behavior : close(socket); is undefined behavior if socket is indeed a socket

Reviewers: jpeg

Reviewed By: jpeg

Subscribers: cedric, jpeg

Projects: #efl

Differential Revision: https://phab.enlightenment.org/D3441

8 years agoExamples Evas: 'far' and 'near' are Windows keywords, so rename them to far_ and...
Vincent Torri [Tue, 15 Dec 2015 05:48:18 +0000 (14:48 +0900)]
Examples Evas: 'far' and 'near' are Windows keywords, so rename them to far_ and near_

Reviewers: cedric, jpeg

Reviewed By: jpeg

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3440

8 years agoExamples: Fix compilation on Windows for eina, eio, elocation and eolian_cxx
Vincent Torri [Tue, 15 Dec 2015 05:43:27 +0000 (14:43 +0900)]
Examples: Fix compilation on Windows for eina, eio, elocation and eolian_cxx

Summary:
When compiling on Windows, Evil.h must be included, so update Makefile.am
accordinglY

Reviewers: cedric, jpeg

Reviewed By: jpeg

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3439

8 years agoEvas VG: Fix clang warning
Jean-Philippe Andre [Tue, 15 Dec 2015 05:35:58 +0000 (14:35 +0900)]
Evas VG: Fix clang warning

8 years agoEina Xml: Fix clang warnings
Jean-Philippe Andre [Tue, 15 Dec 2015 05:35:46 +0000 (14:35 +0900)]
Eina Xml: Fix clang warnings

8 years agoEvas GL: Fix clang warnings
Jean-Philippe Andre [Tue, 15 Dec 2015 05:12:43 +0000 (14:12 +0900)]
Evas GL: Fix clang warnings

Use pragma diagnostic like with GCC since we want to set the function
pointer even though we deprecate the function.

8 years agoEina value: Fix clang warnings (use NULL instead of bool)
Jean-Philippe Andre [Tue, 15 Dec 2015 05:12:21 +0000 (14:12 +0900)]
Eina value: Fix clang warnings (use NULL instead of bool)

8 years agoEvas filters: Temporarily disable failing tests
Jean-Philippe Andre [Mon, 14 Dec 2015 11:28:13 +0000 (20:28 +0900)]
Evas filters: Temporarily disable failing tests

Apparently a special font makes the vflip tests crash. vflip
definitely needs to be fixed.

I'm currently working on the filters, so I'll get back to these.
This is a temporary patch.

8 years agoEolian gen: Don't fail if file is empty (--gi)
Jean-Philippe Andre [Mon, 14 Dec 2015 07:14:29 +0000 (16:14 +0900)]
Eolian gen: Don't fail if file is empty (--gi)

@fix

8 years agoEcore_Win32: let cursors on the non-client area (the borders) be like the default...
Vincent Torri [Sat, 12 Dec 2015 12:41:04 +0000 (13:41 +0100)]
Ecore_Win32: let cursors on the non-client area (the borders) be like the default ones

After the support of the X11 cursors on Windows, the cursors were set for the whole
window (even the borders). Now we let the system use the default cursors for the borders
and we use the cursors set by the API for the client area only

8 years agoEolian test: use eina_environment_tmp_get() instead of deprecated code
Vincent Torri [Sat, 12 Dec 2015 10:54:33 +0000 (11:54 +0100)]
Eolian test: use eina_environment_tmp_get() instead of deprecated code

8 years agoEina test: declare clock_gettime() by including time.h, like it should be
Vincent Torri [Sat, 12 Dec 2015 10:53:35 +0000 (11:53 +0100)]
Eina test: declare clock_gettime() by including time.h, like it should be

8 years agoEina test strbuf: the macro TEXT already exists on Windows so rename it to TEST_TEXT
Vincent Torri [Sat, 12 Dec 2015 10:37:54 +0000 (11:37 +0100)]
Eina test strbuf: the macro TEXT already exists on Windows so rename it to TEST_TEXT

8 years agoEvas example: fix warning
Vincent Torri [Sat, 12 Dec 2015 10:26:36 +0000 (11:26 +0100)]
Evas example: fix warning

Summary:

Test Plan:

Reviewers: cedric

Subscribers:

8 years agoEvas Textblock: Fix NULL dereferencing issue
Youngbok Shin [Sun, 13 Dec 2015 15:13:13 +0000 (17:13 +0200)]
Evas Textblock: Fix NULL dereferencing issue

Summary:
Even if the given two cursor is NULL, it shouldn't be crashed.
@fix

Test Plan:
Test case included in Evas test suite.
Run "make check".

Reviewers: herdsman, tasn

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D3422

8 years agoevas - fix evas module locks to init/del when movile created/destroyed
Carsten Haitzler (Rasterman) [Sat, 12 Dec 2015 14:37:16 +0000 (23:37 +0900)]
evas - fix evas module locks to init/del when movile created/destroyed

we initted when we load and unload. this led to races with
locking/unlocking elsewhere as these expected us to be initted and we
were not yet. this fixes that!

@fix

8 years agoecore-evas-wl: Fix copy/paste errors
Chris Michael [Fri, 11 Dec 2015 16:34:42 +0000 (11:34 -0500)]
ecore-evas-wl: Fix copy/paste errors

This fixes fat-finger copy/paste errors when copying functions from
ecore_evas_wayland_shm to ecore_evas_wayland_egl

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-evas-wl: Defer creating surfaces for wayland canvas
Chris Michael [Fri, 11 Dec 2015 16:26:24 +0000 (11:26 -0500)]
ecore-evas-wl: Defer creating surfaces for wayland canvas

This code adds support for deferring of surface creation and showing
inside Ecore_Evas Wayland. This is needed for Enlightenment so that it
does not try to create or show surfaces until the compositor has had a
chance to sync globals. This fixes an issue where early surface
creation would cause a crash in the compositor due to globals not
being syncd.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: Add new event for sync done
Chris Michael [Fri, 11 Dec 2015 16:24:32 +0000 (11:24 -0500)]
ecore-wl2: Add new event for sync done

This removes the usage of ecore_main_loop_iterate inside of the
display_connect function. It creates a new event type for when display
sync is done, this was we can defer surface creation and EE showing
until the compositor has had a chance to synchronize globals. We need
this for Enlightenment so that it does not try to create error dialogs
too early and thus crash due to not having sync'd globals yet

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoDraw: Fix compile error by removing duplicate definition.
Jaehyun Cho [Fri, 11 Dec 2015 07:58:55 +0000 (16:58 +0900)]
Draw: Fix compile error by removing duplicate definition.

8 years agoRevert "evas: trigger RENDER_FLUSH callbacks during async render"
Carsten Haitzler (Rasterman) [Fri, 11 Dec 2015 06:10:53 +0000 (15:10 +0900)]
Revert "evas: trigger RENDER_FLUSH callbacks during async render"

This reverts commit cbb447c878ef779230ebf4470b615b8276c50fa2.

1. this is wrong because evas_render_pipe_wakeup() is being called IN
THE RENDER THREAD. it... SENDS a wakeup back to the mainloop with

   evas_async_events_put(data, 0, NULL, evas_render_async_wakeup);

and you can see that evas_render_async_wakeup() calls
evas_render_wakeup() and in evas_render_wakeup() flush pre/post are
called, but since the trhead does the flush we cant realyl call
before/after, but it retains order... IF there are updates (haveup).

so calling these callbacks FROM a thread is now leading to apps
mysteriously exiting. this is mucho bad. just at random i now have my
terminals exiting.

8 years agoEvas GL: Fix compile error by comment out removed function.
Jaehyun Cho [Fri, 11 Dec 2015 05:10:43 +0000 (14:10 +0900)]
Evas GL: Fix compile error by comment out removed function.

8 years agoRevert "ecore_cocoa: add support for system cursors"
Jaehyun Cho [Fri, 11 Dec 2015 04:44:40 +0000 (13:44 +0900)]
Revert "ecore_cocoa: add support for system cursors"

This reverts commit 4623d57762094011c5155cbea9512ae6a2db8852.
This commit is duplicated with 2c93c73cbd1966e8cead7e7856044a2eb12138d7.

8 years agoevas: Do not preload data if data is cached.
Minkyoung Kim [Fri, 11 Dec 2015 04:50:40 +0000 (13:50 +0900)]
evas: Do not preload data if data is cached.

Summary: Summary : If data is cached, need not to reload data.

Test Plan: Local tests

Reviewers: jpeg

Reviewed By: jpeg

Subscribers: eunue, spacegrapher, cedric, wonsik, jiin.moon

Differential Revision: https://phab.enlightenment.org/D3418

8 years agoEvas GL: Avoid confusing ERR message when requesting depth & stencil
Jean-Philippe Andre [Fri, 11 Dec 2015 02:32:06 +0000 (11:32 +0900)]
Evas GL: Avoid confusing ERR message when requesting depth & stencil

An ERR message would be printed out mentionning that direct rendering
won't work on this window, even if DR was not requested.
Also, set the DR-specific flags only if DR is enabled.

See T2936.

8 years agoefl -break the "i really know what i'm doing" option to get attention
Carsten Haitzler (Rasterman) [Fri, 11 Dec 2015 00:51:00 +0000 (09:51 +0900)]
efl -break the "i really know what i'm doing" option to get attention

so .. more gentoo "i just copy and pasted and dont know what i'm
doing" land has hit again.

8 years agoeldbus: fix compilation of eldbus user.
Cedric BAIL [Fri, 11 Dec 2015 00:22:55 +0000 (16:22 -0800)]
eldbus: fix compilation of eldbus user.

8 years agoevas: trigger RENDER_FLUSH callbacks during async render
Mike Blumenkrantz [Thu, 10 Dec 2015 20:58:22 +0000 (15:58 -0500)]
evas: trigger RENDER_FLUSH callbacks during async render

these callbacks are used, so ensure that they get called in the expected
order

@fix

 #PatchesFromCedric

8 years agoee-wayland: do not send configure ack if no configure serial exists
Mike Blumenkrantz [Thu, 10 Dec 2015 19:37:33 +0000 (14:37 -0500)]
ee-wayland: do not send configure ack if no configure serial exists

also unset serial to ensure subsequent renders don't trigger the same ack

8 years agoeldbus: add eldbus.model examples
Larry [Wed, 2 Dec 2015 23:20:48 +0000 (21:20 -0200)]
eldbus: add eldbus.model examples

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoeldbus: add efl.model implementation.
Guilherme Lepsch [Fri, 9 Jan 2015 16:20:14 +0000 (14:20 -0200)]
eldbus: add efl.model implementation.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoefl: common Efl.Model implementation functions helpers.
Guilherme Lepsch [Thu, 30 Jul 2015 18:48:41 +0000 (15:48 -0300)]
efl: common Efl.Model implementation functions helpers.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoeina: add EINA_VALUE_TYPE_OPTIONAL, a single-element container that can be empty.
Felipe Magno de Almeida [Thu, 9 Jul 2015 03:57:38 +0000 (00:57 -0300)]
eina: add EINA_VALUE_TYPE_OPTIONAL, a single-element container that can be empty.

Eina Value Optional can be used to create a eina value that can be set
or be empty and can be embedded in a eina_value_struct.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoecore-wl2: ensure client display hash is freed and set NULL on connect failure
Mike Blumenkrantz [Thu, 10 Dec 2015 16:18:24 +0000 (11:18 -0500)]
ecore-wl2: ensure client display hash is freed and set NULL on connect failure

8 years agoedje_cc: Use ecore_file_file_get() and ecore_file_dir_get() for EDC parsing
Vincent Torri [Thu, 10 Dec 2015 11:09:00 +0000 (20:09 +0900)]
edje_cc: Use ecore_file_file_get() and ecore_file_dir_get() for EDC parsing

Replace current EDC parsing code in edje_cc on Windows with the code
using ecore_file_file_get() and ecore_file_dir_get().

Signed-off-by: Jaehyun Cho <jae_hyun.cho@samsung.com>
8 years agoeina_inline_lock_posix: Be far more careful with pthread function error returns
Conrad Meyer [Thu, 10 Dec 2015 10:02:28 +0000 (19:02 +0900)]
eina_inline_lock_posix: Be far more careful with pthread function error returns

Summary:
Related to T2287.

Log lock errors (printf to avoid eina_log locks) and continue or abort,
conditional on EINA_HAVE_DEBUG_THREADS.

Reviewers: raster, cedric

Subscribers: stefan_schmidt, cedric, seoz

Differential Revision: https://phab.enlightenment.org/D2376

Note - fixed review comments on macro names and some formatting and
error strings too - raster.

8 years agoRevert "Evas: Refactor model's savers and loaders."
Jean-Philippe Andre [Thu, 10 Dec 2015 09:20:43 +0000 (18:20 +0900)]
Revert "Evas: Refactor model's savers and loaders."

This reverts commit 01a32f64c0ffbe2ce048f8eb016d685913558cf5.

This broke make check with the following error:
evas_test_mesh.c:123:F:Meshes:evas_object_mesh_loader_saver:0: Failure 'res == 1' occurred

Reopen https://phab.enlightenment.org/D3420

8 years agoEvas textblock: add missing test fonts
Daniel Hirt [Thu, 10 Dec 2015 07:57:39 +0000 (09:57 +0200)]
Evas textblock: add missing test fonts

evas_suite was failing for people that didn't have malayalam fonts
installed in their system. Obviously, we should provide it in the test
suite.

Added malayalam font (from http://www.indlinux.org/) to the
TestFont.eet, along with the license README.

Fixes T2908.

@fix

8 years agoecore_idle_exiter: unlock when Ecore_Task_Cb is NULL
Sung-Taek Hong [Thu, 10 Dec 2015 08:01:25 +0000 (17:01 +0900)]
ecore_idle_exiter: unlock when Ecore_Task_Cb is NULL

Summary:
- When Ecore_Task_Cb is not set, _ecore_idle_exiter_constructor
  returns without _ecore_unlock(), and remains to be locked.

Reviewers: jpeg

Reviewed By: jpeg

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D3424

8 years agoeina: Adding test case for base64 decode function.
Srivardhan Hebbar [Thu, 10 Dec 2015 07:44:27 +0000 (16:44 +0900)]
eina: Adding test case for base64 decode function.

Summary:
Depends on D3381

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric, jpeg

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3382

8 years agoevas/gl_common: Reset current texture target with proper value when creating engine...
Minkyoung Kim [Thu, 10 Dec 2015 07:40:15 +0000 (16:40 +0900)]
evas/gl_common: Reset current texture target with proper value when creating engine gl context.

Summary:
set current target with default value of GL_TEXTURE_2D.
target should be valid value.

Reviewers: jpeg

Reviewed By: jpeg

Subscribers: wonsik, cedric, spacegrapher

Differential Revision: https://phab.enlightenment.org/D3423

8 years agoEvas: Refactor model's savers and loaders.
perepelits.m [Thu, 10 Dec 2015 07:37:30 +0000 (16:37 +0900)]
Evas: Refactor model's savers and loaders.

Summary:
Move common part to a separated document.
Make code more readable using smaller functions. (from Task T2713)

I did it again because somehow test is passing now. It seems like this test suite is unstable.
Please, let me know if there are any errors after running distcheck.

Reviewers: cedric, raster, Hermet, stefan_schmidt

Subscribers: jpeg, artem.popov

Differential Revision: https://phab.enlightenment.org/D3420

8 years agoevas_eet: Add Type Safety check
Pankaj Mittal [Thu, 10 Dec 2015 07:26:56 +0000 (16:26 +0900)]
evas_eet: Add Type Safety check

Summary:
The Function _evas_canvas3d_eet_file_free(void) is referenced in  evas_model_load_file_eet()(file:evas_model_load_eet.c at line 122).
This call is  under condition
        if ((eet_file->mesh == NULL) || (eet_file->header == NULL)).
when  either eet_file->mesh or eet_file->header are NULL, dereference of the corresponding pointer in function "_evas_canvas3d_eet_file_free()"
will generate Segmentation Fault.

@fix

Reviewers: raster, Hermet, tasn, wonsik, spacegrapher, cedric, jpeg

Subscribers: singh.amitesh, sachin.dev, alok25, yashu21985, mvsovani, cedric

Differential Revision: https://phab.enlightenment.org/D3369

8 years agoevas/gl: Update texture when preload is cancelled.
Minkyoung Kim [Thu, 10 Dec 2015 07:04:57 +0000 (16:04 +0900)]
evas/gl: Update texture when preload is cancelled.

Summary:
When preload is cancelled before finishing loading,
should reload the image data and update the texture during rendering object.
So, force texture to be updated on first drawing time.
(It should be guaranteed that preload image object is hidden before preload done.)

Test Plan: Local tests

Reviewers: jpeg

Reviewed By: jpeg

Subscribers: eunue, jiin.moon, wonsik, cedric, spacegrapher

Differential Revision: https://phab.enlightenment.org/D3416

8 years agoecore-evas-wayland: Fix issue of starting resizing causing size jumps
Chris Michael [Wed, 9 Dec 2015 17:03:53 +0000 (12:03 -0500)]
ecore-evas-wayland: Fix issue of starting resizing causing size jumps

Previously, when we started to resize an efl app, the size would
"jump" due to framespace being adjusted. This patch fixes that issue
and resize now works as expected.

@fix

Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
8 years agoecore-evas-wayland: Move configure acknowledge to render_flush_pre
Chris Michael [Wed, 9 Dec 2015 14:38:01 +0000 (09:38 -0500)]
ecore-evas-wayland: Move configure acknowledge to render_flush_pre

If we acknowledge a configure from xdg during post render, we end up
breaking maximize of EFL clients inside Weston (and perhaps other
compositors). In order to fix that, we will now send the configure ack
post render but pre flush.

@fix

Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
8 years agoRevert "edje: calc - remove pointer comparison while finding part desc"
Amitesh Singh [Wed, 9 Dec 2015 15:59:42 +0000 (21:29 +0530)]
Revert "edje: calc - remove pointer comparison while finding part desc"

This reverts commit c892a1cb714fed496cbf5568c4d43880b6fb67b2.

8 years agoedje: calc - add curly braces to avoid ambiguous 'if'
Amitesh Singh [Wed, 9 Dec 2015 10:25:59 +0000 (15:55 +0530)]
edje: calc - add curly braces to avoid ambiguous 'if'

merge two if conditions into one also.

8 years agoedje: calc - remove pointer comparison while finding part desc
Amitesh Singh [Wed, 9 Dec 2015 10:16:41 +0000 (15:46 +0530)]
edje: calc - remove pointer comparison while finding part desc

Only strcmp comparision is realiable.
@fix

8 years agoEvas text: Fix Evas Text truncated text case.
Youngbok Shin [Wed, 9 Dec 2015 07:50:33 +0000 (09:50 +0200)]
Evas text: Fix Evas Text truncated text case.

Summary:
Evas Text only concerns about a advance of each text item.
When a width of last character is bigger than its advance, the last character can be truncated.
And the different line size calculation caused different aligning between Evas Text and Evas Textblock.
So, the width of last character will be considered in Evas Text just like Evas Textblock.
@fix

Test Plan:
The following text shows how the size calculation is different between Evas Textblock and Text.
Get native size from Evas Textblock and get width(geometry) of Evas Text.
You can see the width of Evas Text is bigger than native size of Evas Textblock.
(adv > width)
こんにちは。

The following text will be truncated without this patch.
(adv < width)
ନୂଁ

Reviewers: woohyun, tasn, herdsman

Subscribers: jpeg, cedric

Differential Revision: https://phab.enlightenment.org/D3004

8 years agoEvas: Fix build for ubuntu... again
Jean-Philippe Andre [Wed, 9 Dec 2015 07:58:48 +0000 (16:58 +0900)]
Evas: Fix build for ubuntu... again

See 75fed54f0e4d72b20860
Apparently the proper CFLAGS are not set for SSE compilation.
SSE flags are set only for a specific file (now 2 of them).

As usual, it was worksforme but not for others. This should probably
fix that. Thanks Sub for the report and testing this solution.

8 years agoefl - ecore-x randr - fix coverity issue found after previous leak fix
Carsten Haitzler (Rasterman) [Wed, 9 Dec 2015 05:21:10 +0000 (14:21 +0900)]
efl - ecore-x randr - fix coverity issue found after previous leak fix

fix coverity issue pointed out in the previous commit fix:

dadc1451e7f53d64b9b848d18c4c2514d791c53e

use after free of ptr (free was in wrong spot)

@fix

8 years agoEvas filters: Move 'fill' to its own file
Jean-Philippe Andre [Wed, 9 Dec 2015 02:42:31 +0000 (11:42 +0900)]
Evas filters: Move 'fill' to its own file

8 years agoEvas: Add all 'draw' files to evas static lib
Jean-Philippe Andre [Wed, 9 Dec 2015 01:20:01 +0000 (10:20 +0900)]
Evas: Add all 'draw' files to evas static lib

This should fix the build on old Ubuntu. I didn't get this error myself,
but apparently the following issue happened:

  CCLD   lib/ecore_evas/libecore_evas.la
lib/evas/.libs/libevas.so: undefined reference to `efl_draw_neon_init'
lib/evas/.libs/libevas.so: undefined reference to `efl_draw_sse2_init'
collect2: ld returned 1 exit status

Thanks @mer.kim for the report

8 years agoeina_quaternion: fix ein matrix type comparison
Stefan Schmidt [Wed, 9 Dec 2015 00:22:38 +0000 (01:22 +0100)]
eina_quaternion: fix ein matrix type comparison

Fixing a bug that I fixed before. Sadly got introduced again. See
a486671bce76e405a8d7261f19b0f67ca678f0c4 for the details.

CID: 1341769

8 years agoRevert "evas: refactor model's savers and loaders."
Stefan Schmidt [Tue, 8 Dec 2015 21:23:17 +0000 (22:23 +0100)]
Revert "evas: refactor model's savers and loaders."

This reverts commit 32c33ed64dda542c7cfc952fc656bb711260441b.

This refactor broke the evas test cases for the model loaders and savers. I gave
it a week to get fixed but a first try did not succeed and its blocks a lot of
other automated testing. To be honest, it should have never gone it when it
breaks existing test cases. Once fixed this refactor can happily go in.

Fixes T2905

8 years agoRevert "build: fix distcheck after model saver and loader rework"
Stefan Schmidt [Wed, 2 Dec 2015 11:15:09 +0000 (12:15 +0100)]
Revert "build: fix distcheck after model saver and loader rework"

This reverts commit a7a2781a00ca6557aef6490bc0159cb426441b28.

Fix for a commit that needs reverting so we need to revert this patch as well.
See next commit or bug number for details.

Ref T2905

8 years agoecore-wl2: rewrite iconified set function for consistency
Chris Michael [Tue, 8 Dec 2015 14:56:57 +0000 (09:56 -0500)]
ecore-wl2: rewrite iconified set function for consistency

* santize boolean params
* enforce window state flag setting

@fix

ref T2919

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoEvas textblock: improve and fix line range rectangles
Daniel Hirt [Tue, 8 Dec 2015 15:17:59 +0000 (17:17 +0200)]
Evas textblock: improve and fix line range rectangles

The line range rectangles geometries needed a bit of adjusting. I
started out with fixing T2648. In order to fill the gap between the end
of the line and the margins, the geometry of the last line's character
was used. I am not really sure why. Anyway, we have the line geometry,
so I replaced it with that.

Then, it led me to do some alignment checks, and indeed alignment cases
were not treated. For instance, an LTR paragraph could have a line
aligned with a value greater than 0.0. That means that we should fill
the gap from the left of the line, if it was the last line in a
multi-line selection. The inverse case is for RTL.

I think it now works as it should. Will see if the selection logic is
missing some more stuff once I come up with more example cases.

Fixes T2648.

@fix

8 years agoeina: Adding API for base64 url encoding.
Srivardhan Hebbar [Tue, 8 Dec 2015 11:43:42 +0000 (20:43 +0900)]
eina: Adding API for base64 url encoding.

Summary:
I have put the common encoding code in the common function. The 2 API's
call them with a flag. Thinking of doing the same way to decode
function also.
T2880

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D3414

8 years agoEdje cc: Fix use of uninitialised value.
Tom Hacohen [Tue, 8 Dec 2015 11:41:28 +0000 (11:41 +0000)]
Edje cc: Fix use of uninitialised value.

Slash should be initialised to NULL, as if slash was not found in case
we don't even look for it.

Also, further restrict slash's scope.

@fix

8 years agoEvas render2: Fix always false comparisons.
Tom Hacohen [Tue, 8 Dec 2015 11:35:25 +0000 (11:35 +0000)]
Evas render2: Fix always false comparisons.

This looks like a classic copy and paste error. It didn't make any sense
before, and it seems like no one was going to take a look at it.
This looks correct. If this breaks something (is render2 even used),
someone should remove those lines. I wasn't entirely sure if I should
just remove them, or correct them, as some of the code looked redundant
anyway.

8 years agoEvas test matrix: Silence warning.
Tom Hacohen [Tue, 8 Dec 2015 11:15:23 +0000 (11:15 +0000)]
Evas test matrix: Silence warning.

This happens because this test doesn't really depend on anything evas,
so it doesn't set evas up. We want to be warned when tests forget to set
evas up, just not in this case.

8 years agoEcore con socks: Add missing break statement to switch.
Tom Hacohen [Tue, 8 Dec 2015 11:05:17 +0000 (11:05 +0000)]
Ecore con socks: Add missing break statement to switch.

This looks like an obvious case of missing break. If it wasn't a missing
break, there should have been at least a comment. Looking at the code it
looks like a break is needed. Also, I suspect this code path is never
really tested, and that's why we never hit it.

Tests are not failing either way.

CID1039379

8 years agoEolian legacy generator: Make code a bit clearer.
Tom Hacohen [Tue, 8 Dec 2015 10:57:19 +0000 (10:57 +0000)]
Eolian legacy generator: Make code a bit clearer.

This makes it clear that only one of those conditions can be
true.

This also fixed CID1323087

8 years agoEvas filters: Remove unused variable
Jean-Philippe Andre [Fri, 4 Dec 2015 07:30:54 +0000 (16:30 +0900)]
Evas filters: Remove unused variable

8 years agoEvas: Remove unwanted file added by accident
Jean-Philippe Andre [Tue, 8 Dec 2015 05:54:21 +0000 (14:54 +0900)]
Evas: Remove unwanted file added by accident

From commit 82504d9d77f761f2575f63095459011eba02a594

8 years agoEfreet, Elua : use eina_file_mkstemp instead of mkstemp
Vincent Torri [Tue, 8 Dec 2015 05:24:59 +0000 (14:24 +0900)]
Efreet, Elua : use eina_file_mkstemp instead of mkstemp

Reviewers: jpeg

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D3411

8 years agoEvil: remove useless internal function
Vincent Torri [Tue, 8 Dec 2015 04:53:32 +0000 (13:53 +0900)]
Evil: remove useless internal function

Reviewers: cedric, jpeg

Reviewed By: jpeg

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3410

8 years agoEvil: fix infinite recursive loop in previous addition of setlocale() in Evil
Vincent Torri [Tue, 8 Dec 2015 04:53:23 +0000 (13:53 +0900)]
Evil: fix infinite recursive loop in previous addition of setlocale() in Evil

Summary:
setlocale() called itself because it was defined as a #define
so remove this #define from evil_locale.h and move it in another header file
To avoid future problem, move similar defines to this header file
Also clean all the header file mess in Evil

Reviewers: cedric

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3409

8 years agoedje: introduce size_class attribute
Jee-Yong Um [Tue, 8 Dec 2015 03:15:48 +0000 (12:15 +0900)]
edje: introduce size_class attribute

Edje_Part can change its min or max size in code level with
size_class.

Differential Revision: https://phab.enlightenment.org/D3329

PS: Manual commit, arc refused to work...

@feature

Signed-off-by: Jean-Philippe Andre <jp.andre@samsung.com>
8 years agoedje_cc: Fix parsing including file path in EDC on Windows.
Jaehyun Cho [Tue, 8 Dec 2015 02:01:11 +0000 (11:01 +0900)]
edje_cc: Fix parsing including file path in EDC on Windows.

On Windows, including file path in EDC has not been parsed correctly
because '\' has not been used for path separator.
This commit fixes edje_cc complie to use '\' as path separator on
Windows.

8 years agoevas: fix mistake in descriptions of matrix
se.osadchy [Tue, 8 Dec 2015 00:51:55 +0000 (16:51 -0800)]
evas: fix mistake in descriptions of matrix

Summary: Wrong word in documentation eina_matrix.h.

Reviewers: Hermet, raster, jpeg, cedric

Reviewed By: jpeg, cedric

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D3406

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoevas: recognize duplicated data structure with eina_quternion and ector.
se.osadchy [Tue, 8 Dec 2015 00:12:56 +0000 (16:12 -0800)]
evas: recognize duplicated data structure with eina_quternion and ector.

Summary:
Move data structure and functionality to eina_quaternion from evas_vec4.

Reviewers: raster, Hermet, cedric

Subscribers: jpeg, Oleksander

Differential Revision: https://phab.enlightenment.org/D3327

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoecore-evas-wayland: Don't set window geometry in ecore_evas
Chris Michael [Mon, 7 Dec 2015 20:32:58 +0000 (15:32 -0500)]
ecore-evas-wayland: Don't set window geometry in ecore_evas

As we do not have the proper values for window geometry to be setting
it here, remove calls to set window geometry. We can more accurately
determine the window geometry from inside Elementary as it handles the
theme for the window borders.

@fix

ref T2919

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: rewrite maximize/fullscreen set functions for consistency
Chris Michael [Mon, 7 Dec 2015 19:39:29 +0000 (14:39 -0500)]
ecore-wl2: rewrite maximize/fullscreen set functions for consistency

* sanitize boolean params
* enforce window state flag setting
* use window flag instead of window type for state

@fix

ref T2919

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoevas: Fix definition of framespace in canvas eo file
Chris Michael [Mon, 7 Dec 2015 17:54:21 +0000 (12:54 -0500)]
evas: Fix definition of framespace in canvas eo file

The definition of 'framespace' in the canvas eo file is incorrect.
Framespace is the space occupied by the window frame within the canvas viewport

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: Fix issue of cleanup function always removing from client
Chris Michael [Mon, 7 Dec 2015 17:18:28 +0000 (12:18 -0500)]
ecore-wl2: Fix issue of cleanup function always removing from client
display hash

As the cleanup function is a generic helper to cleanup things in the
Ecore_Wl2_Display structure, we cannot always just explicitly the
display from the client_displays hash.

The removal from the appropriate hashes should be done by the calling
function(s) (ecore_wl2_display_disconnect for clients, and
ecore_wl2_display_destroy for servers)

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: Fix formatting and remove extra blank line
Chris Michael [Mon, 7 Dec 2015 16:30:53 +0000 (11:30 -0500)]
ecore-wl2: Fix formatting and remove extra blank line

NB: no functional changes

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoEvas textblock: fix evas_textblock_cursor_line_set
Daniel Hirt [Mon, 7 Dec 2015 15:23:24 +0000 (17:23 +0200)]
Evas textblock: fix evas_textblock_cursor_line_set

The line_set function should set the cursor to the first logical
position in the line. We can't use the first text position of the
first item in the line, due to BiDi considerations (the line may be
reordered). I've split evas_textblock_cursor_line_char_first to avoid
code duplication, as it already handles these cases.

@fix

8 years agoEvas GL: Fix build for bigendian
Jean-Philippe ANDRE [Mon, 7 Dec 2015 16:03:40 +0000 (01:03 +0900)]
Evas GL: Fix build for bigendian

Fixes T2918

8 years agoevas-wayland-egl: Fix the unnecessary renewal of the engine window.
Seunghun Lee [Mon, 7 Dec 2015 15:06:21 +0000 (10:06 -0500)]
evas-wayland-egl: Fix the unnecessary renewal of the engine window.

Summary:
As we do not set the ob->info->info.win variable anymore inside
Ecore_Evas (old code that was removed), we can omment out the line.
This line was causing unnecessary renewal of the engine window (thus
causing flickering when rotation was applied).

@fix

Test Plan: Rotate or Resize the window on wayland-egl.

Reviewers: gwanglim, devilhorns

Reviewed By: devilhorns

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D3407

8 years agoecore-wl2: Don't free display structure during display cleanup function
Chris Michael [Mon, 7 Dec 2015 15:00:23 +0000 (10:00 -0500)]
ecore-wl2: Don't free display structure during display cleanup function

Due to reference caching, we cannot free the display structure here
yet. During calls to ecore_wl2_display_destroy (or others), if the
number of references reaches zero, then we end up calling
wl_display_destroy (or equivelant). If we free our display structure
during cleanup function, then calls to wl_display_destroy will cause a
segfault as our display->wayland_display has already been freed from
the structure

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoefl - ecore-x xrandr - fix memleaks of properties fetched
Carsten Haitzler (Rasterman) [Mon, 7 Dec 2015 10:34:53 +0000 (19:34 +0900)]
efl - ecore-x xrandr - fix memleaks of properties fetched

while doing some debugging, ifound we have been leaking these xrandr
properites that were malloced in certain cases. this fixes these leaks.

@fix

8 years agoeina: Updated test case to test more scenario.
Srivardhan Hebbar [Mon, 7 Dec 2015 10:22:47 +0000 (19:22 +0900)]
eina: Updated test case to test more scenario.

Summary:
The test cases which I took from wikipedia didn't have the characters
'/' and '+' in the encoded string. So added test case testing these.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D3405