platform/upstream/efl.git
8 years agoecore_evas_wayland: Make alpha set common
Derek Foreman [Tue, 1 Nov 2016 21:34:26 +0000 (16:34 -0500)]
ecore_evas_wayland: Make alpha set common

Moves alpha set into the common implementation - there is a functional
change here - the egl engine now calls alpha_do in render_updates like
the shm engine.

It is probable that this was the intended behaviour all along.

8 years agoecore_evas_wayland: Move more functions into the common implementation
Derek Foreman [Tue, 1 Nov 2016 21:20:49 +0000 (16:20 -0500)]
ecore_evas_wayland: Move more functions into the common implementation

Show and hide can be made common with almost no functional changes.

8 years agowayland evas engines: share engine info structure
Derek Foreman [Tue, 1 Nov 2016 20:46:35 +0000 (15:46 -0500)]
wayland evas engines: share engine info structure

These engines are incredibly similar - by sharing the same engine info
structure we'll be able to simplify the wayland ecore_evas bits and
make them much more maintainable.

8 years agoecore_evas_wayland: Don't include Evas_Engine_Wayland_Egl.h
Derek Foreman [Tue, 1 Nov 2016 20:47:08 +0000 (15:47 -0500)]
ecore_evas_wayland: Don't include Evas_Engine_Wayland_Egl.h

There doesn't seem to be any need for this.

8 years agowayland_egl: Remove unused wl_egl_window variables
Derek Foreman [Tue, 1 Nov 2016 17:38:44 +0000 (12:38 -0500)]
wayland_egl: Remove unused wl_egl_window variables

8 years agowayland evas engines: rename wayland display to wl_display
Derek Foreman [Tue, 1 Nov 2016 17:07:10 +0000 (12:07 -0500)]
wayland evas engines: rename wayland display to wl_display

Continuing to make the shm and egl engines closer to eachother.

8 years agowayland_egl: Rename surface to wl_surface
Derek Foreman [Tue, 1 Nov 2016 16:41:47 +0000 (11:41 -0500)]
wayland_egl: Rename surface to wl_surface

Making this code more closely match the wayland_shm engine

8 years agowayland_egl: Remove pre_post_swap_callback_set
Derek Foreman [Tue, 1 Nov 2016 16:26:14 +0000 (11:26 -0500)]
wayland_egl: Remove pre_post_swap_callback_set

This stuff seems copied from gl_x11 which actually has API to use it.

In this engine nothing can actually set it, so it's just all dead code.

8 years agowayland_egl: Remove Evas_Engine_Info_Wayland_Egl_Swap_Mode enum
Derek Foreman [Tue, 1 Nov 2016 16:24:12 +0000 (11:24 -0500)]
wayland_egl: Remove Evas_Engine_Info_Wayland_Egl_Swap_Mode enum

Nothing uses this.

8 years agowayland_egl: Remove screen from engine info
Derek Foreman [Tue, 1 Nov 2016 13:36:55 +0000 (08:36 -0500)]
wayland_egl: Remove screen from engine info

Doesn't appear to actually be used for anything.

8 years agowayland ecore evas: move move_resize to common
Derek Foreman [Tue, 1 Nov 2016 12:42:23 +0000 (07:42 -0500)]
wayland ecore evas: move move_resize to common

It's identical in both shm and egl variants, share it.

8 years agowayland_egl: Remove some leftovers from www
Derek Foreman [Tue, 1 Nov 2016 12:41:09 +0000 (07:41 -0500)]
wayland_egl: Remove some leftovers from www

The render post callback and "wobbling" variable were only required for
the client side effect.

8 years agoecore-wl2: Update cursor regardless of cursor surface
Chris Michael [Wed, 2 Nov 2016 16:12:04 +0000 (12:12 -0400)]
ecore-wl2: Update cursor regardless of cursor surface

If a NULL surface gets passed into ecore_wl2_window_pointer_set that
would mean we are unsetting the cursor surface, so we should still be
calling wl_pointer_set_cursor even with a NULL surface.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agotheme: remove text.min from bryce sizing text
Mike Blumenkrantz [Wed, 2 Nov 2016 15:29:49 +0000 (11:29 -0400)]
theme: remove text.min from bryce sizing text

this causes bogus min size calc values

ref T4809

8 years agotheme: trigger recalc from time's clock themes when object is resized
Mike Blumenkrantz [Wed, 2 Nov 2016 15:19:54 +0000 (11:19 -0400)]
theme: trigger recalc from time's clock themes when object is resized

8 years agoelput: Add API to allow setting pointer acceleration profile
Chris Michael [Wed, 2 Nov 2016 13:44:13 +0000 (09:44 -0400)]
elput: Add API to allow setting pointer acceleration profile

This patch adds an API that can be used to set a pointer acceleration
profile. This API should be used when addressing T4736

ref T4736

@feature

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoeolian: rename ref to ptr to avoid confusion with eo refs
Daniel Kolesa [Wed, 2 Nov 2016 12:06:38 +0000 (13:06 +0100)]
eolian: rename ref to ptr to avoid confusion with eo refs

8 years agonaviframe: Fix to process title show/hide signal immediately
Jaehyun Cho [Wed, 2 Nov 2016 10:51:47 +0000 (19:51 +0900)]
naviframe: Fix to process title show/hide signal immediately

8 years agoelm_popup: before getting focus, all sub objs should be shown
WooHyun Jung [Wed, 2 Nov 2016 00:28:26 +0000 (09:28 +0900)]
elm_popup: before getting focus, all sub objs should be shown

8 years agounbreak build for OpenSSL < 1.0.2
Gustavo Sverzut Barbieri [Tue, 1 Nov 2016 21:36:20 +0000 (19:36 -0200)]
unbreak build for OpenSSL < 1.0.2

not the ideal solution, but we need a decision if we're going to copy
the long code from OpenSSL into our library just to support legacy
users, given that Efl.Net targeted at the future.

8 years agoelementary: not every call are from ecore_job infrastructure, NULL only when they...
Cedric BAIL [Tue, 1 Nov 2016 18:36:10 +0000 (11:36 -0700)]
elementary: not every call are from ecore_job infrastructure, NULL only when they are.

This is something that our tests suite detected. Please pay attention.

8 years agoedje.object: implement Efl.Observer interface
Jee-Yong Um [Tue, 1 Nov 2016 17:59:09 +0000 (10:59 -0700)]
edje.object: implement Efl.Observer interface

Summary:
To remove duplicated lines to handle edje class (color, text, size),
observer interface is implemented to Edje.Object.

Reviewers: jpeg, cedric

Reviewed By: cedric

Subscribers: bu5hm4n, cedric

Differential Revision: https://phab.enlightenment.org/D4359

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoelementary entry: keep style user when new theme is applied
Youngbok Shin [Tue, 1 Nov 2016 17:54:16 +0000 (10:54 -0700)]
elementary entry: keep style user when new theme is applied

Summary:
The style user should be kept when entry's mode is changed.
@fix

Test Plan:
1. Run "elementary_test -to "entry style user"
2. Click "Singleline Mode" toggle
3. See the result

Reviewers: raster, tasn, herdsman, cedric

Reviewed By: cedric

Subscribers: jpeg

Differential Revision: https://phab.enlightenment.org/D4366

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoelm_multibuttonentry: send "clicked" signal when item got hardware enter key.
Woochan Lee [Tue, 1 Nov 2016 17:51:27 +0000 (10:51 -0700)]
elm_multibuttonentry: send "clicked" signal when item got hardware enter key.

Summary:
It was kind of bug state before.

There was no any action when the user gives a focus on an item. then press the enter key.
The item should be set as selected state after that should call clicked signal.

@fix

Test Plan: elementary test MBE sample.

Reviewers: woohyun, bu5hm4n, Hermet

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D4368

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
8 years agoefl_net_server support systemd socket activation.
Gustavo Sverzut Barbieri [Tue, 1 Nov 2016 18:01:57 +0000 (16:01 -0200)]
efl_net_server support systemd socket activation.

It includes extensive verifications to avoid mistakes and usage of
incorrect sockets.

8 years agoMerge branch 'devs/bdilly/ecore_evas_wayland'
Bruno Dilly [Tue, 1 Nov 2016 18:07:05 +0000 (16:07 -0200)]
Merge branch 'devs/bdilly/ecore_evas_wayland'

This series adds multiseat support to wayland engine.

Evas devices are created for seats (parent), keyboard, mouses and
touch input devices.

This way an application would be able to differentiate between
source seat of input events.

Reviewed By: devilhorns, ManMower, iscaro

Differential Revision: https://phab.enlightenment.org/D4370

@feature

8 years agoexamples/ecore: add example / test for wayland multiseat
Bruno Dilly [Fri, 7 Oct 2016 18:57:04 +0000 (15:57 -0300)]
examples/ecore: add example / test for wayland multiseat

8 years agoecore_wl2: associate evas devices to input ecore events
Bruno Dilly [Thu, 6 Oct 2016 20:43:12 +0000 (17:43 -0300)]
ecore_wl2: associate evas devices to input ecore events

So users would be able to differentiate between source
seats.

8 years agoecore_wl2: add ecore event for seat capabilities change
Bruno Dilly [Wed, 5 Oct 2016 13:36:19 +0000 (10:36 -0300)]
ecore_wl2: add ecore event for seat capabilities change

So when mouse / keyboard are present or not it will
generate events.

ecore_evas/wayland will handle that creating or
deleting evas devices for each one (seat device
will be used as parent).

8 years agoecore_wl2: add ecore event for seat name change
Bruno Dilly [Tue, 4 Oct 2016 18:19:51 +0000 (15:19 -0300)]
ecore_wl2: add ecore event for seat name change

And handle it on ecore_evas/wayland, properly
setting the evas device names.

8 years agoecore_evas/wayland: handle added / removed seats
Bruno Dilly [Mon, 3 Oct 2016 15:34:10 +0000 (12:34 -0300)]
ecore_evas/wayland: handle added / removed seats

Create or delete evas_devices with class EVAS_DEVICE_CLASS_SEAT
for seats on each ecore_evas created so far.

Initially it's named considering its Wayland id.

8 years agoecore_evas/wayland: remove unnecessary NULL attributions
Bruno Dilly [Mon, 3 Oct 2016 14:19:22 +0000 (11:19 -0300)]
ecore_evas/wayland: remove unnecessary NULL attributions

When freeing wdata there is no reason to set each
freed field to NULL.

8 years agoevas: Fix doxygen and normalize function signature
Bruno Dilly [Mon, 3 Oct 2016 13:21:47 +0000 (10:21 -0300)]
evas: Fix doxygen and normalize function signature

On recently added function evas_device_add_full()

8 years agoEvas Device: Avoid invalid ptr indirection.
Guilherme Iscaro [Tue, 1 Nov 2016 18:03:52 +0000 (16:03 -0200)]
Evas Device: Avoid invalid ptr indirection.

Summary:
When Evas is deleted the function _evas_device_cleanup() goes thru all
devices and unref them. Since Evas_Devices are Efl_Input_Device, the user
may still hold a reference to the device (efl_ref()),
thus causing the device to do not be deleted *yet*.
This causes a problem, because when the user calls efl_unref()
and the device itself is deleted the Evas _del_cb
callback will be called and will try to access the Evas_Public_Data from
a deleted object.
In order to avoid this problem all devices will be kept in the devices
list and Evas will unregister the EFL_EVENT_DEL from those devices that
were not deleted.

Reviewers: jpeg, bdilly, barbieri, cedric

Reviewed By: bdilly, cedric

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D4369

8 years agoelementary: Use software engine for wayland client pointers
Chris Michael [Tue, 1 Nov 2016 14:52:49 +0000 (10:52 -0400)]
elementary: Use software engine for wayland client pointers

There seem to be an issue with the ecore_evas_wayland_egl engine when
using them for cursors. The issue is that a black square shows up
behind the mouse pointer. This does not happen with the wayland_shm
engine so use wayland_shm engine for mouse pointers (for now) until
this can be sorted out.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agodocs: add missing docs for the rest of efl interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:18:26 +0000 (14:18 +0100)]
docs: add missing docs for the rest of efl interfaces

8 years agodocs: efl_text: ad missing docs for efl_text* interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:17:50 +0000 (14:17 +0100)]
docs: efl_text: ad missing docs for efl_text* interfaces

8 years agodocs: efl_image: add missing docs for efl_image_* interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:15:30 +0000 (14:15 +0100)]
docs: efl_image: add missing docs for efl_image_* interfaces

8 years agodocs: efl_pack: add missing docs for efl_pack_* interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:14:19 +0000 (14:14 +0100)]
docs: efl_pack: add missing docs for efl_pack_* interfaces

8 years agodocs: efl_io: add missing docs for efl_io_* interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:13:42 +0000 (14:13 +0100)]
docs: efl_io: add missing docs for efl_io_* interfaces

8 years agodocs: efl_gfx: add missing docs for efl_gfx_* interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:11:52 +0000 (14:11 +0100)]
docs: efl_gfx: add missing docs for efl_gfx_* interfaces

8 years agodocs: efl_ui: add missing docs for efl_ui_* interfaces
Stefan Schmidt [Tue, 1 Nov 2016 13:11:10 +0000 (14:11 +0100)]
docs: efl_ui: add missing docs for efl_ui_* interfaces

8 years agoexamples/ecore: command to generate OpenSSL PEM files.
Gustavo Sverzut Barbieri [Tue, 1 Nov 2016 13:52:21 +0000 (11:52 -0200)]
examples/ecore: command to generate OpenSSL PEM files.

it's cumbersome to remember these commands and without them it's hard
to test the SSL server examples, so add a make rule for that.

8 years agoefl_net_server_example: oops, do not assume all servers are Efl.Net.Server.Fd
Gustavo Sverzut Barbieri [Tue, 1 Nov 2016 12:45:38 +0000 (10:45 -0200)]
efl_net_server_example: oops, do not assume all servers are Efl.Net.Server.Fd

SSL server is not an Fd, it contains one internally, but that's
hidden. So call proper methods.

8 years agoScroller test: add a page_size spinner
Dave Andreoli [Tue, 1 Nov 2016 08:53:39 +0000 (09:53 +0100)]
Scroller test: add a page_size spinner

and also #define the PAGESIZE

8 years agoelm: Fix disappearance of some elm images
Jean-Philippe Andre [Tue, 1 Nov 2016 06:49:20 +0000 (15:49 +0900)]
elm: Fix disappearance of some elm images

After the commit 97c9fa64a461c (Remove group_show and group_hide),
some Efl.Ui.Image objects would not render properly. The reason
being that the object call to show() was aborted too early when
the image is still preloading.

This made for really random results as an image preload could
take more or less time, depending on chance.

8 years agoefl_net_{server,dialer}_ssl: TCP + SSL easy to use.
Gustavo Sverzut Barbieri [Tue, 1 Nov 2016 03:31:56 +0000 (01:31 -0200)]
efl_net_{server,dialer}_ssl: TCP + SSL easy to use.

in the previous commit we're manually upgrading an existing TCP socket
to SSL. It is desired since some protocols need to negotiate, like
STARTTLS and the likes

Now we offer 2 classes that does autostart SSL once the socket is
ready.

8 years agoefl_net_ssl: fix build on LibreSSL and BSD
Gustavo Sverzut Barbieri [Tue, 1 Nov 2016 00:28:41 +0000 (22:28 -0200)]
efl_net_ssl: fix build on LibreSSL and BSD

As usual thanks to @netstar for quickly spotting the problem :-)

Fixes: T4811

8 years agoefl_net_socket_ssl: initial SSL wrapper.
Gustavo Sverzut Barbieri [Sat, 29 Oct 2016 00:48:19 +0000 (22:48 -0200)]
efl_net_socket_ssl: initial SSL wrapper.

This is the first step towards SSL connections on top of sockets, with
an example on how to upgrade a dialer and a server client using TCP.

8 years agoefl_io_copier: do not ERROR on EAGAIN.
Gustavo Sverzut Barbieri [Fri, 28 Oct 2016 23:11:17 +0000 (21:11 -0200)]
efl_io_copier: do not ERROR on EAGAIN.

As done by write, if we try to read and we can't, then don't give
up. This happens with streams that wraps another, like SSL, may report
there are data to read, but once you try it may not result in enough
data to upper layers.

8 years agoefl_net_socket_fd: do not act if already closed, do not set eos.
Gustavo Sverzut Barbieri [Fri, 28 Oct 2016 00:26:02 +0000 (22:26 -0200)]
efl_net_socket_fd: do not act if already closed, do not set eos.

It's pointless to reset eos, if it was set, keep it like that.

8 years agoecore_getopt: allow empty strings as parameters.
Gustavo Sverzut Barbieri [Thu, 27 Oct 2016 22:57:47 +0000 (20:57 -0200)]
ecore_getopt: allow empty strings as parameters.

Sometimes we want to specify an empty string, that should be allowed.

8 years agoremove incorrect EINA_UNUSED.
Gustavo Sverzut Barbieri [Thu, 27 Oct 2016 14:05:50 +0000 (12:05 -0200)]
remove incorrect EINA_UNUSED.

8 years agoecore-wl2: Unify surface creation code
Chris Michael [Mon, 31 Oct 2016 18:03:30 +0000 (14:03 -0400)]
ecore-wl2: Unify surface creation code

As we call the same code during ecore_wl2_window_surface_get and
ecore_wl2_window_show functions that basically create the wl_surface
for a given window we can unify that code into one function that can
be called from various places. This also fixes an issue inside
ecore_wl2_window_show where the window surface_id may not have been
getting filled properly.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agodocs: efl_canvas: fill in missing docs for efl_canvas interface
Stefan Schmidt [Mon, 31 Oct 2016 13:21:06 +0000 (14:21 +0100)]
docs: efl_canvas: fill in missing docs for efl_canvas interface

8 years agodocs: efl_animator: fill in missing docs for efl_animator interface
Stefan Schmidt [Mon, 31 Oct 2016 13:20:44 +0000 (14:20 +0100)]
docs: efl_animator: fill in missing docs for efl_animator interface

8 years agodocs: efl_gfx: fill in missing docs for efl_gfx interface
Stefan Schmidt [Mon, 31 Oct 2016 13:02:25 +0000 (14:02 +0100)]
docs: efl_gfx: fill in missing docs for efl_gfx interface

8 years agoAdd 2 buttons in the Scroller test to reveal one more bug
Dave Andreoli [Mon, 31 Oct 2016 16:35:05 +0000 (17:35 +0100)]
Add 2 buttons in the Scroller test to reveal one more bug

the buttons should scroll up and down by 1 px, but nothing move here

That test was still broken in other ways, for example:
play a bit with the "to X Y" buttons and see if it always do the correct thing

8 years agoedje_edit: generate to source code base_scale if it different from 1.0
Vyacheslav Reutskiy [Mon, 31 Oct 2016 11:16:36 +0000 (13:16 +0200)]
edje_edit: generate to source code base_scale if it different from 1.0

Fixes T4767

8 years agoemotion gst1 module - disable subtitles by default as that should be
Carsten Haitzler (Rasterman) [Mon, 31 Oct 2016 10:46:06 +0000 (19:46 +0900)]
emotion gst1 module - disable subtitles by default as that should be

there are spu apis to turn subtitles on and off and this should be off
until turned on by api. you really have to be able to choose the
subtitles to display - eg language etc. to use them effectively.

this fixes T4795

@fix

8 years agoecore audio - fix hang in wayland due to pulse audio connecting to x
Carsten Haitzler (Rasterman) [Mon, 31 Oct 2016 04:55:07 +0000 (13:55 +0900)]
ecore audio - fix hang in wayland due to pulse audio connecting to x

pulse insists on connecting to the xserver on init/setup context if:

1. DISPLAY is set
AND
2. DISPLAY is not empty

so to do a pretty horrible worka-round, empty off the display if its
set so pa doesnt go connect to x and do this if WAYLAND_DISPLAy is set
assuming we'll use wayland then. this is far better than a solid
rock-hard hang. :)

@fix

8 years agoelm_entry: Add voice input panel layout
InHong Han [Wed, 3 Aug 2016 07:40:24 +0000 (16:40 +0900)]
elm_entry: Add voice input panel layout

@feature

Change-Id: I41502d2446b95a4be31fc60a8d995b9d37930844
Signed-off-by: InHong Han <inhong1.han@samsung.com>
8 years agoecore_audio: deprecating enum is apparently not supported in every version of gcc.
Cedric Bail [Sun, 30 Oct 2016 22:49:54 +0000 (15:49 -0700)]
ecore_audio: deprecating enum is apparently not supported in every version of gcc.

8 years agotests: fix eolian_cxx tests with clang
Jean Guyomarc'h [Sun, 30 Oct 2016 18:12:24 +0000 (19:12 +0100)]
tests: fix eolian_cxx tests with clang

eolian_cxx tests failed to link because of massively inexistant symbols.

I assume eolian_cxx tests have been working at some point. Maybe they
were gcc-only? I don't get what's going on with gcc and non-existant
symbols. Is there a sugar-coating of some kind? Because when a symbol
does not exist, clang throws you away. Is it because we are only
referring to the eo implementation functions via function pointers?
That's the second time I'm doing a fix like this. Maybe we should change
a bit our linking flags (see --unresolved-symbols in ld)??

Anyway, now we have our symbols. Clang is happy, make check can go on...

8 years agoecore_con: fix structure declaration
Jean Guyomarc'h [Sun, 30 Oct 2016 17:40:54 +0000 (18:40 +0100)]
ecore_con: fix structure declaration

Clang raised a massive amount of warnings due to the struct sockaddr_un
not being declared before using it. So, include the header that declares
this structure first.

8 years agoecore_cocoa: fix handling of some keys
Jean Guyomarc'h [Sun, 30 Oct 2016 15:35:56 +0000 (16:35 +0100)]
ecore_cocoa: fix handling of some keys

The ascii circumflex (^) can be typed by pressing twice the ^ key on a
mac keyboard. A single press allows composition (e.g. ^+e = ê).
Pressing ^ twice though, led to a segmentation fault in elementary,
because the result character of the operation (^) appeared in the raw
characters stack, and not in the filtered one.
This is a bit weird, as backtick (`) appears in the filtered keys stack.

@fix

8 years agoecore_cocoa: prettify code a bit
Jean Guyomarc'h [Sun, 30 Oct 2016 15:34:35 +0000 (16:34 +0100)]
ecore_cocoa: prettify code a bit

Remove extraneous whitespaces, use appropriate macros when possible,
and mark a memory allocation failure as an unlikely case.

8 years agoecore_cocoa: fix behaviour of option key
Jean Guyomarc'h [Sun, 30 Oct 2016 15:18:41 +0000 (16:18 +0100)]
ecore_cocoa: fix behaviour of option key

Commit e44c48b90408d2518e2708090796988cfd3cacea failed to translate the
deprecated API into the Sierra API... replacing the Command key flags
by the Option key flags. This resulted of Opt+q quitting the program.

@fix

8 years agoecore_cocoa: allow to override termination sequence
Jean Guyomarc'h [Sun, 2 Oct 2016 12:22:53 +0000 (14:22 +0200)]
ecore_cocoa: allow to override termination sequence

8 years agoautotools: stop summoning systemd's infernal cow
Jean Guyomarc'h [Sat, 29 Oct 2016 21:31:28 +0000 (23:31 +0200)]
autotools: stop summoning systemd's infernal cow

Systemd support has now been enabled by default on all capable
platforms. By explicitely providing --disable-systemd, one can disable
its integration to EFL.
When I litterally write that I don't want systemd support, please
don't tell me to enable it. It's almost a passive aggression there ;)

8 years agoecore_audio: drop support for CoreAudio on macOS
Jean Guyomarc'h [Sat, 29 Oct 2016 20:32:19 +0000 (22:32 +0200)]
ecore_audio: drop support for CoreAudio on macOS

CoreAudio support was initially introduced by commit
62e29b39f4df40fd3c0a6b17f7a16f0f8fc1d0c9 as an experimental feature.

It played basic sounds, but suffered from drawbacks: it was controlling
the master channel, and therefore any sound played by ecore_audio would
shut down a previous sound (e.g. background music) for the time of the
sound being played. So that wasn't exactly great... Also, after some
time, some hangs have been reported when playing a sound on input. Most
of the time, it translated as a pause in the main loop (see T3797).
More recently (several months ago), ecore_audio with CoreAudio stopped
working during 1.19 development...

So... CoreAudio support on macOS has never been great. And now it's fully
broken. Instead of trying to revive the thing, let just use PulseAudio.
PulseAudio can be installed without any trouble on macOS thanks to
package managers such as Homebrew. Actually, the efl package provided by
Homebrew already provides PulseAudio as a dependency. And it actually
just works very fine. Dropping CoreAudio seems therefore a nice option:
removes unmaintained code, fixes bugs, and add features.

8 years agoelm win - add ifdef around wl code so it compiles with wl off
Carsten Haitzler (Rasterman) [Sat, 29 Oct 2016 01:15:40 +0000 (10:15 +0900)]
elm win - add ifdef around wl code so it compiles with wl off

8 years agoeio model: remove useless efreet_mime_type_get call
Vitor Sousa [Fri, 28 Oct 2016 23:19:37 +0000 (21:19 -0200)]
eio model: remove useless efreet_mime_type_get call

8 years agoelm_icon: avoid useless assignment
Jean Guyomarc'h [Fri, 28 Oct 2016 21:23:15 +0000 (23:23 +0200)]
elm_icon: avoid useless assignment

Assigning id->edje to EINA_TRUE in a code path triggered when id->edje
is EINA_FALSE instead of unconditionnaly setting it to EINA_TRUE avoids
to assign id->edje to itself.

8 years agoelm_icon: slightly clean-up code
Jean Guyomarc'h [Fri, 28 Oct 2016 21:20:14 +0000 (23:20 +0200)]
elm_icon: slightly clean-up code

It is unnecessary to use an initialized variable on the stack to hold a
new value and then immediately return it.

8 years agoLuncher: Improve indicator theme.
Stephen okra Houston [Fri, 28 Oct 2016 20:04:43 +0000 (15:04 -0500)]
Luncher: Improve indicator theme.

8 years agoelementary_theme: Add 'xterm' cursor into theme
Chris Michael [Fri, 28 Oct 2016 18:27:50 +0000 (14:27 -0400)]
elementary_theme: Add 'xterm' cursor into theme

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoelementary: Remove debug printing
Chris Michael [Fri, 28 Oct 2016 18:27:17 +0000 (14:27 -0400)]
elementary: Remove debug printing

NB: No functional changes, just removing left over prints

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoLuncher: Add the necessary theme files for Enlightenment's new launcher
Stephen okra Houston [Fri, 28 Oct 2016 18:20:41 +0000 (13:20 -0500)]
Luncher: Add the necessary theme files for Enlightenment's new launcher

8 years agoelementary: Make setting cursors for EFL Wayland client apps work
Chris Michael [Fri, 28 Oct 2016 18:01:47 +0000 (14:01 -0400)]
elementary: Make setting cursors for EFL Wayland client apps work

This patch changes els_cursor.c to work with the pointer object
created for EFL Wayland Application windows. This allows any EFL cursors
specified by an application to work in Wayland (ex: elm_test Cursor
tests)

NB: While the code is working, there will still be missing cursors due
to those not being included (yet) in the elm theme. Will continue to
add those as time goes.

@feature

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoelementary_theme: Standardize cursor hotspot swallow names
Chris Michael [Fri, 28 Oct 2016 18:00:55 +0000 (14:00 -0400)]
elementary_theme: Standardize cursor hotspot swallow names

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoelementary: Add internal function to set wayland cursor
Chris Michael [Fri, 28 Oct 2016 17:59:31 +0000 (13:59 -0400)]
elementary: Add internal function to set wayland cursor

Add an internal elm function we can call from withint els_cursor.c so
that we can set window cursor/pointer images based on what is supplied
by els_cursor.

@feature

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore_evas: refactor VNC as an Eina Module.
Guilherme Iscaro [Fri, 28 Oct 2016 16:56:42 +0000 (09:56 -0700)]
ecore_evas: refactor VNC as an Eina Module.

Summary:
This change removes the necessity to link EFL against the libvncserver

Please ignore the first three commits, they're being reviewed here:

https://phab.enlightenment.org/D4323

Reviewers: bdilly, cedric

Reviewed By: cedric

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D4338

Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
8 years agoelementary: Fix build break for non-wayland builds
Chris Michael [Fri, 28 Oct 2016 16:34:18 +0000 (12:34 -0400)]
elementary: Fix build break for non-wayland builds

This fixes an issue where build would be broken if not building with
Wayland support.

Fixes T4778

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: Only call pointer_update_stop once
Chris Michael [Fri, 28 Oct 2016 15:27:59 +0000 (11:27 -0400)]
ecore-wl2: Only call pointer_update_stop once

As we already call _ecore_wl2_input_cursor_update_stop above, we
actually don't need to call it a second time.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoobj caching - add a reuse in key event objects too to nuke leaks
Carsten Haitzler (Rasterman) [Fri, 28 Oct 2016 13:57:25 +0000 (22:57 +0900)]
obj caching - add a reuse in key event objects too to nuke leaks

new leak since sink was added.

8 years agoimprove responsivness of timer sleeping threads for vsync with prctl
Carsten Haitzler (Rasterman) [Fri, 28 Oct 2016 05:50:05 +0000 (14:50 +0900)]
improve responsivness of timer sleeping threads for vsync with prctl

prctl allows us on some platforms to request a thread be woken up more
agressively e.g. due to a timeout bu setting timerslack. since we use
a dedicated thread just for vsync events, this is a very good idea to
ask the kernel to be as exact as possible for this thread as it only
wakes up once per frame (or should only) and accuracy is important. so
use this.

also improve prctl checks to be more explicit in configure.ac and use
these ifdefs in ecore exe too where prctl is used as well.

@feature

8 years agoelementary: Don't set "move" cursor on wayland client apps
Chris Michael [Fri, 28 Oct 2016 13:22:26 +0000 (09:22 -0400)]
elementary: Don't set "move" cursor on wayland client apps

As most (if not all) compositors will draw their own "move" cursor
when moving windows around, there's no real need to set this one
client side.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoelementary: Fix issue of extra pointer being displayed
Chris Michael [Fri, 28 Oct 2016 13:06:48 +0000 (09:06 -0400)]
elementary: Fix issue of extra pointer being displayed

As we only want to show our custom EFL cursors when the mouse is
actually inside the window, use the _elm_win_mouse_in/out functions to
control pointer visibility.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoevas-wayland-shm: Don't post updates to surface if no surface
Chris Michael [Fri, 28 Oct 2016 12:57:10 +0000 (08:57 -0400)]
evas-wayland-shm: Don't post updates to surface if no surface

In the event that an ecore_evas (using wayland_shm) gets hidden then
the corresponding wl_surface gets destroyed. If evas_norender is
called after that, the outbuf_redraws_clear function follows.
Outbuf_redraw_clear function ends up trying to post updates to the
wl_surface however if that wl_surface is gone, then we end up crashing.

This patch addresses that issue by checking for a valid wl_surface
inside the engine before trying to post updates to that wl_surface.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoevas-wayland-shm: Fix formatting
Chris Michael [Fri, 28 Oct 2016 12:53:17 +0000 (08:53 -0400)]
evas-wayland-shm: Fix formatting

NB: No functional changes

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: Disable pointer frame callback if no cursor surface
Chris Michael [Thu, 27 Oct 2016 18:56:36 +0000 (14:56 -0400)]
ecore-wl2: Disable pointer frame callback if no cursor surface

If we have no cursor surface, then we don't need the pointer frame
callback anymore so call cursor_update_stop which will delete the
pointer frame callback

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: Don't create cursor frame callback if there is no cursor
Chris Michael [Thu, 27 Oct 2016 17:48:56 +0000 (13:48 -0400)]
ecore-wl2: Don't create cursor frame callback if there is no cursor
surface

In the event that we have no cursor surface, then we should not be
creating a cursor frame callback.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoelementary: Don't create pointer canvas for inlined_image
Chris Michael [Thu, 27 Oct 2016 17:21:23 +0000 (13:21 -0400)]
elementary: Don't create pointer canvas for inlined_image

As we don't need a pointer or pointer canvas for
ELM_WIN_INLINED_IMAGE, add missing type check and skip creating
pointer canvas there.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoelementary: Use separate checks for pointer object vs pointer canvas
Chris Michael [Thu, 27 Oct 2016 17:13:32 +0000 (13:13 -0400)]
elementary: Use separate checks for pointer object vs pointer canvas

As under the drm engine a softcursor will be created, but not a
pointer canvas, we need to use separate if checks in certain places
like showing, hiding the mouse pointer/canvas.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
8 years agoecore-wl2: only update pointer frame if there is a pointer surface
Mike Blumenkrantz [Fri, 28 Oct 2016 13:53:29 +0000 (09:53 -0400)]
ecore-wl2: only update pointer frame if there is a pointer surface

fix T4777, T4776

8 years agoEdje_Edit: save files name instead of full path on sound add
Vitalii Vorobiov [Fri, 28 Oct 2016 12:56:28 +0000 (15:56 +0300)]
Edje_Edit: save files name instead of full path on sound add

Since file will be inside of edj file, there is no need in having full path to
the place from where it was imported

@fix

8 years agoInput events cache: use the new mechanism to reuse eo objects.
Tom Hacohen [Fri, 28 Oct 2016 12:20:56 +0000 (13:20 +0100)]
Input events cache: use the new mechanism to reuse eo objects.

8 years agoEo: Add a method to mark objects for reuse.
Tom Hacohen [Fri, 28 Oct 2016 12:19:10 +0000 (13:19 +0100)]
Eo: Add a method to mark objects for reuse.

This informas eo an object is going to get reused/cached, so eo can
reset the object appropriately.

@feature.

8 years agoecore-animator allow for animator to skip queued animator ticks
Carsten Haitzler (Rasterman) [Fri, 28 Oct 2016 04:58:56 +0000 (13:58 +0900)]
ecore-animator allow for animator to skip queued animator ticks

set ECORE_ANIMATOR_SKIP to skip queued animtor ticks if multiple are
in the pipeline. optional and not on by default. i would think its not
a good idea to skip these animator ticks and skipping/deferring is a
job higher up.

@feature