platform/upstream/efl.git
10 years agoeet_data: guarantee double-word alignment for temporary stack buffer.
Gustavo Sverzut Barbieri [Mon, 20 Jan 2014 23:37:09 +0000 (21:37 -0200)]
eet_data: guarantee double-word alignment for temporary stack buffer.

The code was giving enough memory to store doubles and longs, but they
could be unaligned as "unsigned char" allows 1-byte alignment, while
double may require 8 bytes.

By specifying the array as "long long" we force certain alignment in a
platform independent way. As this array is small enough and
short-lived, the number of items were not changed, this results in
more bytes on the stack but it shouldn't matter.

10 years agoeina/mp/one_big: fix alignment issues.
Gustavo Sverzut Barbieri [Mon, 20 Jan 2014 23:27:58 +0000 (21:27 -0200)]
eina/mp/one_big: fix alignment issues.

When over-allocating (past "pool->max" items) a memory slice will be
allocated to the new item as a linked list using Eina_Inlist.

The original code was placing the Eina_Inlist structure (3 pointers)
at the beginning of the allocated memory. However the item must have
proper alignment based on "pool->item_size", otherwise a structure may
end with unaligned members. Take for example MIPS 32 bits, it uses 4
bytes pointers with 8 bytes double. A structure containing a double
could have it unaligned as 12 % 8 = 4 (12 is the size of Eina_Inlist,
that contains 3 pointers), and MIPS doesn't allow unaligned access.

Albin Tonnerre (Lutin) spotted this in his Debian MIPS test machine,
it was breaking at eet_data_get_double() that was storing an unaligned
double. This was being called from within edje test suite.

The current code will place the list node after the requested
"pool->item_size", of course guaranteeing the pointer inside the node
is aligned (otherwise a "char" or "short" would break its alignment).

10 years agoecore_getopt: support for per-category help listing
Daniel Kolesa [Mon, 20 Jan 2014 17:10:21 +0000 (17:10 +0000)]
ecore_getopt: support for per-category help listing

10 years agoconfigure: include required NEON build flags.
Gustavo Sverzut Barbieri [Mon, 20 Jan 2014 14:15:56 +0000 (12:15 -0200)]
configure: include required NEON build flags.

It is required to use "-mfpu=neon" to build NEON.

Patch-by: Albin Tonnerre <lutin@debian.org>
10 years ago[ecore_x] Fix to handle paste null data case.
Jaehyun Cho [Mon, 20 Jan 2014 13:15:33 +0000 (22:15 +0900)]
[ecore_x] Fix to handle paste null data case.

Summary: Do nothing when null data is pasted.

Reviewers: Hermet, raster

Reviewed By: raster

CC: seoz, raster, cedric
Differential Revision: https://phab.enlightenment.org/D473

10 years agoecore - fix CRIT -> CRI macro
Carsten Haitzler (Rasterman) [Mon, 20 Jan 2014 11:23:22 +0000 (20:23 +0900)]
ecore - fix CRIT -> CRI macro

10 years agoedje: Call _edje_entry_imf_context_reset in _edje_entry_user_insert.
Youngbok Shin [Mon, 20 Jan 2014 06:59:50 +0000 (15:59 +0900)]
edje: Call _edje_entry_imf_context_reset in _edje_entry_user_insert.

Summary:
When user paste text to elm_entry, _edje_entry_user_insert is called.
We need to reset a prediction buffer just like _edje_entry_text_markup_insert.

Reviewers: seoz, Hermet, woohyun, tasn, jihoon, raster

Reviewed By: raster

CC: cedric
Differential Revision: https://phab.enlightenment.org/D461

10 years agoeina-inlist - handle prepend/append relative if list is null (empty)
Carsten Haitzler (Rasterman) [Mon, 20 Jan 2014 06:16:56 +0000 (15:16 +0900)]
eina-inlist - handle prepend/append relative if list is null (empty)

stable release - cherry-pick me!

10 years agoeet: do not load data when the file is open in read/write mode.
Cedric BAIL [Mon, 20 Jan 2014 03:03:02 +0000 (12:03 +0900)]
eet: do not load data when the file is open in read/write mode.

I do think that it was a left over from previous file format. Removing
this memcpy should make Enlightenment startup faster and should reduce
by 500KB the memory it use.

10 years agoeina: add eina_{str,bin}buf_manage_read_only_new_length().
Cedric BAIL [Thu, 2 Jan 2014 06:36:23 +0000 (15:36 +0900)]
eina: add eina_{str,bin}buf_manage_read_only_new_length().

It is sometime useful to start from a defined buffer, but to not touch it
until needed. This make life of caller more easier as they don't need to
duplicate the buffer themself as Eina will now take care of that.

10 years agoeina: add eina_binbuf_append_buffer() and eina_strbuf_append_buffer().
Cedric BAIL [Thu, 2 Jan 2014 03:16:26 +0000 (12:16 +0900)]
eina: add eina_binbuf_append_buffer() and eina_strbuf_append_buffer().

10 years agoeet: reduce memory usage of Eet internal structure by removing useless structure.
Cedric BAIL [Fri, 17 Jan 2014 09:55:14 +0000 (18:55 +0900)]
eet: reduce memory usage of Eet internal structure by removing useless structure.

10 years agoedje: Add changed signal emit in _edje_entry_imf_event_delete_surrounding_cb
Youngbok Shin [Mon, 20 Jan 2014 03:54:45 +0000 (12:54 +0900)]
edje: Add changed signal emit in _edje_entry_imf_event_delete_surrounding_cb

Summary:
"_edje_entry_imf_event_delete_surrounding_cb" changes text of entry.
When the callback function is called and the entry is changed,
we need to notice to applications that the entry is changed.

Reviewers: seoz, Hermet, tasn, woohyun, jihoon, raster

Reviewed By: raster

CC: cedric
Differential Revision: https://phab.enlightenment.org/D460

10 years agoedje_cc: factorize two strlen()s
Jean Guyomarc'h [Mon, 20 Jan 2014 03:47:46 +0000 (12:47 +0900)]
edje_cc: factorize two strlen()s

Reviewers: cedric, raster

Reviewed By: raster

CC: cedric
Differential Revision: https://phab.enlightenment.org/D454

10 years agoecore_evas_wayland_common: fix compilation, missing bits of 892b1c5
Jérémy Zurcher [Sun, 19 Jan 2014 18:41:27 +0000 (19:41 +0100)]
ecore_evas_wayland_common: fix compilation, missing bits of 892b1c5

10 years agoFix doxy typo
Chris Michael [Thu, 16 Jan 2014 09:23:19 +0000 (09:23 +0000)]
Fix doxy typo

Signed-off-by: Chris Michael <devilhorns@comcast.net>
10 years agoecore_evas - applied macro to check rotation state.
ChunEon Park [Sat, 18 Jan 2014 13:26:10 +0000 (22:26 +0900)]
ecore_evas - applied macro to check rotation state.

also, let the potrait compare on the higher priority.

if you can suggest better macro name, then please modify it.

10 years agoRevert "ecore_evas - use the macro orthogonal check."
ChunEon Park [Sat, 18 Jan 2014 13:04:25 +0000 (22:04 +0900)]
Revert "ecore_evas - use the macro orthogonal check."

This reverts commit 43acf1e82f94dcdca25d6781d8b27b015f4bd656.

some partial changes are incorrectly applied.

10 years agoecore_evas - use the macro orthogonal check.
ChunEon Park [Sat, 18 Jan 2014 12:53:56 +0000 (21:53 +0900)]
ecore_evas - use the macro orthogonal check.

10 years agoefl - updated AUTHORS
ChunEon Park [Sat, 18 Jan 2014 10:12:45 +0000 (19:12 +0900)]
efl - updated AUTHORS

10 years agoevas: Fixed to compare current render_op with previous render_op for textblock. If...
Youngbok Shin [Sat, 18 Jan 2014 10:06:16 +0000 (19:06 +0900)]
evas: Fixed to compare current render_op with previous render_op for textblock. If the render_op is changed, call evas_object_render_pre_prev_cur_add.

Summary: This change is related to the commit: cdb86e34bbf82bd91aefd686bfc3bdf2a018a956

Reviewers: woohyun, tasn, Hermet, seoz

CC: cedric
Differential Revision: https://phab.enlightenment.org/D450

10 years agoefl/ecore_con - compare same types.
ChunEon Park [Sat, 18 Jan 2014 09:50:51 +0000 (18:50 +0900)]
efl/ecore_con - compare same types.

10 years agoefl/emotion - added missing headers for getuid() geteuid().
ChunEon Park [Sat, 18 Jan 2014 09:49:27 +0000 (18:49 +0900)]
efl/emotion - added missing headers for getuid() geteuid().

10 years agopkgbuild -> luajit
Carsten Haitzler (Rasterman) [Sat, 18 Jan 2014 00:25:43 +0000 (09:25 +0900)]
pkgbuild -> luajit

10 years agoedje: use luajit by default, if you want lua use --enable-lua-old
Daniel Kolesa [Fri, 17 Jan 2014 19:07:59 +0000 (19:07 +0000)]
edje: use luajit by default, if you want lua use --enable-lua-old

10 years agoEvas cserve2: Fix invalid memory access
Jean-Philippe Andre [Fri, 17 Jan 2014 09:07:33 +0000 (18:07 +0900)]
Evas cserve2: Fix invalid memory access

In case of load error, we free the file entry, but forgot to
reset the pointer to it in the image entry.

10 years agoedje: use EINA_UNUSED instead of __UNUSED__
Jihoon Kim [Fri, 17 Jan 2014 09:51:36 +0000 (18:51 +0900)]
edje: use EINA_UNUSED instead of __UNUSED__

10 years agoemotion: Use the GStreamer 1.x module instead of the 0.10 one in the border example...
Sebastian Dröge [Fri, 17 Jan 2014 08:32:45 +0000 (09:32 +0100)]
emotion: Use the GStreamer 1.x module instead of the 0.10 one in the border example application

10 years agoemotion gstreamer1: Call _emotion_frame_new() for audio-only streams without visualis...
Sebastian Dröge [Thu, 16 Jan 2014 21:43:23 +0000 (22:43 +0100)]
emotion gstreamer1: Call _emotion_frame_new() for audio-only streams without visualisations too

Without this applications are not notified of any progress.

https://phab.enlightenment.org/T767

10 years agoedje_entry: When preediting, cursor should not be moved.
WooHyun Jung [Thu, 16 Jan 2014 07:31:36 +0000 (16:31 +0900)]
edje_entry: When preediting, cursor should not be moved.

10 years agoecore_getopt: support for categories
Daniel Kolesa [Wed, 15 Jan 2014 20:22:27 +0000 (20:22 +0000)]
ecore_getopt: support for categories

10 years agoFix ecore_con example file to call shutdown functions.
Guillaume Friloux [Wed, 15 Jan 2014 19:42:50 +0000 (20:42 +0100)]
Fix ecore_con example file to call shutdown functions.

Found by Chris Michael.

10 years agoecore_x_gesture: Cast maek pointer correctly.
Daniel Juyung Seo [Wed, 15 Jan 2014 11:47:49 +0000 (20:47 +0900)]
ecore_x_gesture: Cast maek pointer correctly.

Confirmed by the original author of libxgesture, Sungjin Park.

10 years agoTextblock: Add test cases for style padding
Jean-Philippe Andre [Fri, 10 Jan 2014 02:15:48 +0000 (11:15 +0900)]
Textblock: Add test cases for style padding

In the previous commit, style padding has been changed, so
that lines don't get extra space just because there's a special
style (glow, ...)

This adds some test cases that check the actual geometry of the
lines relatively to each other.

NOTE: This test does not fail before the padding commits, as
_relayout_if_needed() adjusts the padding properly.

10 years agoTextblock: Relayout if needed in line geometry get
Jean-Philippe Andre [Fri, 10 Jan 2014 02:43:04 +0000 (11:43 +0900)]
Textblock: Relayout if needed in line geometry get

evas_object_textblock_line_number_geometry_get() did not trigger
a relayout, so it could end up returning FALSE and invalid values.

10 years agoTextblock: Fix padding with styles (part 2)
Jean-Philippe Andre [Wed, 8 Jan 2014 07:39:46 +0000 (16:39 +0900)]
Textblock: Fix padding with styles (part 2)

Summary:
If ONE single item in the whole textblock has a padding of k,
then ALL the lines of the textblock will be padded by k pixels.

Here's a solution to add the padding only to the first line.

Test Plan:
Write any multiline text, without styles, in an entry.
Add some glow to one element. All lines should be spaced by
an extra 2 pixels.

Reviewers: tasn

CC: cedric
Differential Revision: https://phab.enlightenment.org/D442

10 years agoTextblock: Fix padding computation (part 1)
Jean-Philippe Andre [Wed, 8 Jan 2014 06:28:49 +0000 (15:28 +0900)]
Textblock: Fix padding computation (part 1)

Track padding per paragraph, since this is how it is computed.

Problem before this patch:
- If markup text is changed, padding may grow, and the layout is updated (good)
- If the UI itself needs a relayout, the old padding value is NOT reused,
so style paddings will reset the padding to 0.

Test protocol:
- Set some text with style=glow. The whole object should have padding 2,2,2,2
- Relayout the UI, the whole object will have padding 0,0,0,0 (should be 2,2,2,2)

10 years agoecore_x: set cursor with xinput2
Sebastian Dransfeld [Wed, 15 Jan 2014 07:29:05 +0000 (08:29 +0100)]
ecore_x: set cursor with xinput2

For some reason XDefineCursor might malfunction when using xinput2. So
when using xinput2, use XIDefineCursor. Also see:

https://bugs.launchpad.net/compiz/+bug/1179155

10 years agoEvas/cserve2: Add some safety checks when reading socket messages
Jean-Philippe Andre [Tue, 14 Jan 2014 08:36:54 +0000 (17:36 +0900)]
Evas/cserve2: Add some safety checks when reading socket messages

Fixes CID 1039571 and 1039572.

10 years agoEvas/cserve2: Remove dead code
Jean-Philippe Andre [Tue, 14 Jan 2014 08:11:40 +0000 (17:11 +0900)]
Evas/cserve2: Remove dead code

Fixes CID 1039925.

10 years agoecore xcim module - don't printf every time you start to add stdout noise
Carsten Haitzler (Rasterman) [Wed, 15 Jan 2014 00:55:05 +0000 (09:55 +0900)]
ecore xcim module - don't printf every time you start to add stdout noise

10 years agoecore ecore_x_e: check 0 and -1 for ecore_x_window_prop_window_get(). According to...
Seunghun Lee [Tue, 14 Jan 2014 09:48:16 +0000 (18:48 +0900)]
ecore ecore_x_e: check 0 and -1 for ecore_x_window_prop_window_get(). According to note for ecore_x_window_prop_window_get() :

Summary:
If the property was successfully fetched the number of items stored in
val is returned, otherwise -1 is returned.
Note: Return value 0 means that the property exists but has no elements.

Reviewers: seoz

Reviewed By: seoz

CC: cedric
Differential Revision: https://phab.enlightenment.org/D457

10 years agoEvas/cserve2: Stop looking at those dirty pics!
Jean-Philippe Andre [Tue, 14 Jan 2014 04:59:27 +0000 (13:59 +0900)]
Evas/cserve2: Stop looking at those dirty pics!

What I mean is erm... images marked as dirty don't have a
cache_key, and that's perfectly fine when the image file
has changed.

10 years agoEvas/cserve2: Fix potential crash in image_free
Jean-Philippe Andre [Tue, 14 Jan 2014 04:42:17 +0000 (13:42 +0900)]
Evas/cserve2: Fix potential crash in image_free

Don't try to free an object that was already freed.
Basically, allow image_free(NULL).

10 years agoEio monitor: Fix crash on invalid data access
Jean-Philippe Andre [Tue, 14 Jan 2014 04:33:16 +0000 (13:33 +0900)]
Eio monitor: Fix crash on invalid data access

Fix race condition when touching/changing a (theme) file often.
An Eio_Monitor was marked as "delete_me" but the rename callback
was still called, leading to memory access to already freed
objects.

Test protocol was:
ELM_THEME=~/default.edj elementary_test &
watch touch ~/default.edj

10 years agoEvas/cserve2: Keep image_id when reopening a changed file
Jean-Philippe Andre [Mon, 13 Jan 2014 08:42:14 +0000 (17:42 +0900)]
Evas/cserve2: Keep image_id when reopening a changed file

If a file had changed, a new OPEN message was sent, and the
client image ID was then changed, but the LOAD message was
sent with the previous image ID. So cserve2 would not be
able to honor that request.

10 years agoEvas/cserve2: Clear up File_Data when opening a new image
Jean-Philippe Andre [Mon, 13 Jan 2014 08:22:50 +0000 (17:22 +0900)]
Evas/cserve2: Clear up File_Data when opening a new image

After a file changed, the "changed" flag was not reset in the
File_Data structure, although it was supposed to represent a
new entry altogether.

10 years agoEvas/cserve2: Rename "invalid" field to "changed"
Jean-Philippe Andre [Mon, 13 Jan 2014 08:17:36 +0000 (17:17 +0900)]
Evas/cserve2: Rename "invalid" field to "changed"

This flag indicates that a file changed, not that the entry
is invalid. There is another "valid" flag already.

10 years agoEvas/cserve2: Improve error messages a bit
Jean-Philippe Andre [Mon, 13 Jan 2014 08:15:20 +0000 (17:15 +0900)]
Evas/cserve2: Improve error messages a bit

10 years agoEvas/cserve2: Add another load error check
Jean-Philippe Andre [Mon, 13 Jan 2014 06:52:00 +0000 (15:52 +0900)]
Evas/cserve2: Add another load error check

This one is very unlikely as it would mean the image is
invalid while being in the hash.

10 years agoEvas/cserve2: Fix crash on image file change
Jean-Philippe Andre [Mon, 13 Jan 2014 06:42:14 +0000 (15:42 +0900)]
Evas/cserve2: Fix crash on image file change

When an image file is changed, it is discarded from cserve2,
so the references become invalid. In case we were loading a
scaled version of that image, no proper error checking was
done, leading to obvious crashes.

10 years agoEvas/cserve2: Reduce error logs on invalid glyph load
Jean-Philippe Andre [Mon, 13 Jan 2014 06:06:48 +0000 (15:06 +0900)]
Evas/cserve2: Reduce error logs on invalid glyph load

Change message level from ERR to WRN, when a glyph is not
loadable because FT fails to load it or it contains 0 pixel.

cserve2 used to complain about invalid glyph 3, on a few fonts

10 years agoEvas/cserve2: Fix crash in server on file change
Jean-Philippe Andre [Mon, 13 Jan 2014 05:21:44 +0000 (14:21 +0900)]
Evas/cserve2: Fix crash in server on file change

An inotify callback is triggered when an image file changes,
and it is supposed to cleanup all references to this image.

Unfortunately, it was doing it in a very unsafe way as pointers
could become invalid, because of hash free callbacks in the
referenced image list. Add some list safety and always assume
the pointer might be dead after free operations.

TBH, this _file_changed_cb function looks very confused to me
(as it tries to bypass eina_hash_del).

10 years agoEvas/cserve2: Fix memleak during shutdown
Jean-Philippe Andre [Mon, 13 Jan 2014 02:58:16 +0000 (11:58 +0900)]
Evas/cserve2: Fix memleak during shutdown

Don't reset the dead slave callback, as it is used to free the
slave worker's resources.

10 years agoEvas/cserve2: Use calloc instead of malloc
Jean-Philippe Andre [Mon, 13 Jan 2014 02:44:21 +0000 (11:44 +0900)]
Evas/cserve2: Use calloc instead of malloc

This will silence some warnings from valgrind (and add some
safety as well...)

10 years agoecore-x present fixups
Mike Blumenkrantz [Mon, 13 Jan 2014 17:14:50 +0000 (12:14 -0500)]
ecore-x present fixups

10 years agoecore-x now supports XPRESENT
Mike Blumenkrantz [Fri, 10 Jan 2014 07:26:06 +0000 (02:26 -0500)]
ecore-x now supports XPRESENT

see http://cgit.freedesktop.org/xorg/proto/presentproto/plain/presentproto.txt

10 years agoecore_evas_win32: Fix build regression in ecore evas win32 engine
Stefan Schmidt [Mon, 13 Jan 2014 12:36:28 +0000 (13:36 +0100)]
ecore_evas_win32: Fix build regression in ecore evas win32 engine

Regression introduced with fb59ac34a1ec65680ca4ff3fc62d1fc42aa72f5a
Simple typo during Eina_Bool convertion.

10 years agoemotion gstreamer1: Set visualizations if the vis name is not NULL
Sebastian Dröge [Mon, 13 Jan 2014 09:02:34 +0000 (10:02 +0100)]
emotion gstreamer1: Set visualizations if the vis name is not NULL

and not the other way around.

10 years agoemotion gstreamer1: Also handle DATE_TIME tag
Sebastian Dröge [Sun, 12 Jan 2014 19:44:41 +0000 (20:44 +0100)]
emotion gstreamer1: Also handle DATE_TIME tag

10 years agoemotion gstreamer1: Fix wrong FIXME comment
Sebastian Dröge [Sun, 12 Jan 2014 19:42:49 +0000 (20:42 +0100)]
emotion gstreamer1: Fix wrong FIXME comment

10 years agoemotion gstreamer1: Don't forget to allocate metadata struct
Sebastian Dröge [Sun, 12 Jan 2014 19:42:01 +0000 (20:42 +0100)]
emotion gstreamer1: Don't forget to allocate metadata struct

10 years agoemotion gstreamer1: Implement seekability check via seeking query
Sebastian Dröge [Sun, 12 Jan 2014 19:36:14 +0000 (20:36 +0100)]
emotion gstreamer1: Implement seekability check via seeking query

10 years agoemotion gstreamer1: Add support for selecting subtitle channels
Sebastian Dröge [Sun, 12 Jan 2014 19:31:19 +0000 (20:31 +0100)]
emotion gstreamer1: Add support for selecting subtitle channels

10 years agoemotion gstreamer1: Volumes > 1.0 are valid too
Sebastian Dröge [Sun, 12 Jan 2014 19:29:32 +0000 (20:29 +0100)]
emotion gstreamer1: Volumes > 1.0 are valid too

10 years agoemotion gstreamer1: Add support for buffering and clock-lost handling
Sebastian Dröge [Sun, 12 Jan 2014 19:28:50 +0000 (20:28 +0100)]
emotion gstreamer1: Add support for buffering and clock-lost handling

10 years agoemotion gstreamer1: Major cleanup of internal and GStreamer state handling
Sebastian Dröge [Sun, 12 Jan 2014 19:25:01 +0000 (20:25 +0100)]
emotion gstreamer1: Major cleanup of internal and GStreamer state handling

10 years agoemotion gstreamer1: Don't set pipeline to PAUSED during seeking
Sebastian Dröge [Sun, 12 Jan 2014 18:33:57 +0000 (19:33 +0100)]
emotion gstreamer1: Don't set pipeline to PAUSED during seeking

This is not required and is handled transparently inside GStreamer
already.

10 years agoemotion gstreamer1: Use GST_PLAY_FLAG_VIS instead of the corresponding integer
Sebastian Dröge [Sun, 12 Jan 2014 18:26:08 +0000 (19:26 +0100)]
emotion gstreamer1: Use GST_PLAY_FLAG_VIS instead of the corresponding integer

10 years agoemotion gstreamer1: Don't use the NATIVE_AUDIO/VIDEO flags
Sebastian Dröge [Sun, 12 Jan 2014 18:21:57 +0000 (19:21 +0100)]
emotion gstreamer1: Don't use the NATIVE_AUDIO/VIDEO flags

There are too many video formats we don't handle, so we should let GStreamer
handle the conversion if necessary.

10 years agoemotion gstreamer1: Use refcounting instead of in/out variable counting
Sebastian Dröge [Sun, 12 Jan 2014 18:18:37 +0000 (19:18 +0100)]
emotion gstreamer1: Use refcounting instead of in/out variable counting

10 years agoemotion gstreamer1: Simplify pipeline creation
Sebastian Dröge [Sun, 12 Jan 2014 18:16:58 +0000 (19:16 +0100)]
emotion gstreamer1: Simplify pipeline creation

Also remove the webcam support, this could've never worked and
needs to be handled differently.

10 years agoemotion gstreamer1: Some code re-flow in the GStreamer bus handler
Sebastian Dröge [Sun, 12 Jan 2014 11:59:03 +0000 (12:59 +0100)]
emotion gstreamer1: Some code re-flow in the GStreamer bus handler

10 years agoemotion gstreamer1: Use GStreamer's path/URI handling functions instead of writing...
Sebastian Dröge [Sun, 12 Jan 2014 11:54:52 +0000 (12:54 +0100)]
emotion gstreamer1: Use GStreamer's path/URI handling functions instead of writing our own

10 years agoemotion gstreamer1: Rename functions and variables for clarity
Sebastian Dröge [Sun, 12 Jan 2014 11:03:44 +0000 (12:03 +0100)]
emotion gstreamer1: Rename functions and variables for clarity

Also remove some useless variables.

10 years agoUnbreak ecore_evas wayland engines:
Chris Michael [Mon, 13 Jan 2014 07:52:24 +0000 (07:52 +0000)]
Unbreak ecore_evas wayland engines:

Val variable is not declared here. Use the variable 'on' which was
passed in to this function.

Signed-off-by: Chris Michael <devilhorns@comcast.net>
10 years agoecore_evas - use eina_bool instead of the char for interal data.
ChunEon Park [Mon, 13 Jan 2014 05:26:36 +0000 (14:26 +0900)]
ecore_evas - use eina_bool instead of the char for interal data.

10 years agoevas - cserve2 - avoid segv accessing im->cache
Carsten Haitzler (Rasterman) [Sun, 12 Jan 2014 23:50:52 +0000 (08:50 +0900)]
evas - cserve2 - avoid segv accessing im->cache

10 years agoevas gl - fix egl/gles build after alpha4 change for fonts
Carsten Haitzler (Rasterman) [Sun, 12 Jan 2014 22:02:24 +0000 (07:02 +0900)]
evas gl - fix egl/gles build after alpha4 change for fonts

10 years agoevas - fonts - move to using 4bit and rel 4 bit compressed font glyphs
Carsten Haitzler (Rasterman) [Sun, 12 Jan 2014 20:13:00 +0000 (05:13 +0900)]
evas - fonts - move to using 4bit and rel 4 bit compressed font glyphs

this changes the internal encoding of font glyphs in evas to use 4bit
uncompressed if small, or 4bit rle (run length encoded) if larger.
this caves at least 50% of memory on fonts - and more if bigger. with
large fonts (40-80pixel size) we can save in the region of 80% of
memory used for glyphs. this also happesn to allow speedups in
rendering too.

10 years agoevas gl - prefer alpha4 format for font textures if possible (save memory)
Carsten Haitzler (Rasterman) [Sun, 12 Jan 2014 20:09:29 +0000 (05:09 +0900)]
evas gl - prefer alpha4 format for font textures if possible (save memory)

if alpha4 is possible (desktopgl) then use it for fonts as this should
cut memory in half for them and possibly speed things up due to less
memory bandwidth needed

10 years agoecore_evas_x: Call _ecore_evas_x_sync_clear() after
Daniel Juyung Seo [Sun, 12 Jan 2014 08:22:40 +0000 (17:22 +0900)]
ecore_evas_x: Call _ecore_evas_x_sync_clear() after
_ecore_evas_x_gl_window_new() to ensure eglDestroySurface is called
before XDestroyWindow call.

In case of gl engine, eglDestroySurface should be called before
XDestroyWindow.
Special thanks to Changwoo Kim <ce2010.kim@samsung.com> for the idea.

10 years agoecore_evas_x: Fix gl_sync_draw_done initialize routine to handle
Daniel Juyung Seo [Sun, 12 Jan 2014 08:01:02 +0000 (17:01 +0900)]
ecore_evas_x: Fix gl_sync_draw_done initialize routine to handle
exceptional case.

As ECORE_EVAS_GL_SYNC_DRAW_DONE is an environment variable, that could
have unexpected value like 2 or 3. So set gl_sync_draw_done to 1 only
when the env variable is 1.

10 years agoecore_evas_x: Refactor/flourish gl_dync_draw_done member from Ecore_Evas
Daniel Juyung Seo [Sun, 12 Jan 2014 07:37:30 +0000 (16:37 +0900)]
ecore_evas_x: Refactor/flourish gl_dync_draw_done member from Ecore_Evas
structure.

- moved gl_sync_draw_done for a better structure packing.
- changed gl_sync_draw_done type from int to char which is enough.
- added more description to gl_sync_draw_done

10 years agoeina: fix memory leak of inserting the wrong value in the hash.
Cedric Bail [Sat, 11 Jan 2014 12:36:47 +0000 (21:36 +0900)]
eina: fix memory leak of inserting the wrong value in the hash.

We where inserting the pointer data instead of the pointer, leading to
unaligned access on Sparc (Thanks Lutin to report it and Debian tools/infra
to help us catch it) and also a memory leak.

10 years agoecore - clean up code.
ChunEon Park [Sat, 11 Jan 2014 09:19:43 +0000 (18:19 +0900)]
ecore - clean up code.

fix indentation and use Eina_Bool instead of char.

10 years agoecore_evas - fixed indent
ChunEon Park [Sat, 11 Jan 2014 09:01:07 +0000 (18:01 +0900)]
ecore_evas - fixed indent

10 years ago[eina] add test case - eina_list_clone, eina_list_reverse_clone
Shinwoo Kim [Sat, 11 Jan 2014 06:46:52 +0000 (22:46 -0800)]
[eina] add test case - eina_list_clone, eina_list_reverse_clone

10 years agoevas gl example - make bg blue to make onefang not freak out. :)
Carsten Haitzler (Rasterman) [Sat, 11 Jan 2014 02:12:16 +0000 (11:12 +0900)]
evas gl example - make bg blue to make onefang not freak out. :)

10 years agoevas cserv2 - remove logical dead code
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 11:16:09 +0000 (20:16 +0900)]
evas cserv2 - remove logical dead code

fixes CID 1113441

10 years agoformatting - fix
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 10:58:21 +0000 (19:58 +0900)]
formatting - fix

10 years agoedje - formatting - indenting fix
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 10:54:01 +0000 (19:54 +0900)]
edje - formatting - indenting fix

10 years agocomment out unreachable code
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 10:04:32 +0000 (19:04 +0900)]
comment out unreachable code

fixes CID 1132633

10 years agoevas gl - comment out unreachable code
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 09:59:49 +0000 (18:59 +0900)]
evas gl - comment out unreachable code

fixes CID 1132634

10 years agoevas - remove dead code (comment it out)
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 09:48:42 +0000 (18:48 +0900)]
evas - remove dead code (comment it out)

removes entirely dead code and keeps it for reference only

10 years agoedje_cc - remove dead code
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 09:45:01 +0000 (18:45 +0900)]
edje_cc - remove dead code

fixes CID 1132639

10 years agocserve2 - debug bin - fix theoretical leak of socket fd
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 09:40:52 +0000 (18:40 +0900)]
cserve2 - debug bin - fix theoretical leak of socket fd

fixes CID 1039286

10 years agoeina lock posix - remove dead code in eina_spinlock_take
Carsten Haitzler (Rasterman) [Fri, 10 Jan 2014 07:58:39 +0000 (16:58 +0900)]
eina lock posix - remove dead code in eina_spinlock_take

this solves CID 1106340

10 years agoemotion: remove some warning in the example.
Cedric BAIL [Fri, 10 Jan 2014 07:33:22 +0000 (16:33 +0900)]
emotion: remove some warning in the example.

10 years agoexamples: fix various warnings in ecore and eina examples.
Vincent Torri [Thu, 9 Jan 2014 08:24:04 +0000 (09:24 +0100)]
examples: fix various warnings in ecore and eina examples.

Signed-off-by: Cedric BAIL <cedric.bail@samsung.com>