platform/upstream/efl.git
10 years agofix ecore-x shadow warning
discomfitor [Thu, 12 Dec 2013 14:31:08 +0000 (09:31 -0500)]
fix ecore-x shadow warning

10 years agorebase dns.c from upstream
discomfitor [Thu, 12 Dec 2013 14:29:02 +0000 (09:29 -0500)]
rebase dns.c from upstream

10 years agoevas - fix gif loader drop of image data sometimes (image dump/flush)
Carsten Haitzler (Rasterman) [Thu, 12 Dec 2013 09:42:13 +0000 (18:42 +0900)]
evas - fix gif loader drop of image data sometimes (image dump/flush)

stable release - cherry-pick me!

this fixses some still gifs going missing.

10 years agopkgbuild: follow aur efl-git version naming.
Cedric BAIL [Thu, 12 Dec 2013 02:04:43 +0000 (11:04 +0900)]
pkgbuild: follow aur efl-git version naming.

10 years agopkgbuild: let's tests all backend in our package.
Cedric BAIL [Thu, 12 Dec 2013 01:28:13 +0000 (10:28 +0900)]
pkgbuild: let's tests all backend in our package.

10 years agoautotools: detect systemd to warn user they are required to use --enable-systemd...
Cedric BAIL [Thu, 12 Dec 2013 01:27:01 +0000 (10:27 +0900)]
autotools: detect systemd to warn user they are required to use --enable-systemd to turn it on.

10 years agoembryo - embryo_cc - argv parsing MIGHT leave str buf not nul terminated
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 13:50:34 +0000 (22:50 +0900)]
embryo - embryo_cc - argv parsing MIGHT leave str buf not nul terminated

should fix CID 1039726

10 years agoevas - evas gl - fix realloc failure case to handle it properly
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 13:44:42 +0000 (22:44 +0900)]
evas - evas gl - fix realloc failure case to handle it properly

patched shader string realloc failure not handled right - could leak
patched string and p.

10 years agoevas - evas_gl - fix shader patch to not free static strings
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 13:40:15 +0000 (22:40 +0900)]
evas - evas_gl - fix shader patch to not free static strings

parsing problem with opengl_strtok() which would free the previous
token "p", but in some cases it would be a const string. this should
fix CID 1039653

10 years agoeet data encode - dont free data if its not separately allocated
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 11:44:51 +0000 (20:44 +0900)]
eet data encode - dont free data if its not separately allocated

in one case data is not separately allocated but is part of the
Eet_Variant_Unknow struct where it is allocated as extra space on the
end of the data blob. in this case don't free it, otherwise do (pass
in true) as before. this should fix CID 1039728

10 years agoecore - ecore systemd module - unused var warning - remove unused var
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 11:21:32 +0000 (20:21 +0900)]
ecore - ecore systemd module - unused var warning - remove unused var

10 years agoemotion - generic module - re-order code to make coverty happy
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:54:15 +0000 (19:54 +0900)]
emotion - generic module - re-order code to make coverty happy

this fixes CID 1039884 which isn't a real problem as the callback del
never dereferences the data pointer - just uses it as a value, but
this is really to ensure that it doesn't come back if the code were to
change.

10 years agoevas - font dir - remove from hash before free/del
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:49:36 +0000 (19:49 +0900)]
evas - font dir - remove from hash before free/del

this addresses CID 1039889 but was otherwise harmless

10 years agoethumb - client lib - fix access of async item in idler after free
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:42:07 +0000 (19:42 +0900)]
ethumb - client lib - fix access of async item in idler after free

this fixes CID 1039892

10 years agoephysics - fix real access-after-free issue on delete of constraint
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:38:46 +0000 (19:38 +0900)]
ephysics - fix real access-after-free issue on delete of constraint

this fixes CID 1039895

10 years agoecore-con - dns.c - fix double free
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:32:54 +0000 (19:32 +0900)]
ecore-con - dns.c - fix double free

this fixes CID 1039909

10 years agoedje - edje_pick - fix coverty access-after-unref issue
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:24:42 +0000 (19:24 +0900)]
edje - edje_pick - fix coverty access-after-unref issue

fixes CID 1039916

10 years agoedje - edje_cc - fix coverty issue of not exiting loop for box items name
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:13:16 +0000 (19:13 +0900)]
edje - edje_cc - fix coverty issue of not exiting loop for box items name

this shoul fix CID 1039918

10 years agoevas - textblock - fix dbg_info_get to avoid leak if root is NULL
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 10:00:15 +0000 (19:00 +0900)]
evas - textblock - fix dbg_info_get to avoid leak if root is NULL

fixes CID 1040007

10 years agoevas - cserve2 - fix uninitialized member "image_id".
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 09:47:39 +0000 (18:47 +0900)]
evas - cserve2 - fix uninitialized member "image_id".

fixes CID 1042303.

10 years agoevas - gl - ifdef out unused var with non-gles build
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 09:26:45 +0000 (18:26 +0900)]
evas - gl - ifdef out unused var with non-gles build

10 years agoevas - gl - use snprintf for safety (never likely anyway, so code was fine)
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 09:25:29 +0000 (18:25 +0900)]
evas - gl - use snprintf for safety (never likely anyway, so code was fine)

10 years agoemotion - test - exit if calloc fails
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 09:22:59 +0000 (18:22 +0900)]
emotion - test - exit if calloc fails

spotted this while looking at coverty issue. likely analysis tools
will complain.

10 years agoedje - edje_cc - try and make coverty ignore the leak (invalid)
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 09:20:13 +0000 (18:20 +0900)]
edje - edje_cc - try and make coverty ignore the leak (invalid)

hope that this does not reappear in future if changes happen in src.
CID 1132627

10 years agoevas gif loader - address animated always being true
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 09:10:15 +0000 (18:10 +0900)]
evas gif loader - address animated always being true

this addresses CID 1135790 and removes useless checks.

10 years agopkgbuild: we do conflict with efl-git from AUR.
Cedric BAIL [Wed, 11 Dec 2013 04:56:08 +0000 (13:56 +0900)]
pkgbuild: we do conflict with efl-git from AUR.

10 years agoautotools: only turn on auto testing on linux.
Cedric BAIL [Wed, 11 Dec 2013 04:46:33 +0000 (13:46 +0900)]
autotools: only turn on auto testing on linux.

10 years agoautotools: fix complain about using subdirs without the right option.
Cedric BAIL [Wed, 11 Dec 2013 04:46:01 +0000 (13:46 +0900)]
autotools: fix complain about using subdirs without the right option.

10 years agoautotools: make systemd dependency detection automatic.
Cedric BAIL [Wed, 11 Dec 2013 02:39:09 +0000 (11:39 +0900)]
autotools: make systemd dependency detection automatic.

10 years agoautotools: let's try this variant for detecting c++ compiler availability.
Cedric BAIL [Wed, 11 Dec 2013 01:40:13 +0000 (10:40 +0900)]
autotools: let's try this variant for detecting c++ compiler availability.

10 years agogif loader - handle theoretical NULL loader_data case
Carsten Haitzler (Rasterman) [Wed, 11 Dec 2013 00:23:09 +0000 (09:23 +0900)]
gif loader - handle theoretical NULL loader_data case

10 years agoEvas tests: Add invalid markup test
Jean-Philippe Andre [Tue, 10 Dec 2013 11:42:58 +0000 (20:42 +0900)]
Evas tests: Add invalid markup test

Commit 0cb048f9b681d43fe9 fixed a crash in Textblock where
a couple "key=val" had an empty value ("key=").

10 years agoecore_x: ensure positive malloc size
Sebastian Dransfeld [Tue, 10 Dec 2013 07:50:18 +0000 (08:50 +0100)]
ecore_x: ensure positive malloc size

size_ret is used later as an argument for malloc, so it should be
positive. In addition this should ensure that
ecore_x_window_porp_property_get returns a positive value and is true if
we malloc data.

Hopefully also fixes CID 1135636

10 years agoevas - fix gif loader cpu overuse - used too much cpu to decode anims
Carsten Haitzler (Rasterman) [Tue, 10 Dec 2013 07:27:19 +0000 (16:27 +0900)]
evas - fix gif loader cpu overuse - used too much cpu to decode anims

stable release - cherry-pick me!

the evas gif loader used way too much cpu to decode animated gifs
because in the rewrite that made it correct, it did not store the
current gif file handle and state, thus each frame it would have to
decode all frames before that one before finally decoding the final
one. that means to decode frame 200, it decoded frame 1, 2, 3, 4 etc.
all the way up to 199 THEN decoded 200 on top, so decode cost became
progressively more then further through the animation you were.

this fixes that by storing state and file handle and allowing you to
iterate through.

10 years agoecore_x: remove double free
Sebastian Dransfeld [Tue, 10 Dec 2013 07:28:21 +0000 (08:28 +0100)]
ecore_x: remove double free

We first free'd the data in the else, then at the end.

Now we free old_data just before return, and at the end.

Fixes CID 1135637

10 years agoecore-evas - disable comp syncing by default (can be turned on though still)
Carsten Haitzler (Rasterman) [Tue, 10 Dec 2013 05:02:27 +0000 (14:02 +0900)]
ecore-evas - disable comp syncing by default (can be turned on though still)

10 years agoEvas textblock: Fix crash with empty markups values
Jean-Philippe Andre [Tue, 10 Dec 2013 02:39:15 +0000 (11:39 +0900)]
Evas textblock: Fix crash with empty markups values

Markup parsing will segv if a value string is empty,
as in "<style=>". Sure, this is invalid, but hey, it could
definitely be used from an app or even by a user writing
his own markups :)

The internal doc says this function expects an item to be
of the form "key=val" but there are no checks beyond the
presence of "=" in the string before calling it.

10 years agoautotools: try to figure out what our buildbot get here.
Cedric Bail [Tue, 10 Dec 2013 02:30:43 +0000 (11:30 +0900)]
autotools: try to figure out what our buildbot get here.

10 years agoecore_imf/wayland: Only call hide_input_panel on im_context_hide().
Rafael Antognolli [Mon, 9 Dec 2013 16:21:29 +0000 (14:21 -0200)]
ecore_imf/wayland: Only call hide_input_panel on im_context_hide().

There's no need to call it on text_input_leave too, otherwise this would
be called twice, the one from text_input_leave possibly being called
after the focus was regain already by a text input, causing the bug
described in T237.

This fixes T237.

10 years agoecore_evas: added missing documentation for ecore_evas_manual_render_set/get().
Daniel Juyung Seo [Mon, 9 Dec 2013 11:17:05 +0000 (20:17 +0900)]
ecore_evas: added missing documentation for ecore_evas_manual_render_set/get().

10 years agoEvas textblock: Slightly improve paragraph items freeing.
Tom Hacohen [Mon, 9 Dec 2013 13:33:46 +0000 (13:33 +0000)]
Evas textblock: Slightly improve paragraph items freeing.

Use the EINA_LIST_FREE macro instead of a safe loop and a list free.
This is faster and cleaner.

10 years agoautotools: work around buggy macro to force C++ compiler detection.
Cedric Bail [Mon, 9 Dec 2013 12:35:56 +0000 (21:35 +0900)]
autotools: work around buggy macro to force C++ compiler detection.

10 years agoecore_evas/wayland: Update withdrawn property, and inform state_changed.
Rafael Antognolli [Wed, 4 Dec 2013 20:50:19 +0000 (18:50 -0200)]
ecore_evas/wayland: Update withdrawn property, and inform state_changed.

10 years agofix seb "fix" commit that causes lots of segvs.
Carsten Haitzler (Rasterman) [Mon, 9 Dec 2013 09:39:07 +0000 (18:39 +0900)]
fix seb "fix" commit that causes lots of segvs.

this fixes 04e0a6d95ec5610e35febbbb9b1be1a361f177ae which introduced
copy & paste segv's by properly terminating the right buffer.

10 years agoEvas/cserve2: Fix crash during shutdown
Jean-Philippe Andre [Mon, 9 Dec 2013 08:48:32 +0000 (17:48 +0900)]
Evas/cserve2: Fix crash during shutdown

If an image failed to load, and cserve2 returned an error message,
then the File_Entry was freed, but not removed from the hash.
Solution: remove entry from the hash, let the callback free the data.

10 years agoEvas: Fix crash in GL engine(s) during preload cancel
Jean-Philippe Andre [Mon, 9 Dec 2013 08:28:02 +0000 (17:28 +0900)]
Evas: Fix crash in GL engine(s) during preload cancel

In my config, running terminology with the GL engine and under
cserve2, some image could not be loaded. The tex argument
in evas_gl_preload_target_[un]register was NULL, leading to
an immediate crash.

10 years agoFix opaque windows becoming transparent after hide() then show().
Chris Michael [Mon, 9 Dec 2013 07:32:45 +0000 (07:32 +0000)]
Fix opaque windows becoming transparent after hide() then show().

This fixes Phab #T611. Previously, windows which were opaque would
become transparent after being hidden and then shown. We fix that by a
call to ecore_wl_window_alpha_set when the window gets shown. This
patch also brings the wayland_egl hide code more inline with the shm
hide code by testing if the surface does not match the existing one.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
10 years agoEcore: add DnD abort from source
Daniel Zaoui [Sun, 1 Dec 2013 12:48:21 +0000 (14:48 +0200)]
Ecore: add DnD abort from source

This can be used to cancel a DnD drag operation in the middle.

10 years agoecore evas - add animator update syncing to only render on animator ticks
Carsten Haitzler (Rasterman) [Sun, 8 Dec 2013 06:07:40 +0000 (15:07 +0900)]
ecore evas - add animator update syncing to only render on animator ticks

this adds a ifdefable feature to sync rendering only to animator
slots. this should reduce over-render of more frames than a user can
see when updates are triggered by things like mouse movements (which
may come in many times faster than the framerate). this is an
experiment to see if this helps smoothness and load. it also has
problems in e grabs x while rendering - this is now fixed in e18
alreadey, but it is just a config you can turn off.

10 years agoecore loop + animator - add call to get if an animator cb has run this iter
Carsten Haitzler (Rasterman) [Sun, 8 Dec 2013 06:05:51 +0000 (15:05 +0900)]
ecore loop + animator - add call to get if an animator cb has run this iter

this adds a simple call and infra to get if an animator has run this
iteration. it's simple and not really useful other than internally to
efl.

10 years agobuild: fix installing in DESTDIR for packaging and buildbot.
Cedric Bail [Mon, 9 Dec 2013 01:57:12 +0000 (10:57 +0900)]
build: fix installing in DESTDIR for packaging and buildbot.

This should fix T628 .

10 years agoefl: remove uuid from all libtool files.
Cedric Bail [Sun, 8 Dec 2013 02:50:54 +0000 (11:50 +0900)]
efl: remove uuid from all libtool files.

This change is needed for windows as this library is statically linked into
evil, but libtool doesn't get it.

10 years agoecore_x: Fix buffer overrun
Sebastian Dransfeld [Sat, 7 Dec 2013 17:32:17 +0000 (18:32 +0100)]
ecore_x: Fix buffer overrun

Use strncpy to prevent buffer overrun on buf, and '\0' terminate.

Fixes CID 1039584

10 years agoecore_x: Fix bug in ecore_x_region_invert
Sebastian Dransfeld [Sat, 7 Dec 2013 17:25:22 +0000 (18:25 +0100)]
ecore_x: Fix bug in ecore_x_region_invert

As bounds is a pointer, bounds + num will always be postive, so this is
an eternal loop. In addition, XFixesInvertRegion only accepts one
region, so num must be 1.

Fixes CID 1039469

10 years agoecore_x: Remove NULL check
Sebastian Dransfeld [Sat, 7 Dec 2013 17:02:21 +0000 (18:02 +0100)]
ecore_x: Remove NULL check

to get here, old_data != NULL.

Fixes CID 1039648 and 1039649

10 years agoecore_x: Use free, not XFree
Sebastian Dransfeld [Sat, 7 Dec 2013 17:00:27 +0000 (18:00 +0100)]
ecore_x: Use free, not XFree

Data returned from ecore_x_window_prop_property_get are malloced, so use
free, not XFree. Which also is consistent, as free was used on other
occasions.

10 years agoecore_x: supported isn't an optional argument
Sebastian Dransfeld [Sat, 7 Dec 2013 16:56:55 +0000 (17:56 +0100)]
ecore_x: supported isn't an optional argument

if supported is NULL, this function isn't of any use.

Fixes CID 1039430

10 years agoecore_x: Only use att if call succeeds.
Sebastian Dransfeld [Sat, 7 Dec 2013 16:53:34 +0000 (17:53 +0100)]
ecore_x: Only use att if call succeeds.

XGetWindowAttributes isn't guaranteed to succeed.

Fixes CID 1039692

10 years agoecore_x: Only do memset if len < 20
Sebastian Dransfeld [Sat, 7 Dec 2013 16:50:15 +0000 (17:50 +0100)]
ecore_x: Only do memset if len < 20

If len is 20, n is 0, and nothing to memset.

Fixes CID 1039311

10 years agoecore_x: always free data and return value
Sebastian Dransfeld [Sat, 7 Dec 2013 16:46:29 +0000 (17:46 +0100)]
ecore_x: always free data and return value

if we are the selected parser, we are responsible for free'ing data and
returning a selection value.

10 years agoecore_x: Check if we have data
Sebastian Dransfeld [Sat, 7 Dec 2013 16:43:46 +0000 (17:43 +0100)]
ecore_x: Check if we have data

data could be NULL and size == 0, so check for this.

Fixes CID 1039264

10 years agoecore_x: '\0' terminate converted text
Sebastian Dransfeld [Sat, 7 Dec 2013 16:14:11 +0000 (17:14 +0100)]
ecore_x: '\0' terminate converted text

This makes ecore_x text converter behave like elm text converter.

Fixes CID 1039310

10 years agoecore_x: Use eina_inlist
Sebastian Dransfeld [Sat, 7 Dec 2013 15:31:55 +0000 (16:31 +0100)]
ecore_x: Use eina_inlist

ecore_x_selection implemented it's own single linked list. Use
eina_inlist instead.

10 years agoecore_x: Only send XdndEnter when we have types
Sebastian Dransfeld [Sat, 7 Dec 2013 14:49:45 +0000 (15:49 +0100)]
ecore_x: Only send XdndEnter when we have types

If ecore_x_window_prop_property_get fails, we have no types to drop so
don't send XdndEnter.

Fixes CID 1039693

10 years agoecore_x: Also check for !prop here
Sebastian Dransfeld [Sat, 7 Dec 2013 14:44:04 +0000 (15:44 +0100)]
ecore_x: Also check for !prop here

We check if prop is NULL after first property check, do it here too.

Fixes CID 1039429

10 years agoeldbus: Check return value of eina_value_pget
Sebastian Dransfeld [Sat, 7 Dec 2013 14:38:24 +0000 (15:38 +0100)]
eldbus: Check return value of eina_value_pget

If eina_value_pget returns FALSE, we shouldn't continue.

Fixes CID 1039685

10 years agoeldbus: use alloca to allocate local mem
Sebastian Dransfeld [Sat, 7 Dec 2013 14:34:55 +0000 (15:34 +0100)]
eldbus: use alloca to allocate local mem

Since we free the allocated mem some lines down, we can use alloca. This
also fixes CID 1039179

10 years agoeldbus: unref msg on error
Sebastian Dransfeld [Sat, 7 Dec 2013 14:31:46 +0000 (15:31 +0100)]
eldbus: unref msg on error

If we return here, the msg isn't sent, so unref it.

Fixes CID 1039178

10 years agoeldbus: on error unref msg
Sebastian Dransfeld [Sat, 7 Dec 2013 14:30:52 +0000 (15:30 +0100)]
eldbus: on error unref msg

This is done on all other errors, EINA_SAFETY macros ain't no silver
bullet.

Fixes CID 1039177

10 years agoeldbus: free allocated mem on error
Sebastian Dransfeld [Sat, 7 Dec 2013 14:30:09 +0000 (15:30 +0100)]
eldbus: free allocated mem on error

Fixes CID 1039177

10 years agoeldbus: print error when fwrite fails
Sebastian Dransfeld [Sat, 7 Dec 2013 13:17:33 +0000 (14:17 +0100)]
eldbus: print error when fwrite fails

Return value of fwrite should be checked.

Fixes 1039715

10 years agoeldbus: Check return value of _type_size
Sebastian Dransfeld [Sat, 7 Dec 2013 13:10:44 +0000 (14:10 +0100)]
eldbus: Check return value of _type_size

_type_size could return 0, which would fail in the modulo call next.
Check for 0, and return.

Fixes CID 1039436

10 years agoeldbus: create object when needed
Sebastian Dransfeld [Sat, 7 Dec 2013 13:07:29 +0000 (14:07 +0100)]
eldbus: create object when needed

We allocate the strbuf at the beginning of the function, but do several
return checks after this which does not free the strbuf before return,
and we don't use the strbuf.

So create the strbuf object after checks, just before we need it.

Fixes CID 1039287

10 years agoeina: Set it to NULL before next iteration
Sebastian Dransfeld [Sat, 7 Dec 2013 12:59:16 +0000 (13:59 +0100)]
eina: Set it to NULL before next iteration

it might have been free'd by the user, so set it to NULL before next
iteration. This is an attempt to fix CID 1039913 and 1039914.

We don't use the pointer value, only the pointer, so the error is wrong.
Could flag the error in coverity, but if this fixes it, we wont see the
error in other situations.

10 years agoRevert "evas/render - make sure render pre in proxy rendering."
ChunEon Park [Sat, 7 Dec 2013 18:14:14 +0000 (03:14 +0900)]
Revert "evas/render - make sure render pre in proxy rendering."

This reverts commit b113f555e0ea0b0261ad870fca35d427a415c20e.

found side effect... revert this code until fix them.

10 years agoevas/render - make sure render pre in proxy rendering.
ChunEon Park [Sat, 7 Dec 2013 17:40:35 +0000 (02:40 +0900)]
evas/render - make sure render pre in proxy rendering.

if the some of children are the mapped object in source object tree as well as the the mappped object is invisible,
then they wont be render_pre() called.

this make sure those render_pre() in proxy rendering.

10 years agoupdate .mailmap
Boris Faure [Sat, 7 Dec 2013 16:25:27 +0000 (17:25 +0100)]
update .mailmap

10 years agoupdate .mailmap
Boris Faure [Sat, 7 Dec 2013 16:11:30 +0000 (17:11 +0100)]
update .mailmap

10 years agoconfigure: Switch back to dev profile in master branch.
Stefan Schmidt [Fri, 6 Dec 2013 15:05:45 +0000 (16:05 +0100)]
configure: Switch back to dev profile in master branch.

This also enables v_rev again. Thanks to Doug for pointing it out.

10 years agoevas - silence coverty (false positive leak)
Carsten Haitzler (Rasterman) [Fri, 6 Dec 2013 00:07:44 +0000 (09:07 +0900)]
evas - silence coverty (false positive leak)

the hash can't leak where coverty thinks... because the hash find can
never work ... because the hash is empty.

10 years agodoc build - finally fix images.mk to stop changing order every make
Carsten Haitzler (Rasterman) [Fri, 6 Dec 2013 00:02:18 +0000 (09:02 +0900)]
doc build - finally fix images.mk to stop changing order every make

sort changes output based on locale. even between C and en_US etc.
letalone all the other interesting ones. this causes the sorted order
to keep changing of the images. this forces locale to C to make it
always the same order.

10 years agoevas - Clip mark performance improvement
Vostokov Sergey [Thu, 5 Dec 2013 12:05:44 +0000 (21:05 +0900)]
evas - Clip mark performance improvement

stable release - cherry-pick me!

We propose a patch that reduces graph traversal work in
evas_object_child_map_across_mark(). It fixes a few particular
slowdowns around Tizen applications, including 0.6 seconds slowdown.

evas_object_child_map_across_mark() does not seem to need to
recursively call itself on the same object many times. Yet we have
noticed that in some scenarios it repeatedly traverses the same
subtrees of objects over and over again, whenever there is more than
one way of reaching these subtrees. In the production issue mentioned
above, certain elm_object_part_content_set() call results in millions
of recursive calls of evas_object_child_map_across_mark(), taking
~0.6sec total.

We propose to allocate a hash table during top-level call to store all
objects visited, and return from sub-calls instantly whenever we are
called over an object we already visited.

10 years agoAdd code to deal with min, max, step, aspect, and base sizes.
Chris Michael [Thu, 5 Dec 2013 11:48:25 +0000 (11:48 +0000)]
Add code to deal with min, max, step, aspect, and base sizes.
Properly fix efl wayland elm window resize problem.

This adds support for min, max, step, aspect, and base size properties
when resizing a canvas under EFL Wayland.

This Also Properly fixes raster's report for EFL Wayland elm windows not
resizing properly. Previously, when resizing an elm window in wayland,
a portion of the window would draw outside the frame.

Signed-off-by: Chris Michael <cp.michael@samsung.com>
10 years agoSet engine aspect function pointer
Chris Michael [Thu, 5 Dec 2013 11:47:44 +0000 (11:47 +0000)]
Set engine aspect function pointer

Signed-off-by: Chris Michael <cp.michael@samsung.com>
10 years agoAdd common function prototype for setting ecore_evas aspect ratio
Chris Michael [Thu, 5 Dec 2013 11:47:12 +0000 (11:47 +0000)]
Add common function prototype for setting ecore_evas aspect ratio

Signed-off-by: Chris Michael <cp.michael@samsung.com>
10 years agoefreet: propagate scan error to caller
Sebastian Dransfeld [Thu, 5 Dec 2013 11:15:07 +0000 (12:15 +0100)]
efreet: propagate scan error to caller

Current code ignored return from recursive calls, so if cache_add failed
during recursion, the parent call would still succeed.

10 years agoevas - fix overdraw + too many rects problem found in some expedite tests
Carsten Haitzler (Rasterman) [Thu, 5 Dec 2013 11:02:39 +0000 (20:02 +0900)]
evas - fix overdraw + too many rects problem found in some expedite tests

stable release - cherry-pick me!

there way a problem with software rendering - it rendered some areas
more than once per frame due to overlapping rectangles. it also had
more rectangles to cover the same update area that it should have had.
this fixes this.

10 years agoefreet - fix recusrion checks need to pop as well as push.
Carsten Haitzler (Rasterman) [Thu, 5 Dec 2013 08:41:48 +0000 (17:41 +0900)]
efreet - fix recusrion checks need to pop as well as push.

stable release - cherry-pick me!

this fixes a lack of popping the efret recursion stack handler. this
wasn't done in 1.8 and is a bug.

10 years agoconfigure: Bump version to 1.8.99 to indicate development cycle.
Stefan Schmidt [Thu, 5 Dec 2013 10:14:03 +0000 (11:14 +0100)]
configure: Bump version to 1.8.99 to indicate development cycle.

1.8.0 is out and we are in the dev cycle for 1.9 so indicate this with
the version 1.8.99 for builds from git master.

10 years agoeina: this is actually a better way of improving Eina_Hash performance.
Cedric Bail [Thu, 5 Dec 2013 07:25:01 +0000 (16:25 +0900)]
eina: this is actually a better way of improving Eina_Hash performance.

This reduce the size of the rbtree part dedicated to the hash key match,
but reuse only bit that weren't matched by the bucket.

10 years agoximmodule: fix issue the cursor of preedit shows the in front of preedit string
Jihoon Kim [Thu, 5 Dec 2013 07:06:39 +0000 (16:06 +0900)]
ximmodule: fix issue the cursor of preedit shows the in front of preedit string

Before fixing this issue, the cursor of preedit appears the in front of preedit string.
The cursor of preedit string will appear in the proper position.

10 years agoximimmodule: remove useless code in preedit_string_get
Jihoon Kim [Thu, 5 Dec 2013 07:00:39 +0000 (16:00 +0900)]
ximimmodule: remove useless code in preedit_string_get

'cursor_pos' variable will be set in the below code, so 'cursor_pos = 0' is useless.

10 years agoFix raster bug report about elm window resize issue (content drawing
Chris Michael [Wed, 4 Dec 2013 11:30:36 +0000 (11:30 +0000)]
Fix raster bug report about elm window resize issue (content drawing
outside the window).

Be sure that the EEs requested geometry gets updated in our
common_resize function After min/max have been taken into account.

Elm is using ecore_evas_request_geometry_get in it's resize_job code
(Why...I have no clue lol. Does not seem like a good thing to check).

Signed-off-by: Chris Michael <cp.michael@samsung.com>
10 years agoUpdate the engine resize edge before we call ecore_wl_window_resize
Chris Michael [Wed, 4 Dec 2013 11:29:32 +0000 (11:29 +0000)]
Update the engine resize edge before we call ecore_wl_window_resize

Ecore_wl_window_resize essentially tells the shell to start the resize
process. We should have the evas engine info's resize_edge updated
Before we start that process so that Evas knows where the resize is
occuring from

Signed-off-by: Chris Michael <cp.michael@samsung.com>
10 years agoefreet - protect efreetd from recursing too far to save overhead and mem
Carsten Haitzler (Rasterman) [Wed, 4 Dec 2013 11:22:26 +0000 (20:22 +0900)]
efreet - protect efreetd from recursing too far to save overhead and mem

an errant path made its way into my efreet cache. this had a
side-effect of causing efreetd to scan my entire $HOME recursively
to monitor everything. while the original cause was $HOME getting in,
we shouldn't have efreetd sit and consume scan all of $HOME when this
is far from a normal situation. the recursive scanning is there ot
handle some minimal levels of subdirs in app directories, but not an
entire filesystem, so this mitigates the effects of errant cache data
by limiting the amount of recursion allows for icon dirs and desktop
dirs to 8 and 3 levels respectively.

10 years agoyet another useless images.mk change...
Carsten Haitzler (Rasterman) [Wed, 4 Dec 2013 11:20:20 +0000 (20:20 +0900)]
yet another useless images.mk change...

10 years agoecore ecore_glib.c: fixed documentation about glib integration always configure option.
Daniel Juyung Seo [Wed, 4 Dec 2013 10:43:26 +0000 (19:43 +0900)]
ecore ecore_glib.c: fixed documentation about glib integration always configure option.

--enable-glib-integration-always was changed to --with-glib=always in efl 1.8.

10 years agoeina: improve cmp function of eina_hash string related hash.
Cedric Bail [Wed, 4 Dec 2013 10:05:03 +0000 (19:05 +0900)]
eina: improve cmp function of eina_hash string related hash.

For strings, we already know the length, so let's just not call the
heavy strcmp if the length is not already the same. Get some easy
speedup here.

10 years agoeina: fuzze test the eina_hash_string_superfast to.
Cedric Bail [Wed, 4 Dec 2013 10:04:24 +0000 (19:04 +0900)]
eina: fuzze test the eina_hash_string_superfast to.

10 years agoeina: ignore benchmark binary.
Cedric Bail [Wed, 4 Dec 2013 09:32:29 +0000 (18:32 +0900)]
eina: ignore benchmark binary.