after i noticed some jitters, it seesms one thread to listen for vsync
events then to wake up the mainloop can suffer from irregular
scheduling jitters. it seems to be highly depenedent on both gpu
driver and cpu but it seemed the vsync thread itself was more reliably
woken than the mainlooop it then signalled, so merging this back is
just batter. it's configurable via an environment variable so we can
try either right now and see, but default is now to be in main loop.
if try_no_queue is used - its a thread that clearly wantt to be out of
the queue to run on its own so probably wants to wake up accurately
and thus not be a backgroun rpriority task but and urgent one. ensure
we set up priority accordingly as we didn't before. this should fix
scheduling when under load for vsync
@fix
Summary:
evas_vg_container should return NULL when the parent is NULL
because the function does not allot NULL parent
@fix
Reviewers: kimcinoo, jsuya, Hermet
Reviewed By: kimcinoo
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11867
This is necessary because the LuaJIT FFI implements the null
pointer to nil equality behavior, which is not possible to
do in standard Lua (as equality metamethod needs identical types
to trigger).
So replace it with ffi.nullptr as cffi-lua implements; in luajit
ffi.nullptr will just become nil and it'll work like before.
Summary:
efl_wref_del makes pointer to NULL.
So we should copy pointer before efl_wref_del is called.
Reviewers: SanghyeonLee, cedric
Reviewed By: SanghyeonLee
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11866
As we run some intermediate binary like edje_cc during efl build , and due to
those binary has some memory leaks the address sanitizer build of efl fails.
This Patch runs those binary with 'ASAN_OPTIONS=detect_leak=false' to disable the
memory leak check for those binaries.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11137
On Windows 64bits, long is a 4 bytes type. Declare count as being a intptr_t instead of long
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11836
On Windows 64bits, long is a 4 bytes type. Declare i as being a uintptr_t instead of unsigned long
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Felipe Magno de Almeida <felipe@expertisesolutions.com.br>
Differential Revision: https://phab.enlightenment.org/D11837
Summary:
evas_render_updates_internal > eina_array_remove is calling pending_change
pending_change > render_post could call evas_object_change >
evas_render_object_recalc.
The eina_array is removing its item from pending_objects, while
the evas_object_change that is calling evas_render_object_recalc is adding
item to pending_objects.
As a result, the pending_objects.count is incorrect, and it breaks render.
Reviewers: Hermet, jsuya, herb
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11835
Summary:
current cubic bezier function isn't accurate at sometime.
to make it more accurate, this patch rework bezier curve
by using a cardano's algorithm.
(refer to https://pomax.github.io/bezierinfo/)
Reviewers: Hermet, bu5hm4n
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11819
Summary: .9.png is not 9 patch file, but a png file which name is .9
Test Plan: evas_object_image_file_set(image, "./.9.png", NULL);
Reviewers: Hermet, jsuya, herb
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11848
Summary: EAPI must be defined to dllexport when building DLL, and to dllimport when using these DLL. To achieve this, define EFL_BUILD for each library and module, and set DLL_EXPORT unconditionally. Static library are and will be not supported
Test Plan: compilation
Reviewers: zmike, raster, jptiz
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11834
so i've moved all systemd and elogind support to be runtime only with
dlopen (eina_module) of libsystemd.so.0 (or libelogind.so.0 for elput)
and finding of symbols manually at runtime (if the right code paths or
env vars are set), thus remvoing the need to decide at compile time if
efl needs systemd support or not as it no longer needs systemd
headers/libs at compile time and just at runtime. this simplifies
building a bit and makes efl more adaptive to the final target system
at runtime.
Summary:
Some applications have used "clicked" smart callback from elm_image which
does not have an image in it.
To keep backward compatibility, hit_rect needs to be resized properly to
call smart callback function.
@fix
Reviewers: Hermet, jsuya, kimcinoo
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11840
Summary:
evas_vg_shaep_add should returned NULL when the parent is NULL
because the function does not allow the NULL parent.
@fix
Reviewers: jsuya, Hermet
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11839
Summary:
this was duplicated for a number of widgets in slightly different ways,
so it can be unified as a private layout function to reduce maintenance
costs
Reviewers: Hermet, YOhoho, Jaehyun_Cho, jsuya
Reviewed By: jsuya
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11818
Summary:
Users can use multiple jsons per description in one part in edc.
At that time, multiple files are loaded,
if a file is already loaded, it must be unloaded.
Test Plan:
edje_cc -beta -id ./ json_edc.edc
make
./test
Push Change button -> shutdown application
{F3882716}
Reviewers: Hermet, herb, kimcinoo
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11820
Summary:
When call elm_check_add, theme_apply of layout is called and
_efl_ui_check_legacy_efl_ui_widget_theme_apply is called.
Then it calls the signal emit for the icon.
this call is unnecessary.
Test Plan: N/A
Reviewers: akanad, Jaehyun_Cho, Hermet, YOhoho, zmike
Reviewed By: zmike
Subscribers: zmike, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11812
Summary:
When call elm_progressbar_add, theme_apply of layout is called and
_efl_ui_progressbar_legacy_efl_ui_widget_theme_apply is called.
Then it calls the signal emit for the icon.
this call is unnecessary.
Test Plan: N/A
Reviewers: Hermet, bowonryu, Jaehyun_Cho, YOhoho, zmike
Reviewed By: zmike
Subscribers: zmike, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11811
Summary:
void* is an invalid type for calculating pointer offsets, so ensure that
this is always cast to something else (e.g., char*) in the few cases
where it's been misused
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11816
Summary:
When call efl_canvas_object_animation_animation_pause_set,
pd->in is freed by calling efl_canvas_object_animation_stop in _animator_cb.
So when set pause_state, crash occurred
This patch add null check to prevent null access.
Test Plan: N/A
Reviewers: Hermet, kimcinoo, herb, bu5hm4n
Reviewed By: Hermet, bu5hm4n
Subscribers: bu5hm4n, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11815
Summary:
This patch reduces calculations for layouting textblock when it is not needed.
Exactly in **evas_object_textblock_render_pre**, layouting was done (if needed) regardless of object visibility.
evas_object_render_pre will async called if object status is changed, for example show->hide
In short words: **We do not layout textblock content if textblock t is hidden.**
```
// suppose textblock is shown
// and user want to hide it and set content in it
// to be visible later on if needed
evas_object_hide(textblock);
evas_object_textblock_text_markup_set(textblock, "Hello World");
//Layouting will be done on textblock regardless of its visiblity, becase render_pre
//will be called and will make it relayout
```
Reviewers: woohyun, zmike, tasn, raster, bu5hm4n
Reviewed By: zmike
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11508
Summary:
When calling style_user_pop in edje, current style was not freed by
style_user_pop in evas properly (since old_ts was not cared at all).
Because of this, memory was leaked by none accessible textblock styles which
should have been freed.
@fix
Test Plan:
1. call edje_object_part_text_style_user_pop
2. call edje_object_part_text_style_user_push with new style-text
3. repeat #1 and #2 several times and check the memory (ps -u)
Reviewers: ali.alzyod, bowonryu, bu5hm4n, zmike
Reviewed By: ali.alzyod
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11801
Summary:
The geometry info of evas_vg_node is based on Efl.Path_Bounds.
The size of vg_node does not change unless path data is changed.
Therefore, this API does not work normally.
Test Plan: N/A
Reviewers: Hermet, kimcinoo, herb
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11809
Summary:
There are wrong implemenation codes in the above and below get objects of efl_canvas_vg_node functin.
and made it working correctly.
Reviewers: Hermet, jsuya
Reviewed By: jsuya
Subscribers: singh.amitesh, cedric, #reviewers, #committers
Tags: #efl
Maniphest Tasks: T8700
Differential Revision: https://phab.enlightenment.org/D11781
Instead of forcing someone to look at this again in a couple of years we
can simply extend it to 100 years and expect this code does no longer
exist. :-)
For more details on the steps I used to generate and the problem in
ecore_con please see the original commit 3b835b57ff
Current mask is shared across the objects hierarchy,
children would try to apply this mask on the rendering
when it's enabled by the parent.
but if the map is enabled from one of parents, children should not apply
the parents mask but leaves it to the parent's job.
@fix
so if we call a callback that has e do something like set new pixel
data in a pixel data provider and we do it from the is pixel inside
call ebcause pixels were marked as dirty - pixels may no longer exist
after that, so re-get to make sure and handle properly.
this was found with tiling + e + shaped windows + moving them around
in tiling. it's a valid efl bug though.
@fix
Summary:
this should match all other similar cases when determining whether to
emit the event by checking whether at least one coord has changed, not both
Depends on D11791
Reviewers: devilhorns
Reviewed By: devilhorns
Subscribers: raster, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11792
Summary:
when defined, BUILD_EDJE_FP causes some of these struct members to not
be floating types and additionally defines special macros to be used
for comparisons to handle this case
Depends on D11786
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11787
- Put some methods in a group so they are accessible and can be linked to.
- Remove references to Efl_ methods, unavailable in Legacy.
- Incorrect param names and typos
Summary:
There was a mistake caused by "2a0eeba evas: fix png regression issue".
The flag is_for_data was close_file before this commit.
_evas_image_load_file_internal_head_png closes file if it is called only for
header. So the close_file does not mean is_for_data.
This patch is changing is_for_data to is_for_head, and make
_evas_image_load_file_internal_head_png use EINA_FILE_SEQUENTIAL
for data decoding.
Reviewers: Hermet, jsuya, herb, raster
Reviewed By: raster
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11783
Summary:
verify that ready is called (preload done) and error is not called
ref D11758
Reviewers: Hermet, jsuya, kimcinoo
Reviewed By: Hermet
Subscribers: stefan_schmidt, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11760
For our unit testing of ecore_con we have a sself-signed certificate for
localhost to do some of the tests. It expired in 2016 and this was not
really a problem so far. Starting from Debian 10 and Ubuntu 20.04 this
resulted in failures thought. Either the newer openssl version is more
strict or the config for Debian and Ubuntu is.
Anyway, time to generate a new one. Its valied until Sep 23 12:33:13
2024 GMT and I used the follwoing commands to generate the cert/key and
the second command to rmeove the key passphrase.
openssl req -x509 -newkey rsa:4096 -keyout server.key -out server.pem -days 1600 -subj '/CN=127.0.0.1'
openssl rsa -in server.key -out server2.key
Someone has to do this again in 4 years. if we stillhave this code
around.
Fixes CID 1409576 pointing that _node was unitialized.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Felipe Magno de Almeida <felipe@expertisesolutions.com.br>
Differential Revision: https://phab.enlightenment.org/D11770
Summary:
Accidentally commit "382c580 evas: add support for .9.png file to PNG loader."
adding the 9 patch feature with small code refactoring made use of setjmp
incorrectly.
[Problem]
evas_image_load_file_data_png calls _evas_image_load_file_internal_head_png,
and _evas_image_load_file_internal_head_png calls setjmp and returns without
problem. And png_read_row calls longjmp. This causes jumping into a function
which was exited. Problematic png file will be attached.
[Solution]
Save calling environment i.e. call setjmp, after returning from
_evas_image_load_file_internal_head_png.
Test Plan:
Problematic png file
{F3876983}
And example code.
{F3876986}
Reviewers: Hermet, jsuya, herb
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11782
Since we moved to floating calculation for msaa, this comparsion
is not working for direct image drawing though it's doable.
To resolve this, we can compare floatings regardless of msaa enable.
Summary: this patch will reduce amount of unnessesary calculations, if user set both font_size_array and size_range for content fit.
Reviewers: woohyun, bowonryu
Reviewed By: woohyun
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11498
Summary: Clear paragraph during markup set, instead of waiting until next text layout to happen, this will release unused memory right away
Reviewers: woohyun, zmike, bu5hm4n, raster
Subscribers: raster, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11773
This test seems to fail on Fedora32, we increase font size margin to
prevent it from happening
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11776
Summary:
Linking under Fedora32 toolchain got more picky. Ecore_drm_device.c
needs the symbol so make sure we have it there.
Reviewers: devilhorns, raster
Reviewed By: raster
Subscribers: raster, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11772
that is a unsigned int, if its 0 the timer is called as fast as
possible. Not doing that breaks exactness.
@fix
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11774
Personally I find it quite annoying to have these in every build.
Especially if they have not been attempted to get fixed even after we
started to stabilize some part of the unified API.
Reviewed-by: Felipe Magno de Almeida <felipe@expertisesolutions.com.br>
Differential Revision: https://phab.enlightenment.org/D11763
In our current Ecore_Drm2_Context structure, we are missing support
for using page_flip_handler2. This patch adds that ability. This could
prove useful when working with ecore_drm2 as it allows
page_flip_handler callbacks to setup a handler for page_flip_handler2,
which when the callback gets received, passes back the crtc_id which
the pageflip occured on.
@feature
This has been unused since the move to a preloaded lib and now makes
trouble durign compilation on Fedora32.
@fix
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11767
Summary:
the fileselector can store exactly 1 model and monitor that for changes,
which works normally just well. However, when we expand a items, another
model comes into play, which is not monitored, and no additional
reference is taken. Which means, _process_last will simply delete it,
and elements that did not come into play at the _populate call are never
displayed.
With this commit, we check that we do not have a parent item, and
explictly wait for the model to publish all its items, before we call
_process_last.
Additionally, that showed that the tracking of the list/grid item is
ultimatly buggy, as there is no assertion that the parent model is
really the parent directoy. Hence we need to hand pass the parent item
here, as a fallback option to set the correct parent item.
fixes T8689
Reviewers: zmike
Reviewed By: zmike
Subscribers: zmike, cedric, #reviewers, #committers
Tags: #efl
Maniphest Tasks: T8689
Differential Revision: https://phab.enlightenment.org/D11765
Ecore_Drm_Input had SOME docs, so a group for them is created.
Ecore_Drm_Sprite had NO DOCS at all, so the link is just removed.
This is a deprecated library anyway.
some tests to explicitly verify callback and internal behavior
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11752
useful when we want to verify that e.g., a certain object item is passed
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11751
sometimes a test might only want to verify the width or height of an object
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11750
just being extra paranoid here since there's some classes which implement
some layout/content interfaces without implementing signals
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11749
basic test to ensure that list sizing is homogeneous and works as it
"should", even though list sizing internals are a black hole of lost time
and dreams
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11748
putting raw evas and edje objects in here has historically been done,
so handle it without errors
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11747
something is happening during shutdown, when a fileselector was created
before. This is ending up in a strcmp to a NULL ptr, which crashes, with
the now, we ensure that we are not crashing anymore, when that is
happning.
TO solve this properly we should really fix the cause of a invalid model
beeing returned, however, i do not feel compftable doing that so short
before a release.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11762
when setting the relation from two widgets a<->b a is used as well as b.
This now correctly sets these flags.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11730
Each child_map_across_mark() takes 0.026ms with eina_hash
we can reduce the performing time by replacing the hash to array to 0.0041ms in a device.
Before Average: 0.026ms
Mostly it takes 23 times in the scrolling. thus it takes 3%(0.48ms) in one frame rendering (16ms)
After Average: 0.0041ms
Mostly it takes 23 times in the scrolling. thus it takes 0.6%(0.096ms) in one frame rendering (16ms)
result: map calc 5x faster than original (0.384ms improved)
Thanks Subhransu for report this.
@optimization
this makes no sense
../src/lib/elementary/elm_cnp.c:54:15: warning: type qualifiers ignored on function return type [-Wignored-qualifiers]
54 | static inline const Elm_Sel_Format
Reviewed-by: Christopher Michael <devilhorns@comcast.net>
Differential Revision: https://phab.enlightenment.org/D11759
"Evas_Group" does not exist, and "Evas" is a top-level group without content
(only subgroups). So I guess all these methods belong to the "Evas_Canvas" group.
This brings even more docs into existence, fixing multiple missing references.
All widgets copy&pasted the same description, including an invalid reference
to "Layout", which should be "Elm_Layout".
SOME of them had been fixed over the years but this commit fixes all of them.
This significantly reduces the number of Doxygen warnings and adds meaningful
links to the docs.
Docs did not match the intended enum value due to incorrect usage of the
doxygen tags.
Also, Evas_Render_Op was the only member of the top-level Evas group, so
I moved it inside one subgroup, where methods using it reside.
Summary:
In D11392, fixed to call file_unload when doing memfile_set.
Therefore, "load,error" callback was called while unloading object
that was not loaded when first load with memfile_set.
So fix it to avoid unnecessary unloading.
Test Plan:
{F3871804}
cp ./test_imge.c ./efl/
cd efl
gcc -o test_image test_image.c -g `pkg-config --cflags --libs elementary`
./test_image
Loading image...
Async file load failed.
Image is ready to show.
Reviewers: Hermet, kimcinoo, herb
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers, zmike
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11758
we were losing cached chaders and overwriting the cache all the time
when apps finished compiling new shaders and at other points... when
they should have already had a populated shader cache that was all
fine and happy. this fixes that so once the cache is fully populated
it wont write anymore and it wont lose existing shaders in the cache
@fix
If the given events list is NULL the data pointer would be as well. Make
sure we check for NULL here before accessing.
CID: 1419843
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11754
Instead of doing our own parsing here we should use ecore_file_dir_get()
which uses dirname() already to solve this.
CID: 1422196
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11727
These to code lines should be in one block and not one exectued without
the if.
CID: 1422198
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11726
it may report an exe is installed if it happens to be in cwd but cwd
isnt in path. the "full/relative" path handling case mishandled this.
fixes T8678
@fix
This enables a user to build EFL with wayland support enabled
on FreeBSD. It is NOT functioning, but everything starts at
some point.
This requires also linking against -lepoll-shim.
Meson arguments:
-Deeze=false -Dv4l2=false -Dfb=false -Ddrm=false -Dwl=true \
-Dsystemd=false
@fix T8659
this is sadly needed due to the fact that we have a .eo defnition in the
.eot file which contains the flags keyword.
Differential Revision: https://phab.enlightenment.org/D11739
there is no reason to have them in the .eot files. They are also not
namespaced in the way we want to have types.
Differential Revision: https://phab.enlightenment.org/D11738
Vincent Torri
Carsten Haitzler
Taehyub Kim
Daniel Kolesa
Hosang Kim
Subhransu Mohanty
Yeongjong Lee
Shinwoo Kim
Wonki Kim
Christopher Michael
WooHyun Jung
Mike Blumenkrantz
junsu choi
Hermet Park
Ali Alzyod
Stefan Schmidt
Xavi Artigas
João Paulo Taylor Ienczak Zanette
Marcel Hollerbach
Boris Faure
Alastair Poole