Previous vg didn't take care of cached buffers which were
allocated in it's lifetime because the cache buffers are managed
by its own cache buffer mgr, it has a limitation count of buffers also
buffers can be cleared when engine is shutdown.
This behavior is actually working properly but not well optimized
since it lost a chance to clear grown buffers.
Now vg do clear used buffers when object is invalidated.
for better precise buffer cache key,
We make a unique key name combining root node + size + frame index.
Now, we can reuse the root node for animation and caching buffers.
Container must have a set parent after construction.
efl_add_ref() with a parent won't work this case
because container needs some jobs in overriding parent_set() after proper intialization.
While we render vector drawing on ector,it uses ector buffers for subsequent compositions.
In this sequence, it switches ector buffer with a composition target,
while resuing the global ector context.
Previously, it trickly switched ector buffer, it broke the ector begin/end pair.
Now, it needs to recover global ector buffer once the composition finished.
Thus ector buffer need to return the current pixels information for this.
We need a more precise caching mechanism to save memory.
Current root node(key) is limited for caching buffers,
when to share by the multiple animated instances.
This patch is a step for further optmization work,
at the moment we disable the caching buffers for the animated instances.
Summary:
These APIs to get and set frames for a specific section for playing section.
Depends on D10505
Test Plan: N/A
Reviewers: Hermet, smohanty, kimcinoo
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D10506
Summary:
The current color is affected by the parent's opacity.
If p_opacity is set, it will be applied to the current color.
Test Plan: N/A
Reviewers: Hermet, smohanty, kimcinoo
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D10399
Summary:
Even if the same window, the same file, and the same size,
different images may be requested due to property changes caused by value_provider.
Test Plan: N/A
Reviewers: Hermet, smohanty, kimcinoo
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D10303
Summary:
If value_provider is applied to Efl.Ui.Animation_View, it is passed as vector object.
The passed value_provider list passes to Vg_File_Data.
If there is a value_Provider list in Vg_File_Data,
it apply the property to Lottie_Animation using lottie_animation_property_override api.
Depends on D9874
Test Plan:
...
Eo *vp = efl_add(EFL_GFX_VG_VALUE_PROVIDER_CLASS, p);
efl_gfx_vg_value_provider_keypath_set(vp, "**");
efl_gfx_vg_value_provider_fill_color_set(vp, 100, 0 ,0 ,255);
efl_ui_animation_view_value_provider_override(anim_view, vp);
vp = efl_add(EFL_GFX_VG_VALUE_PROVIDER_CLASS, p);
efl_gfx_vg_value_provider_keypath_set(vp, "**");
efl_gfx_vg_value_provider_stroke_width_set(vp, 50.0);
efl_gfx_vg_value_provider_stroke_color_set(vp, 0, 255, 0, 100);
efl_ui_animation_view_value_provider_override(anim_view, vp);
...
Reviewers: Hermet, kimcinoo, smohanty
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D9897
Summary:
It is not correct to change the parent and hope that refcount will work
for you. The parent chain is a decision made by the user, not something
a widget can touch. Only ref count are to be used when given an outside
object. As things where not triggering destruction, there was also no
point where the image buffer would be getting to zero and be freed. This
commit also fix this.
Reviewers: Hermet, zmike
Reviewed By: Hermet
Subscribers: #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D10169
Summary:
Vg_entry is shared by different child windows.
When two or more objects load the same file
When resizing, cache_vg_entry_render can delete an active entry
while creating a new entry and deleting an existing entry.
Therefore, use cache data added evas
Test Plan:
[enable json loader]
--- a/meson_options.txt
+++ b/meson_options.txt
@@ -203,7 +203,7 @@ option('evas-loaders-disabler',
type : 'array',
description : 'List of modular image/vector load .....
choices : ['gst', 'pdf', 'ps', 'raw', 'svg', 'rs .....
- value : ['webp', 'json']
+ value : ['webp']
[Test]
elementary_test -> Animation View click -> play -> Animation View click again -> ...
Reviewers: Hermet, smohanty, kimcinoo
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D9740
Current vector drawing requires several methods for compositing,
it's shouldnt have only masking but other blending functions
such as matte in lottie, porter&duff in android, etc.
Previously we didn't specify this method name,
just reserved the options and now we started to define the methods
with ector_renderer_composite_method() with Efl.Gfx.VG_Composite_Method
instead of mask_set().
Summary:
Prerequisite: Partial rendering ON + Evas VG content set in elm_layout + color/alpha animation of layout
It seems that expected clip rects were not merged when color change was taking place on vg object.
@fix
Summary:
Enum and object decleared in efl_canvas_vg_object_eo.legacy.*
use Efl_Canvas_Vg_* nameing rule.
This skips the definition of type declared in Efl.Canvas.Vg.Object class.
So change these to legacy names.
Test Plan: N/A
Reviewers: Hermet, smohanty, kimcinoo
Reviewed By: Hermet
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D9718
Current vector drawing requires several methods for compositing,
it's shouldnt have only masking but other blending functions
such as matte in lottie, porter&duff in android, etc.
Previously we didn't specify this method name,
just reserved the options and now we started to define the methods
with ector_renderer_composite_method() instead of mask_set().
Summary:
This patch extends efl_canvas_vg_object class to implement efl_gfx_frame_controller
to suppor any playable animation on it.
Plus, vector object takes care of lottie animation by using json loader.
it's caching mechanism is changed to cache only static frame, not all frames.
vg_cache supports json loader and make it animation request properly.
This feature has been stabilized enough, it's using in Samsung Galaxy Watch active product,
proved its stability enough.
Depends on {D8940}
Co-authored-by: JunsuChoi <jsuya.choi@samsung.com>
Reviewers: #committers, jsuya
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8941
Summary:
When the precomp layer(parent layer) has alpha transparency and has more than 1 child layer
and they overlap each other if vg object just propagate the alpha to child layer
it will be applied twice in overlapped area.
Even if the child layer does not have alpha transparency, parent alpha is applied to each child.
Test Plan: N/A
Reviewers: Hermet, smohanty
Reviewed By: Hermet
Subscribers: cedric, #reviewers, kimcinoo, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D9072
Summary:
This file is full of functions called as:
foo(eo_obj, obj);
Most of them can be reduced to foo(obj); and internally get the eo_obj
with obj->object
This would make it impossible to screw up calling them passing an
unrelated pair, and make calling code a little more readable.
ref T7230
Depends on D9048
Reviewers: raster, cedric, zmike
Reviewed By: zmike
Subscribers: #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7230
Differential Revision: https://phab.enlightenment.org/D9049
Summary:
This file is full of functions called as:
foo(eo_obj, obj);
Most of them can be reduced to foo(obj); and internally get the eo_obj
with obj->object
This would make it impossible to screw up calling them passing an
unrelated pair, and make calling code a little more readable.
ref T7230
Depends on D9045
Reviewers: raster, cedric, zmike
Reviewed By: zmike
Subscribers: #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7230
Differential Revision: https://phab.enlightenment.org/D9046
Summary:
This file is full of functions called as:
foo(eo_obj, obj);
Most of them can be reduced to foo(obj); and internally get the eo_obj
with obj->object
This would make it impossible to screw up calling them passing an
unrelated pair, and make calling code a little more readable.
ref T7230
Reviewers: raster, cedric, zmike
Reviewed By: zmike
Subscribers: #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7230
Differential Revision: https://phab.enlightenment.org/D9045
There is a wrong case that fails caching ector surfaces.
Those vector objects manually constructing shapes,
passed wrong pointer as the cacahing key instead of root node pointer.
This fixes it.
Coverity reports that both pname & file variables can be null here,
and that we are potentially passing NULL to strcmp. Check for valid
variables before passing to strcmp
Fixes Coverity CID1399423, CID1399421
@fix
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D8350
Here is a replacement to use eina_file from a vg obj instance
to map file data by vg loaders.
This brings a benefit that integrated access to load data
between vg object and vg loaders.
It may occur some threading corruption, since ector raster uses thread
internally. Unfortunately the issue is rarely occured, hardly catch
the exact point. But I guess this ector drawing range may be the root cause.
Commented out them as it did before, until we could make the issue sure.
this takes the current generated output from eolian for legacy code in
evas and adds it to the tree, then removes legacy references from the
corresponding eo files. in the case where the entire eo file was for
a legacy object, that eo file has been removed from the tree
ref T7724
Reviewed-by: Marcel Hollerbach <marcel-hollerbach@t-online.de>
Differential Revision: https://phab.enlightenment.org/D8107
Summary:
instead of passing a string which requires reading docs to know which
arbitrary string key=value sets can be passed, use an extensible struct
which contains more easily referenced values
ref T7672
Depends on D8035
Reviewers: segfaultxavi
Reviewed By: segfaultxavi
Subscribers: segfaultxavi, cedric, #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7672
Differential Revision: https://phab.enlightenment.org/D8039
Summary:
the previous implementation/api had a number of issues:
* "file" property contained both "file" and "key" values
- also performed file loading operation
* "load_error" property which was specific to image objects
* no methods for controlling file loading/unloading
this patch attempts the following changes:
* split "file" property into "file" and "key" properties
- also remove "key" from existing "mmap" property
* remove "load_error"
* directly return error codes from operations
* add "load" and "unload" methods for directly controlling load state
* add implicit file loading if file/mmap is set during construction
* rewrite all efl.file implementations to move file loading into load() method
* rewrite all usage of efl.file api based on these changes
* add C extension functions to mimic previous behavior
ref T7577
Reviewers: segfaultxavi, bu5hm4n, cedric
Reviewed By: segfaultxavi
Subscribers: vitor.sousa, #reviewers, #committers
Tags: #efl_api
Maniphest Tasks: T7577
Differential Revision: https://phab.enlightenment.org/D8018
the convention for event naming is to use $property,changed where possible
and to always emit related data with the event to reduce function calls
ref T7558
Reviewed-by: Marcel Hollerbach <marcel-hollerbach@t-online.de>
Differential Revision: https://phab.enlightenment.org/D7987
the save method is implemented very rarely, and most objects which use efl.file
will not be able to use a save method
ref T5719
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Reviewed-by: Marcel Hollerbach <marcel-hollerbach@t-online.de>
Differential Revision: https://phab.enlightenment.org/D7786
Previous masking image is generated on vg rendering time,
though context is changed to main vg rendering to masking,
This had a issue that ector context switching which is not allowed
in software backend because of asynchronou method.
Now, this improvment brings the masking rendering move to
vg render pre step with synchronous rendering method,
then use the masking surface in the main vg rendering.
Cleaned up the case when null parent is coming.
and found out efl_data_scope_get() could return invalid data
when invalid CLASS type is given. it's handled as well.
Dectected huge amount of unnecessray internal events triggering
which were signaled via vg nodes.
By events, nodes were connected with each others,
and finally reaches to vg object to update its rendering properly.
However GFX_PATH_CHANGE signal is too commonly triggered for scenarios,
listening and response it is too burdensome.
We acutally don't need to do this if all nodes could share the
vg object. Nodes directly notify vg object to update it.
Next patch will come to remove stupid GFX_PATH_CHANGE that's aweful
in performance wise.
Hermet Park
Taehyub Kim
Hermet Park
junsu choi
Cedric Bail
Pankaj Kumar
Christopher Michael
Mike Blumenkrantz
Mike Blumenkrantz