in the previous commit we started to only receive the parent_obj from
the evas object when its not a widget. However, we still need to ensure
that the parent is equal to the current one in order to remove it when a
widget.
While this is one more call, the call to get the parent is not as
heavy as the data_get call, as this directly addresses a struct, and
does not read from a hash table.
This also fixes a test case failure.
in genlist testcase the number of calls reduced by 30000.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Reviewed-by: Hermet Park <<hermetpark@gmail.com>>
Reviewed-by: Youngbok Shin <youngb.shin@samsung.com>
Differential Revision: https://phab.enlightenment.org/D11984
As we always make sure the parent is a widget class we don't have to check anymore.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Reviewed-by: Hermet Park <<hermetpark@gmail.com>>
Differential Revision: https://phab.enlightenment.org/D11985
this here needs to pass the flag that gets set, not the flag that
results. This is important for the cases, where a subtree in a widget
tree keeps the flag. As this would leave the wrong counter.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11556
Summary:
the internals of those two methods have been the same. Both setted the
internal numeric flag as either as 1 more or equal to the flag in the
parent object (depending on the internal state). Further details can be
found in the comment in code.
Depends on D11554
Reviewers: zmike
Reviewed By: zmike
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11555
Summary:
eval all children is ending up in the complete discovery of the whole
subtree, which is unnessesary here, as we are already discovering the
whole subtree with the calls, these changes are in, so simple
evalulating this is enough.
Depends on D11551
Reviewers: zmike
Reviewed By: zmike
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11554
Summary:
this fixes disabled set behaviour. This ensures that when setting
disabled twice, that unsetting it once does not break the overall state.
This never appeared in any real life example, because
elm_object_disabled_set is already checking for equalness. However, this
is not wanted here, because the simple setter can also be used to sync
the state with the parent, which appears to be helpfull.
Depends on D11550
Reviewers: zmike
Reviewed By: zmike
Subscribers: zmike, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11551
Summary:
when setting twice the same value, unsetting the same value would not
restore the same state in the tree again. With this commit, we ensure
this is working correctly.
Reviewers: zmike
Reviewed By: zmike
Subscribers: zmike, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11550
In EFL we have multiple hirachies of parts. One in Efl.Layout namespace
(alias edje) and one in Efl.Ui namespace. The seperation of these two
makes sense from the perspective of hiding functionality. However, a
functionality that we want to have on both is: we want to be able to
check which type of part this is. In order to do so, this commit
introduces a common interface, which allows that.
This is required because eo is currently undergoing some works, where
only APIs on a object are allowed, that are also inheriting its type,
which is normal in OOP, but sometimes, due to the lack of limitation, we
did that. This commit resolves one more case of that.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11516
this safes in the runtime of elm_test round about 9kb (*).
Additionally, using a array here is jumping way fewer times arround in
memory, as we do not need to jump from node to node in a list.
Additionally, this commit replaces a few abort disabler macros with a
error checking macro. (which cleans the log).
*: explanation: we have round about 600 widgets in elm_test, every
widget is normally refered once, every list node has 4 pointer, makes
round about 9600 bytes or rougly 9 KB. So the messured savings are more
or less explaining the reality.
Reviewed-by: Carsten Haitzler (Rasterman) <rasterman.com>
Differential Revision: https://phab.enlightenment.org/D11374
so sometimes widgets dont have smart data - some situtions happen...
but let's not segfault. instead we now see the usual eina log eerrs
and things march on... and no segv.
@fix
if there was a spacer around elm.swallow.content of scroller,
position of pan object would be different with position of the scroller obj.
this patch modifies a fomular that calculates some points relative to scroller
to a fomular that calculates the points relative to pan obj.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D10745
Summary:
The internal and the API we would like is mostly a canvas API. A lot of the code
in evas is working around the fact that efl_input_device is not defined inside Evas.
This patch is the first step to try to clean this up.
Depends on D10487
Reviewers: zmike, raster, bu5hm4n, Hermet
Reviewed By: zmike
Subscribers: #reviewers, #committers
Tags: #efl
Maniphest Tasks: T8321
Differential Revision: https://phab.enlightenment.org/D10488
Summary: these should error so the user can detect that they screwed up
Reviewers: devilhorns
Reviewed By: devilhorns
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D10563
The idea of this widget is to provide to MVVM what Efl.Ui.Collection provide and
leverage the same shared logic for layout.
Co-authored-by: Mike Blumenkrantz <zmike@samsung.com>
Co-authored-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D9958
Summary:
Based on the discussion T7883, there is no reason to keep those
interfaces separately. So, now, efl_ui_scrollable will support
the events and methods together.
@ref T7883
Reviewers: eagleeye, bu5hm4n, zmike, segfaultxavi
Reviewed By: segfaultxavi
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7883
Differential Revision: https://phab.enlightenment.org/D10072
7321911468 moved the constructor to to
top, which leads to the fact that a few widgets cannot fetch the shared
data of the window. Which made the test suite fail.
With this commit we are just moving the sub_object_add after the
constructor, which fixes the flip crash.
This fixes the test suite
the object is not properly set up yet for child classes -0 parent like
the main evas object or smart obj havent been constructed yet so call
this really early, not in the middle of construction of an efl ui
widget...
this fixes a segv in elementary_test -to filp where it segv's on
object construction where evas object ->cur/prev are NULL (not set up
yet).
Efl.Color and Efl.File cannot be done due to composition. Efl.Gfx.Image
can, so we can solve this via composition, this also resolves all the
other cases for Efl.Gfx.Image.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D9991
Summary:
first of all, in efl-ui we should probebly ensure that a widget is
always created in a window object. Otherwise we are looking for trouble.
Additionally, calling efl_ui_win_shared_data_get on anything else than a
window object will result in a returned NULL value.
If we are not having a widget parent, there is also not much point in
calling a API that is only defined on the widget base class, so we also
move that away
Reviewers: kimcinoo, raster
Reviewed By: kimcinoo
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D9762
Summary:
`Efl_Ui_Scrollable_On_Show_Region` is only used for legacy EAPI. it moved to
elm_widget.h from eo file. also, it renamed `Elm_Widget_On_Show_Region_Cb`.
Reviewers: Jaehyun_Cho, zmike
Reviewed By: zmike
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D9896
unload is a crucial method for classes that implement load, as this is
the method which is called during e.g., efl_file_simple_load() when
a new file is specified in order to destroy the previous object data
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D9786
It was possilbe to add an image with improper parent object before aa2d94f and
56752e0. This patch makes it possible. Sure there are error messages when user
adds an image object using non widget object as below but you can see image.
ERR<28822>:elementary ../src/lib/elementary/efl_ui_widget.c:4801 _efl_ui_widget_efl_object_constructor() You passed a wrong parent parameter (0x400000007ced (null)). Elementary widget's parent should be an elementary widget.
ERR<28822>:elementary ../src/lib/elementary/efl_ui_widget.c:4803 _efl_ui_widget_efl_object_constructor() No widget data for object 0x400000007ced ((null))
ERR<28822>:eina_safety ../src/lib/elementary/efl_ui_win.c:9450 efl_ui_win_shared_data_get() safety check failed: pd == NULL
ERR<28822>:eo ../src/lib/eo/eo.c:579 _efl_object_call_resolve() in src/lib/elementary/efl_ui_widget.eo.c:256: func 'efl_ui_widget_sub_object_add' (698) could not be resolved for class 'Evas.Canvas'.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D9747
if we are walking this method twice (spoiler we do!) then we subscribe
twice to the event, which leads (depending on the order of executed
deletion) to the accessing of obj after its deleted.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9684
as an example: elm_test uses roughly 500 widgets, this pointer takes 8
bytes, this is wasting quite some time, which can be perfectly safed by
just structuring the code a little bit more. In some callbacks we need
to call one more time efl_data_scope_get but that is not a that big
issue, as most calls after that are also doing a eo call on that object,
so the eo_id is already cached.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9683
since 2dd596084b, tree_unfocusable is changed from
Eina_Bool to int.
Double negation ensure binary return value of `elm_widget_tree_unfocusable_get`
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D9673
this has to use tree_unfocusable not disabled.
ref D9644
Reviewed-by: YeongJong Lee <yj34.lee@samsung.com>
Differential Revision: https://phab.enlightenment.org/D9671
Dereference null return value issues is reported by Coverity.
If `sd` is NULL, it will return zero values.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D9668
this is used for storing the position and size of the widget. However,
just in elm_test this eats 2KB of ram, just for saving them. The cases
where they have been used do not seem like a great performance hotspot,
additionally, in the panel cases the privat data of the widget is
achived anyways, so eo is involved and we are spending some time there
anyways.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9646
walking here the parent chain can be very costly, esp. if you make a lot
of widgets visible or hide them. With this commit we ensure that we only
walk the parent chain when there is a custom chain, which is only the
case in legacy. This gets us about 2-3 FPS on the rpi with the
item_container benchmark.
With earlier commits, tree_unfocusable and disabled is honoring the
whole parent chain, so checking that alone, without the parent chain, is
fine.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9645
Before this patch tree unfocusable set just set a simple flag, getting
this state however forced you to traverse the whole parent chain to get
the real result.
With this patch the setting of the unfocusable flag is heavier as it
walks all the children of the widget, however, the getting of the flag
is way easier now. The next revision will refactor the focus related
APIs for that.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9644
asking for a parent provider only makes sense if we are having a custom
focus parent provider, otherwise this will just result in the normal
parent relation. Not doing this safes us a lot of time that we spent for
nothing in the focus evalulation code.
The gengrid change is needed as the parent provider should be applied
after the constructor is called, otherwise there is no garantee that
shared_win_data is not NULL.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9643
this introduces a shared place to all widgets. The shared pointer is
allocated in the window, as the window outlifes every widget that is
part of it.
This struct will be later used for further optimizations like:
1. There are really heavy focus operations which are only needed for
gengrid/genlst, there is no point in executing them if there is no
gen**** added to the window object. So we can skip the custom
parent_provider logic that is only introduced for gengrid / genlist.
2. Legacy focus APIs must do list walks, which means, on every focus
operation we always have to walk the full list up to the parent, which
is annoying and slow, as we *most of the time* do not use legacy focus
API.
This list can be continued, the above two cases are fixed in the next
revisions.
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9642
calling geometry set here is again calling the API in canvas object that
splits this call to size_set and position_set which means we spent quite
a bit of time in eo, just to call the same APIs we could call directly.
With this commit here, the calls are directly going to the right
objects, with the right API.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Differential Revision: https://phab.enlightenment.org/D9619
Summary:
historically there have been two methods of calculating sizes in elm:
* elm_layout_sizing_eval
* evas_object_smart_calculate (now efl_canvas_group_calculate)
the former was used to set size hints on widgets, while the latter was
used to perform internal size calcs for the widget. for things to
work correctly, these functions had to be triggered in just the right
order at just the right time. many hard-to-fix bugs related to widget
sizing over the years have been the result of this split
this patch removes elm_layout_sizing_eval implementations so that all
widgets perform both internal size calcs and size hint setting all
in the same function, ensuring that these are always in sync
the result is that in the vast majority of cases, far fewer recalcs
happen for widgets, and they are quicker to achieve their final size
Depends on D9438
Reviewers: bu5hm4n
Reviewed By: bu5hm4n
Subscribers: bu5hm4n, cedric, #reviewers, #committers
Tags: #efl_widgets
Maniphest Tasks: T8059
Differential Revision: https://phab.enlightenment.org/D9439
Elyes HAOUAS
Marcel Hollerbach
Subhransu Mohanty
Mike Blumenkrantz
Hosang Kim
Cedric BAIL
Carsten Haitzler
Wonki Kim
Xavi Artigas
WooHyun Jung
Yeongjong Lee
Shinwoo Kim