Cancelling a promise will fulfill it but won't actually free the memory. This
memory is under custody of the owner, who must either call value_set or
error_set to finish it.
We treated as errors when Eina_Accessor was NULL, which meant
that we could never get an empty range os children. Treat it
as success and pass an empty Eina_Accessor
We used to keep a reference to the parent object and have it in the call
structure although we were actually calling the function on the embedded
object. This was needed because we wanted to unref the parent correctly.
This was incorrect (and marked as a hack) and now I finally gotten
around to implementing the (amazingly simple) fix to remove this
workaround.
Essentially we just ref the comp object, unref the parent, and let the
normal eo call flow to unref the comp object correctly later on, like it
would have unreffed the extra ref we had for the parent.
I guess eeze scanner was not operational at all because the SUID bit was
never set. The reason for that was that where it was put in the makefile
made it not be a make rule (where @ would have worked) but used by a special
autofoo rule.
Pffff... another tricky one...
-windowDidResize is actually also called when the window is not
resize by the user, leading to multiple event posting, and of
course the infamous call of ecore_main_loop_iterate() when it
was not paused at all, leading to messy events handling... the
most visible being the initial resizing of the window going rogue.
We now ensure with -windowWillStartLiveResize and -windowDidEndLiveResize
that we only send an event when the user requested it. Since the main
loop is paused at this point, calling ecore_main_loop_iterate()
becomes safe.
Fixes T3648
Autotools really really really sucks.
Because of the above, we need to manually list both source dirs and
build dirs, because some of the include files are in the source dirs and
some are in the build dir (generated).
When compositing objects, we were checking that class_of(B) is in A's
inheritance tree before allowing attaching B as a composite object of A.
This is wrong and breaks a few cases, for example: B extends a class that
is in A's inheritance tree or B implements an interface that is in A's
inheritance tree.
Thanks to Marcel for reporting.
Evas events recently changed, and involved the use
of the 'multi' field within a mouse mouve event.
It was used afterwards for mouse scrolling, but since
it was never set by ecore_cocoa, scrolling went
broken...
Fixes T3789
The bug came from the fact we need to handle the destruction of the
main loop which destroy the underlying timer. The event handler that
catch the destruction of the timer can not make the difference between
eo_del call from the timeout code and eo_del from the main loop
destruction. By removing the event handler, the double free is properly
avoided.
As we add more object in the main loop, they can't live in the top
namespace as they make little sense there (Efl.Fd !). For coherence,
everyone should in the loop namespace, so move timer there.
Now when dealing with pointer types, we will not get pointer to
pointer semantics in callbacks and eina_promise_owner_value_set
for Eina_Promise.
It will work as expected:
Eina_Promise_Owner* promise = eina_promise_add();
void* p = malloc(sizeof(T));
eina_promise_owner_value_set(promise, p, &free);
The call to eina_promise_then steals the first ref'count, so it is
possible that the promise is freed after the eina_promise_then,
so we need to eina_promise_ref before eina_promise_then.
Summary:
Implemented interface Efl.Gfx.Buffer functions bufer_map/unmap for Efl.Canvas3D.Scene.
Added function e3d_drawable_texture_rendered_pixels_get to module evas_gl_3d
to getting pixels from FBO. Added wrappers for functions
e3d_drawable_texture_rendered_pixels_get and e3d_drawable_texture_id_get
to have possibility call it through engine functions.
Reviewers: cedric, Hermet, raster, jpeg
Reviewed By: jpeg
Subscribers: jpeg
Differential Revision: https://phab.enlightenment.org/D3978
So... I had issues with evas-fb engine which was massively leaking,
one image per frame.
After investigating a bit with @cedric on IRC, the reference count
of the cache entries was always 2 before the engine dropped.
So, for each frame with an animation, we could never drop a cache
entry, leading to a trumendous amount of memory leaking.
Now for non-async rendering, we copy the behaviour of
evas_render_pipe_wakeup() which is called in async-mode,
and actually drops a reference in the cache entry.
Fixes T3763
Not sure if this test case is really good, but it was
necessary to prove that input event refeed can work from
application side. This is a feature that should work
but shouldn't be used :)
This does:
1. Forward keyboard events from evas to win
2. Allow feeding external input events
Input events can be faked by apps by simply forging
eo objects of the proper type (key or pointer evt) and
calling eo_event_callback_call().
Such events will be forwarded to the internal Evas, and
some bool flags prevent infinite refeeding loops.
efl_event_dup() returns fake events for this to work.
@feature
This reverts commit ac5a0b7e79.
I added assert(pd == sd) in the modified code and couldn't find
anything suspicious. The proper class is used in eo_data_scope_get().
I also can't find a crash or a bug currently in hover. I know there
was a problem earlier, but that was fixed in 3092e08a83.
Now, if there is still a problem, we should check it carefully.
Pinging @cedric
Cedric revert patch 6f7608befd same days
ago.
<quote Cedric>
Your patch 6f7608befd did break elemines
which does relly on the proper index to be given to it. I think
Edje_Edit should be fine and use the index if no name is provided (See
edje_load.c around L1180). In all case this was clearly an ABI break and
needed to be reverted.
</quote Cedric>
So now edje_cc not guarantee name for part items. This is bad, because
edje_edit use item name for access to item attributes. If item have not
name we cann't get it attributes, because edje_edit compare the given
name with part items name from edj-file, strcmp dislake NULL args.
For now I see one way mark all edje_edit_part_item* APIs as deprecated
and add new APIs for manage part items by index.
For clearly and history:
The eason for revert - https://phab.enlightenment.org/D3820
The proposed solution - https://phab.enlightenment.org/D3907
We can not assume that the elementary parent will be also somewhere in the eo parent
hierarchy. We need to explore both when doing an eo_provider_find in elementary for it
to be really useful.
As we already link to Eeze and use it throughout this code, there is
no point in adding an extra dependency on libudev itself when we can
just use Eeze functions to get what we need. This is a better fix for
T3712
@fix
Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
This lets me narrow down the remaining cases of pointers across the EFL.
The void pointers will later need to be reevaluated on per-case basis and
replaced appropriately where possible/feasible.
1. remain elm_win_type_set
2. remain elm_win_alpha_set
3. remain elm_win_available_profile_set/get
app usually set availble profile into window, and app fw or other lib can read it.
so I remain get/set api both.
4. remove elm_win_profile_set/get from eo
5. remove illume feture in eo and move it to legacy
Apparently I broke some inputs in E (efm) like mouse wheel.
Somehow the list of objects where the pointer is in was NULL.
This was because the mouse_in/out events were not matched to
the proper window ID.
Fixes T3760
Yup, I broke everything again. This time, mouse move inputs
would not move the cursor, since I was bypassing the regular
_ecore_evas_mouse_xxx callbacks.
Fixes T3766
So, I was stupid. I was relying on legacy callbacks to
trigger eo events, which means that only when a legacy
callback was registered would my new eo events be triggered.
Instead, I can pass the eo event desc & info whenever
calling evas_object_event_callback_call().
We only care whether the process terminated successfully or not.
Non-zero values will be anything but a success, so we don't even care
about the exit code of the program.
This should fix compiling on OpenBSD (where WEXITSTATUS needed
to be defined by including a system header) as well on Windows.
Fixes T3733
This code was written when eo_del() was removed and eo_unref() was the
recommended practice. Since we added eo_del() back we now need to adjust
this new code accordingly.
These files are not getting cleaned up as they are not integrated into the rest
of our build system. These are the extra makefiles to build build only eina, eo,
etc.
The trick with -C clean does not work here for me and this is the only way I
found to get a distcheck going through. Which is kinda important given we want to
do the first alpha tarballs for 1.18 next Monday.
This is a new incarnation of 0a03e63350. Our list
has grown to big again as people insist of adding the generated eolian files to
DISTCLEAN while BUILT_SOURCES will get removed durign the clean anyway.
Adding this file list twice will just make the argument list for rm to long to
work.
This reverts commit 546ff7bbba.
It seems that eo_del() is useful and removing it was creating bugs.
The issue is that the way we defined parents in eo, both the parent and
the programmer share a reference to the object. When we eo_unref() that
reference as the programmer, eo has no way to know it's this specific
reference we are freeing, and not a general one, so in some
circumstances, for example:
eo_ref(child);
eo_unref(child); // trying to delete here
eo_unref(container); // container is deleted here
eo_unref(child); // child already has 0 refs before this point.
We would have an issue with references and objects being freed too soon
and in general, issue with the references.
Having eo_del() solves that, because this one explicitly unparents if
there is a parent, meaning the reference ownership is explicitly taken
by the programmer.
eo_del() is essentially a convenience function around "check if has
parent, and if so unparent, otherwise, unref". Which should be used when
you want to delete an object although it has a parent, and is equivalent
to eo_unref() when it doesn't have one.
1. remain borderless property.
2. shaped property was removed because wayland not support this.
3. remain noblank property.
4. remain role, name properties.
5. remove title property because it can be supported using name property
1. remain iconified, maximized, fullscreen property
2. remain sticky property
3. change return value type of modal from bool to enum for the future.
4. merge urgent_set and demand_attention
5. remain activate method
1. remain focus_highlight_style,focus_highlight_enabled,
focus_highlight_animated,prop_focus_skip,focus. because each window need to set the focus
feature.
2. autohide, autodel is remained for convenient.
3. icon_name will be removed. elm_win_icon_object_get can used instaed of icon_name.
Summary:
- added is_active property in Evas_Key_Grab.
- Evas key down and key up events transferred to active grabs only.
- If evas grabs has a exclusive grab, other grabs will be deactivated.
- If a exclusive grab is ungrabbed, remained grabs will be activated.
Reviewers: jypark
Reviewed By: jypark
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D3996
elm_win have three feature releated with screen.
1. screen_rotation_get
2. scrren_size_get
3. screen_dpi_get
so create efl_screen interface, and elm_win implement that interface
Summary:
We did the same for evas_canvas3d_scene_pick(see c850cc0d80),
but forget for evas_canvas3d_scene_exist.
Reviewers: Hermet, raster, cedric
Reviewed By: cedric
Subscribers: jpeg
Differential Revision: https://phab.enlightenment.org/D3973
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
It is logically break cycle in case object is found. In case need pick all objects
that lay in screen ray, we should use evas_canvas3d_scene_pick_member_list_get
function.
Reviewers: Hermet, raster, cedric
Reviewed By: cedric
Subscribers: jpeg
Differential Revision: https://phab.enlightenment.org/D3975
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
- elm_map has a scrollbale interface and it is set
as ELM_SCROLLER_SINGLE_DIRECTION_SOFT by default.
- elm_map can be rotated by gesture or by an API
elm_map_rotate_set, so this single direction
makes scroll unnatural.
Reviewers: Hermet, cedric
Subscribers: conr2d, jpeg
Differential Revision: https://phab.enlightenment.org/D3986
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
We do properly unref promise while calling all the then callback. There
is no need to check it a second time (which actually lead to a 100%
bad access).
T3759
Small patch to deprecate Ecore_Drm. This patch also adds a configure
option to enable ecore_drm for older code. This option is disabled by
default, so must be explicitly specified during build.
Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
For ecore_evas drm engine(s), disable setting of
ecore_event_window_direct_callback as this Completely Breaks all input
when running Enlightenment Wayland.
NB: This can likely be re-enabled at some point, when the jpeg
breakage is over ;)
@fix
Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
This adds a few classes, in particular Efl.Event and Efl.Event.Point
which are used as the event info for all pointer (mouse, multi) call
Using an eo object as event info will allow for future extensions
really easily. We don't need to expose any of the internals, also
a single type can be used for all pointer events.
Still TODO:
- Keyboard events (urgent)
- Axis / Joystick events (needs porting from Tizen)
- Gestures support (probably later)
- Event feeding from app side (manual feed)
Not going to be ported to EO:
- Hold API and hold event. Seems barely used.
The new APIs are not very much tested at this point, and rely on the
old legacy event system. The most important thing right now is to
ensure that nothing was broken so far. Unfortunately I can only do
this much testing myself...
Merge branch 'devs/jpeg/work'
If on_hold or on_scroll is set in an eo callback, the subsequent
calls to the legacy callbacks will also have this flag set.
Inversely the legacy callbacks should affect all subsequent eo
callbacks.
Note: those are just indicative flags.
This is going back to the same idea as legacy. We will have
events such as:
- move
- down
- up
- in
- out
- wheel
- cancel ("new" - very rare)
Now the question is whether/how we should divide "multi" events
which start from the 2nd finger from standard mouse events. The first
multitouch finger should by default look like a mouse event.
This moves Efl.Pointer.Event back to Evas. Originally I wanted
to share this class with Ecore but eventually I didn't need to
do so, since only ecore_evas (which depends on evas) really needs
access to these.
The internal data struct is not moved out of efl (yet?)
This continues the work started in the previous commit:
forward full event info (Efl_Pointer_Event) from evas to
the upper layers.
This also includes more support for IN and OUT.
This is still VERY experimental and not fully done yet.
All other pointer events need to be sent as well.
The legacy event system is used as a transportation mechanism,
as it is too hard to change the logic. This only adds an extra
eo event in case of move. Obviously for performance we might
want to listen to callback_add,del but that's an optimization
for later.
The whole point of sending those pointer events is to carry more
information than can be sent over legacy evas events, and unify
the events in a common format.
This whole input system is a massive mess. It looks like spaghetti.
Long live the giant flying monster.
This commit changes how some events are propagated in X.
Before:
ecore_x -> evas_event -> evas
After:
ecore_x -> ecore_input_evas -> ecore_evas -> evas_event -> evas
There are still inconsistencies between events and between X and WL,
but ecore_evas should be used for all events since it rotates the
inputs.
This way, ecore sends eo events to evas, which can then
be listened to by other clients (FIXME: the events will
need to be propagated from evas to the elm window).
Current support:
- mouse/multi down
- mouse/multi up
- mouse/multi move
- mouse wheel
Since the direct input event callback returns true if the
event has been processed, we can easily support legacy and
progressively implement full support for eo input events.
This object is the data carried over in an event data pointer.
The private data should be accessible by Ecore and Evas, but
not externally. This means we should be able to easily extend
the feature set, adding more and more information, without
breaking API / ABI.
Also, this should allow applications to create fake input
events easily without exposing our internal structures, or
functions with more and more parameters (such as feed multi).
This is only a storage class, shouldn't contain any logic.
In the future, some logic may be added for gestures support
for instance, or input smoothing / resampling (eg. if input
frequency is 90Hz and screen refresh rate is 60Hz).
The aim is to replace:
- Evas_Event_Mouse_Xxx
- Evas_Event_Multi_Xxx
- Ecore_Event_Mouse_Xxx
We might want to also support Axis, Gestures, etc... with the
same model or even same storage class.
So, this is not a very clean solution, but this mostly
makes Evas_Device an Eo object of class Efl.Input.Device.
Since evas_device relies on some Evas knowledge (evas
callbacks, canvas private data), it can't be fully moved
to lib/efl/.
Making the input device an interface rather than a class
was also not a great solution, as the goal is to share
the data structure around EFL internals (Ecore and Evas).
Test wasn't enabled when modifications to function prototypes to
eio_job_file_xattr_list and eio_job_file_xattr properties were
made, so it got pushed inadvertidly without the necessary
changes to user code.
previously, elm_image refered to align values
set by evas_object_size_hint_align_set function
to position the internal image inside an image object.
size hint value is for positioning the image object inside its container object,
so the previous code is inappropriate.
this commit adds 'align' property for positioning the internal image.
this is an args event. right now we don't use it, but this should be
done by some of the setup/init of an app and then produce an args
event. the idea would be that this can be used by single-instance apps
like web browsers, terminology to treat launch as an event.
Summary:
Currently when an item in gengrid is focused and if it is scrolled
off the viewport halfly and if it user clicks on it, it won't be
scrolled into viewport. Only non-focused items will be scrolled into
viewport when clicked.
Test Plan:
elementary_test -> Gengrid focus -> Click an item ->
Scroll it little bit out of viewport -> Click again
Reviewers: prince.dubey, shilpasingh, cedric, raster, SanghyeonLee
Subscribers: seoz, rajeshps, govi, jpeg
Differential Revision: https://phab.enlightenment.org/D3977
Previously, preserved content was shown unexpectedly after its naviframe
item was popped because the preserved content was not removed but was
unset from its naviframe item.
Since popped naviframe item becomes hidden on the screen, the preserved
content of the popped naviframe item should become hidden by default as
well.
So, I was storing a single "sd" pointer (class private data)
but using it in two different places, ie. as if it were two
different classes. Crashes happened, complaints were heard.
See T3714.
This fixes T3714
Efl.Part uses a basic inheritance for layout objects, but somehow
this completely falls apart when the child class overrides all
the functions. Weird. I'll check that more.
Summary:
When the widget is unset from any container, a parent of the widget
doesn't exist. So we should set its parent to the top object.
But if we just set sd->parent, the parent can not find the widget as a
child. So the container widgets set the parent-child relation when
sub object is unset.
This commit is related to 0822ad2195.
@fix
Test Plan:
Check this issue.
https://phab.enlightenment.org/D3855
The unset widget don't added any widget as child.
So when it set scale, the widget can not reload the thmeme.
Reviewers: raster, cedric, Hermet, reutskiy.v.v
Reviewed By: Hermet, reutskiy.v.v
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D3957
Conflicts:
src/lib/elementary/elm_mapbuf.c
This was actually difficult...
Mac OS X can use addr2line (sometimes called gaddr2line in
function of the package managers). However, addr2line does
NOT handle specific cases.
It was therefore necessary to use Mac OS X' own tool: atos,
which gracefully handles all backtraces, including the one
containing objective-c messages or fat archives.
eina_btlog now tests different utilities one by one, and
determines whether it is supported or not.
Fixes T3711
The default scale type should not be none as this breaks elm icon and elm image as it doesn't follow the default view of elm image before elm image scale was added. The default behavior is fill so this commit makes fill the default.
This patch fixes an issue where starting Enlightenment would not
center the mouse pointer. Basically the issue is that we cannot warp
the mouse pointer until After the ee->prop.window has been set else
E will not process the mouse_move event
@fix
Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
This adds a new API function that Ecore_Evas can call to calibrate
input devices given size of the output
Signed-off-by: Chris Michael <cpmichael@osg.samsung.com>
We need to continue to tick even if we can't find a buffer - otherwise
we'll stop updating forever.
We also need to reset buffer busy bits to 0 if we run out of buffers or
we'll never try to assign one. (The page flip handler would eventually
clear the busy bit for the current buffer, but a few lines ago we just
set that to NULL so it will never happen...)
If we don't block the current buffer from being assigned to the new
current buffer we'll end up rendering into the same buffer over and over
and never actually double buffer.
Larry Jr
Lauro Moura
Felipe Magno de Almeida
Tom Hacohen
Davide Andreoli
Jean Guyomarc'h
Cedric BAIL
Derek Foreman
Oleksandr Shcherbina
Jean Guyomarc'h
Vyacheslav Reutskiy
Carsten Haitzler
Ji-Youn Park
Jean-Philippe Andre
Jaehwan Kim
Mike Blumenkrantz
Chris Michael
Daniel Kolesa
Jee-Yong Um
Stefan Schmidt
JEONGHYUN YUN
Subhransu Mohanty
Sungtaek Hong
junsu choi
Andy Williams
Jaeun Choi
godly.talias
Jaehyun Cho
Hermet Park
Stephen Houston