for release we need to test performance - esp of evas and eo in real life usage
and scroling is just such one. this adds a simple automated scrolling
test to genlist test (hit the button or ELM_TEST_AUTOBOUNCE=1
elementary_test -to genlist) so you can get consistent input and benchmark info
This change is the simplest I could do, but I think it would be better
to move _item_theme_hook inside _item_new. This is a further attempt
to fix T2777.
Summary:
Currently path is strdup() when getting real path, but it is not freed
in some cases.
@fix
Reviewers: seoz, Hermet, woohyun, cedric
Reviewed By: cedric
Differential Revision: https://phab.enlightenment.org/D3194
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Currently, longpressed event can be called when map is
scrolled, zoomed or rotated. This might cause unintentional
result, because mouse need to be down at lease once to
execute scroll, zoom, or rotate, and longpressed event will start
when those events takes long enough time.
Reviewers: seoz, Hermet, woohyun, cedric
Reviewed By: cedric
Differential Revision: https://phab.enlightenment.org/D3195
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
libatspi implementation allows to call methods on objects which
are not added to client cache with "AddAccessible" at-spi signal.
Bridge should allow to perform such calls by updating its internal
cache as soons as it returns a reference to the dbus object to client.
Elementary is not using much the calculate callback from Evas. This
create all sort of problem where it transform O(n) algorithm. Like
sizing a toolbar for next frame to O(n^n) by actually sizing the
toolbar every time an object is inserted.
T2777
_sizing_eval is very costly on toolbar as it recalc all edje part and sends
signal to all of them. _item_theme_hook was already calling _sizing_eval
doubling the cost of adding new item.
various fixes have occurred over the years to improve proxy rendering.
as a result, this theme, which is apparently the only other user of proxy
edje parts besides deskmirror, was broken.
@fix
Summary: clang reports SIG_BLOCK_CLICKED and SIG_TIMEOUT as being
unused variables, so remove
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: clang reports passing EINA_FALSE here is being treated as an
Eo object, so pass NULL instead
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: clang reports passing EINA_FALSE here is being treated as an
Eo object, so pass NULL instead
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: clang reports ELM_APP_SERVER_VIEW_PROPS_ENTRY as being an
unused variable, so remove.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Scenario:
const char * accel = accel_get();
accel_set("gl:depth");
Now accel may not be valid anymore since the original
Eina_Stringshare might have been destroyed. So there was
no point in saving the const char * pointer. It was not valid
after the call to set(). This is a bit confusing here. Returning
char* would be clear: the caller would own the data.
too many rels and complexity to track this to a more coherent
message
probably broken since 1.0 since there are zero "noicon" toolbar
tests
@fix
fix T2782
Elm_Image rely on Evas_Image ability to rotate content. There is no
pixels manipulation involved that require an update_add here. In fact
this cause a bug where the content would disapear due to evas trying
to get some pixels data that are not where it think they are.
Summary:
In hover.edc, hoversel dismiss animations are defined,
but it doesn't work at all. (Hover is removed before animation)
This patch makes hoversel remove its hover after dismiss animation.
@fix
Test Plan: elementary_test hoversel
Reviewers: Hermet, cedric
Reviewed By: cedric
Subscribers: woohyun
Differential Revision: https://phab.enlightenment.org/D3070
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary: As ecore_evas_wayland_move ends up updating the location of
the ecore_wl_window, we should be calling move with the proper
geometry as this ends up updating the internal ecore_wl_window
geometry (which in turn sends to xdg_surface_set_window_geometry).
This is part of a fix where xdg_surface_set_window_geometry was
getting sent inproper values.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This reverts commit 1094fb77a0ee23cbe1e8b15783e406a8d93b7fd4.
Feature is freezed for new released version.
This patch must updated after release finished.
Toolbar creates a more_item when shrink mode is
ELM_TOOLBAR_SHRINK_EXPAND.
If select mode is ELM_OBJECT_SELECT_MODE_ALWAYS and there is no item
appended, then toolbar sets a selection on a newly created item.
In this case, sd->more_item is referenced before the new more_item is
assigned to sd->more_item.
@fix
Summary:
Add elm_gengrid_item_all_contents_unset API for reusing contents in application side.
User can unset all contents or re-use their contents for increase scrolling performance.
genlist already support this API.
@feature
Test Plan: Need to add test in elementary_test.
Reviewers: raster, singh.amitesh, cedric
Subscribers: woohyun, Hermet, seoz
Differential Revision: https://phab.enlightenment.org/D3066
previous config ignored elm theme if ithas cursors. this was just
wrong, so switch default back to using theme and update configs
accordingly etc. this is houw it should have worked.