This explicit reset() was included for the editor_benchmark in
19037a2e69ef1aa9303eea5af8a200b403fdeba4 but it should be at the end
of the close() function.
This fixes a failing ASSERT() in ~DocObserverWidget when we destroy
the ContextBar.
This can happen in certain benchmarks that don't create a layer. We've
added a return value to getDrawableLayers() to know if the returned
range is valid.
CMakeCache.txt and other files (rules.ninja, other .cmake files) save
the full path to the cl.exe binary, including the specific MSVC
version. We've added an option to update that specific version to
continue using all builds/configurations with the new updated path to
cl.exe.
When auto-scroll is disabled, we can even paint moving the mouse
outside the editor viewport bounds, we should keep moving the brush
preview cursor anyway.
When we touch the editor viewport bounds, this patch:
1) Avoids flickering the cursor in auto-scroll mode, and
2) keeps the cursor moving when auto-scroll move is disabled
Unused functions as now we always use native custom cursors (or in
case we require this in the future, we could just use an UILayer to
show the mouse cursor).
* Replaced ui::Overlays with ui::UILayer
* Refactored BrushPreview to use its own UILayer (+ new blender shader
for the negative black & white effect).
* Add a default back UILayer to each ui::Display
* Use ui::move_region() in this "back layer" when GPU is disabled.
Although we should completely remove ui::move_region in a future, for
CPU rendering sometimes it's faster to just move memory instead of
redrawing the widgets.
This patch includes a new ui::Manager::dontWaitEvents() function to
avoid waiting for any user interaction. The old editor_benchmark now
uses this new function too.
Now ui::Theme is not an abstract class anymore and can be used for
simple tests/benchmarks. Some member functions are empty, but can be
implemented later if needed.
There is a moment inside Timeline::updateUsingEditor() where the
TimelineAdapter is nullptr between detachDocument() and the
updateTimelineAdapter() call, so just using a m_editor->getSite() was
going to ask for the timeline adapter if the selection was kept intact
("Keep selection" option enabled).
With this patch we simplify updateUsingEditor() just avoiding the
getSite() call at all (using the Editor fields directly) and handling
the special case where we already are in the same editor as a fast path.
With the combination of 8b0b1d8d1e2624eaa8433df58b83845902cd7975
(allowing transactions in docs without context) and
be17c4832405be44193a4d24268c38da80369743 (getting the active DocRange
from Doc context), we were getting a crash accessing a nullptr context.
