Instead of cmpletely replacing the IME snippet for every update, expand
the old range if there is overlap. This change avoids never-ending
restarts of the IME on Android where snippets are expanded in two
calls, one for expanding before the selection and one for exanding after
the selection.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
It's much simpler to map the Android back button to a key.Event and
let the usual key filtering determine whether to block its default
behaviour.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This change adds key.NameUp/Down/Left/Right and maps the Android TV
remote directional keys to them. As a side-effect a key.InputOp can
now receive directional keys (and block their focus movement).
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Before this change, Tab and Shift-Tab would always result in focus
movement. This this change, a key.InputOp with a matching Keys set
will block focus movement and deliver the events to it.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
We should return as soon as possible from FrameEvent.Frame to allow the main
goroutine to continue processing other tasks. Whereas GPU.Frame may touch
the frame ops, GPU.Present will not, so this change moves Present to after
returning from FrameEvent.Frame.
This is a variant of 38ff78df5d that
works on OpenGL where context locking is required.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
We should return as soon as possible from FrameEvent.Frame to
allow the main goroutine to continue processing other tasks.
Whereas GPU.Frame may touch the frame ops, GPU.Present will not,
so this change moves Present to after returning from FrameEvent.Frame.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
List was recently changed to include an extra child at each end, to
automatically scroll when reaching the end of a focus direction. However,
if List includes unfocusable children that strategy may fail. This change
adds another fallback where app.Window will scroll a constant amount in
the focus direction, to reveal more children.
For https://github.com/tailscale/tailscale/issues/4278.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
A focused widget may be partially or completely off-screen in which case
the user will have difficulty interacting with it. This change attempts to
scroll the focused widget into view by issuing synthetic scroll events.
For https://github.com/tailscale/tailscale/issues/4278, but doesn't completely
solve it because layout.Lists won't layout focusable widgets outside its visible
bounds. A follow-up change deals with that.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This is a refactor to make it easier to add higher level logic to
focus moves. A follow-up will add automatic scrolling to bring
focused widgets into view.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
vkDeviceWaitIdle and vkQueueWaitIdle are expensive; a vkFence is cheaper
and the usual way to ensure a previous frame has completed before starting
another.
References: https://todo.sr.ht/~eliasnaur/gio/375
Signed-off-by: Elias Naur <mail@eliasnaur.com>
After 34f10d9cbb, the display link callback
will never block. However, if the main thread is blocked for another reason,
say a bug in the user program, callback requests will pile up as blocked
goroutines.
This change ensures that a redraw request is queued only if no other request
is pending.
References: https://todo.sr.ht/~eliasnaur/gio/370
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Previously, the window size is equal to the screen size. That doesn't
consider the size of the taskbar, causing the height be bigger than the
real height. Now, the maximized have the same behavior of windowed,
since both of them must include decorations and taskbar.
Signed-off-by: Inkeliz <inkeliz@inkeliz.com>
GioView.onCreateView may call methods on the imm field of GioView.
Make sure it is initialized before use.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The monitor may have a different backing store scale, yet the
system doesn't redraw automatically when that happens
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This reverts commit 11f39582b8 that
introduced pacing of display link callbacks. Blocking the display link
callback introduced deadlocks with code that updates the display link
display.
Not pacing the display link may introduce extra view invalidations,
but invalidates are consolidated by the system and won't result in
extra redraws.
Fixes: https://todo.sr.ht/~eliasnaur/gio/370
Signed-off-by: Elias Naur <mail@eliasnaur.com>
It's now possible to directly user pointer.Cursor to add to the ops.
pointer.CursorText.Add(gtx.Ops)
This is an API change. Use pointer.Cursor directly instead of CursorNameOp.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
Add most of the common cursors defined by different systems.
Normalize cursor names to match CSS.
This is API change: some cursor names have changed, and the
underlying type is no longer a string.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
Events such as system.CommandEvent may result in Invalidate being
called. Ensure animation state is properly updated.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Mapping it to key.NameReturn confuses widgets such as Editor that
treats clicks separate from return key presses.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This change solves an issue where an editor key press such as
backspace would apply before the IME had a chance to process it.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Apparently, a Wayland server such as Sway and River may not issue
a frame pacing callback created before a top level configure event.
This change ensures the frame pacing callback is always refreshed
when animating, regardless of the reason for the frame.
Fixes: https://todo.sr.ht/~eliasnaur/gio/364
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This change implements reporting of the caret position from Editor, as well
as Windows, macOS, Android support. As a result, the IME composition window
on Windows and macOS is now positioned correctly.
References: https://todo.sr.ht/~eliasnaur/gio/246
Signed-off-by: Elias Naur <mail@eliasnaur.com>
When a text range in the IME snippet is replaced, the replacement
is discarded if the range don't overlap with the snippet range. However,
the replacement is more relevant than whatever snippet is current.
This change discards the snippet in case of no overlap.
As a bonus, IMEs that leaves the snippet range at [0:0] will have the
snippet track the composing region now.
The FuzzIME test is adjusted to always generate replacements that overlap
Editor content; otherwise the IME snippet and editor state can't be expected
to match.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Elias Naur