When Focus.unfocus is called, the caller usually just thinks about wanting to remove focus from the node, but really, unfocus is a request to automatically pass the focus to another (hopefully useful) node.
This PR removes the focusPrevious flag from unfocus, and replaces it with a disposition enum that indicates where the focus should go from here.
The other value of the UnfocusDisposition enum is UnfocusDisposition.scope.
UnfocusDisposition.previouslyFocusedChild is closest to what focusPrevious used to do: focus the nearest enclosing scope and use its focusedChild field to walk down the tree, finding the leaf focusedChild. This PR modifies it slightly so that it walks up to the nearest focusable enclosing scope before trying to focus the children. This change addresses #48903
A new mode: UnfocusDisposition.scope will focus the nearest focusable enclosing scope of this node without trying to use the FocusScopeNode.focusedChild value to descend to the leaf focused child. This is useful as a default for both text field finalization and for what happens when canRequestFocus is set to false. It allows the scope to stay focused so that nextFocus/previousFocus still work as expected, but removes the focus from primary focus.
In addition to those changes, unfocus called on a FocuScope that wasn't the primary focus used to unfocus the primary focus instead. I removed that behavior, since it was buggy: if the primary focus was inside of a child scope, and you called unfocus on the parent scope, then the child scope could have focused another of its children instead, leaving the scope that you called unfocus on with hasFocus returning true still. If you want to remove the focus from the primary focus instead of the scope, that's easy enough to do: just call primaryFocus.unfocus().
Fixes#48903
This implements VisualDensity changes for text fields*. By default, the layout of the text field does not change.
If the ThemeData.visualDensity is set to a value other than zero, then the density of the UI will increase or decrease. See the VisualDensity docs for more information.
(*In reality, the changes are on the InputDecorator class, not on the text field.)
I also fixed a problem that I think I found with _Decoration where it doesn't compare isDense or isCollapsed as part of its operator==.
This fixes a problem where if you press "Shift" and then "A", then release "Shift" and then "a", then the "A" key will be "stuck" on because the logical key for the key down message is different (capital "A") from the logical key for the key up message (lowercase "a").
This PR changes the pressed keys logic so that it uses the physical key to add/remove keys from the list of pressed keys, but keeps the associated logical key.
This does mean that after the "Shift" key goes up, the pressed keys contains a capital "A" and it doesn't switch to be a lowercase "a", but there isn't currently any mechanism we can use to do that remapping. This is far less surprising than the current behavior, but is still not quite correct.
I fixed the event simulation code to take a physicalKey so that it could be matched with the logical key, but the event simulation code isn't up to the task, since it can only simulate keys that appear in the key maps. The new platform key event design should fix that (added TODOs).
This PR is a follow up of https://github.com/flutter/flutter/pull/42640
Some changes of that PR includes redundant changes using FlutterError.fromParts constructor even though it's not necessary.
Some minor changes are:
- Remove one unnecessary todo
- Fix indent consistencies
* Adds an opt-in flag to fix floating snackbar's offset when no floating action button is present. This flag will be removed once the migration for the fix is complete.
Co-authored-by: filaps <filip1997.28@mail.ru>
This re-lands #49235 with the addition of includeSemantics flag on the Focus widget so that the FocusTraversalGroup can create a Focus widget without affecting the semantics tree.
The FocusTraversalGroup uses the Focus widget to create a grouping of descendants for traversal, but doesn't actually participate in focus (canRequestFocus is always false), so we don't want it to add a Semantics widget in that case, since that can cause semantics changes. The canRequestFocus attribute can also be used when a widget is disabled, so we do sometimes want to include Semantics even if that is false, but not in the case where it is always false, as for FocusTraversalGroup.
- Added a test to make sure that FocusTraversalGroup doesn't add any semantics information.
