* Fix NUnit test context not being properly set
* Add failing tests
* Capture ExecutionContext to keep async locals
* Remove explicit EstablishExecutionEnvironment call, as it was a bad idea
* Make ExecutionContext usage disabled by default, and only enabled for NUnit
* Add a failing test for #15712.
* Validate StartU at the start of a measure pass.
If any container U size has changed since the last layout pass then `StartU` must be considered unstable as the average container height will have changed.
* Correctly position focused element.
If the focused element has been moved outside the visible viewport due to a realized container size change, then we need to ensure it's positioned correctly.
* We can skip check if StartU is already unstable.
* Don't invalidate virt. panels more than necessary.
* Add another virt panel test.
And revert the expected results for another test to the way they were at the beginning of this PR.
* Tweak container size estimation.
Use the desired size of _measured_ containers instead of the bounds: a layout pass may not had completed on the containers yet, so the bounds may not be up-to-date. Was easier to move the estimation methods out of `RealizedStackElements` and into `VirtualizingStackPanel` itself in order to do this, and arguably makes more sense.
* Add failing test for #16113.
* Convert delegate to ICommand in style setter.
When compiling a binding to e.g. `Button.Command` in a style `Setter`, we were not converting `XamlIlClrMethodPathElementNode` to `XamlIlClrMethodAsCommandPathElementNode` as we were only testing whether the property that the binding was being assigned to is an `ICommand`.
If we detect that we're assigning the binding to a `Setter.Value` then we need to look in the `Setter.Property` to see check whether the property is an `ICommand` too.
Fixes#16113
* Add tests for binding negation operator.
Expected results come from 11.0.x branch.
* Unset and null need to be distinct.
Also remove `ExpressionNode.Reset` as it was doing unneeded stuff and should just be the same as `SetSource(AvaloniaProperty.UnsetValue, null);`.
* Make ExpressionNode.OnSourceChanged accept null.
The previous design assumed that a source of `null` was an invalid input to all types of expression nodes. It turned out that there was a single exception to this rule: the `!` operator can in fact operate on a null value. With this new design we instead have to explicitly check for a null value in every override of `OnSourceChanged ` except in `LogicalNotNode`.
Due to this change, we also have to distinguish between `null` and `(unset)` in `ExpressionNode.SetSource` as well.
Fixes#16071
* Fix comment.
* Add failing KeySpline tests
* Fix Animator for progress values less than zero
---------
Co-authored-by: Jumar Macato <16554748+jmacato@users.noreply.github.com>
* Added failing test for #15201.
* Handle nested BindingNotifications.
When #13970 was written, [a check](https://github.com/AvaloniaUI/Avalonia/pull/13970/files#diff-cfb25a491b9452e1815aa2c0d71465aaf81e99792a88a04a1a2ed572fd1930ffR60) was added to ensure that nested `BindingNotification`s didn't happen, and the refactor was written with the assumption that they wouldn't happen.
The problem is that they _do_ happen: when a source object implements both `INotifyDataErrorInfo` and had data annotations, then the nested data validation plugins would each wrap the value coming from the previous plugin in a new `BindingNotification`, resulting in nested `BindingNotifications`.
This adds support for nested binding notifications back in - even though IMO nesting binding notifications is a bug, if we're doing it and we previously supported it then we should continue to support it.
Fixes#15201
* Added failing tests for #15081.
* Provide target property in BindingExpression ctor.
Usually it is not necessary to provide the target property when creating a `BindingExpression` because the property will be assigned when the binding expression is attached to the target in `BindingExpressionBase.Attach`.
This is however one case where `Attach` is not called: when the obsolete `binding.Initiate` method is called and then an observable is read from the `InstancedBinding` without the binding actually being attached to the target object. In this case, prior to the binding refactor in #13970 the value produced by the observable was still converted to the target type. After #13970, because the target property (and hence the target type) is not yet set, the conversion is to the target type is no longer done.
`DataGrid` uses this obsolete method when editing cells, causing #15081. Ideally we'd fix that in `DataGrid` but I'm not happy making this change so close to 11.1, so instead fix this use-case to behave as before.
Fixes#15081
By default disable the `UseSaveLayerRootClip` option: this will re-enable subpixel rendering, fixing #15015 but causing #14270 to reappear.
Until the required API is added to SkiaSharp you have to choose one or the other :(
* Added support for Compatibility profile in Wgl and Glx code
* Fixes after code review
---------
Co-authored-by: Olivier DALET <olivier.dalet@addupsolutions.com>
Avoid race condition where a dispatcher timer callback exectues right after we stopped the timer
Fix not closing a tooltip when its pointer exit event is the last input sent to Avalonia
* test: CommandParameter does not change between CanExecute and Execute
* feat: CommandParameter does not change between CanExecute and Execute
* test: update
* add isStroked overload for IGeometryContext
* move IGeometryContextEx
* fix Fill Path line
* remove added whitespaces
* close a figure to non-stroke lines with a line
* add geometry tests
* add wpf test files
* Added a test for tracking path closure with missing strokes for various line caps/joins
* add IsStroke overload for other segments
* update tests
* Skip line join test for closed geometry with holes for now
---------
Co-authored-by: Nikita Tsukanov <keks9n@gmail.com>
* use tap size as default size for scrolling start. reset IsGestureRecognitionSkipped when pointer is released
* use static default constant for scroll distance
* fix typo
* Ported the old Vulkan PR
* chop-chop
* Support for external objects in vulkan backend
* Fixed structure type
* Removed debug code
* sln fix
* Don't force vulkan on windows
* Fixed ShowGridLines = false crash
Fixed "System.InvalidOperationException: 'The control GridLinesRenderer already has a visual parent Grid (Name = grid) while trying to add it as a child of Grid (Name = grid).'" when set ShowGridLines = false and resize.
* Fixed grid lines renderer (fixes ShowGridLines binding)
When changing ShowGridLines property GridLinesRenderer is not updating (create/remove/set bounds). Need to call InvalidateArrange -> ArrangeOverride -> EnsureGridLinesRenderer.
* Fix native menu bar tooltip
* Make ToolTip tests more reliable
* Increase MacOS_Native_Menu_Has_ToolTip_If_Defined sleep delay
* Disable MacOS_Native_Menu_Has_ToolTip_If_Defined, as it fails way too often
Max Katz
Steven Kirk
Jumar Macato
Benedikt Stebner
Nikita Tsukanov
Emmanuel Hansen
Julien Lebosquain
József Molnár
zabolotnev
Bartosz Korczyński
afunc
ijklam
odalet
Tom Edwards
Stephen Monaco
Steve
workgroupengineering
Lubomir Tetak
zii-dmg
dbriard
satanikia1300
Alex
Dawei Huang