The `isRTLAtPosition` method had a bug, it used `NSInteger max = [_selectionRects count]` instead of `NSInteger max = [_selectionRects count] - 1`. But I realized we don't even need the function any more, it was used in a few places in previous iterations of #36643, but in the only place remaining, we actually already have the selection rect and don't need to search for it by position.
Btw as an explanation of the crash, I guess there is some mismatch between code point and character count somewhere. UIKit was asking for `caretRectForPosition:2` when we only had 1 character. This could have only crashed when floating cursor selection was used, but actually when switching to CJK keyboard, UIKit turns out to use `caretRectForPosition` to calculate something about the composing rect.
Fixes https://github.com/flutter/flutter/issues/128031
Fixes a few issues with the Linux text editing delta implementation that was causing some issues when composing CJK text.
`im_preedit_changed_cb` dispatches a delta with the delta range set as the current composing region causing a crash when composing because the new composing region range might not yet exist in the `oldText` that the delta is applied to. We should instead send the composing region before the change to the text as that will be the correct range that is being modified in the `oldText`.
`im_preedit_end_cb` dispatches a delta with an empty `oldText`, a not empty `deltaText`, and a delta range of `(-1,-1)` causing a crash when composing ends because the delta range will never be within the range of the text. We should instead send the current text value as the `oldText` and not send a `deltaRange` or `deltaText` since a `(-1,-1)` range essentially means nothing in the text changed.
`im_commit_cb` does not account for the case when we were previously composing when the text was committed. This causes the dispatched delta to have a delta range that always inserts at the current the selection, when it should instead replace the previous composing range. We should account for the case when we are previously composing before committing the text, and dispatch a delta with the previous composing region.
Fixes https://github.com/flutter/flutter/issues/113909
Sometimes the time point created late will less than the time point create earlier since the CLOCK SOURCE has changed.
We must make sure the CLOCK SOURCE (fml::TimePoint) is always keep the same while we are testing.
Corrects the test case is needed before we fix the problem completely.
## Description
This PR fixes an issue related to text input plugin, on IOS, sending a wrong action string to the engine for `FlutterTextInputActionContinue`.
## Related Issue
Fixes https://github.com/flutter/flutter/issues/126922
## Tests
Adds 1 test.
Previously, when creating native platform views on macOS, we ignored any parameters passed via the framework side "params" argument in the "create" method call, and instead always passed a nil value to the FlutterPlatformViewFactory. This made it impossible for users of macOS platform views to pass constructor arguments to the NSView subclass implementing the platform view.
We now decode the arguments data using the codec specified by the `FlutterPlatformViewFactory` and pass them through to the `[FlutterPlatformViewFactory createWithIdentifier:arguments:]` method where the platform view author can make use of them.
Fixes: https://github.com/flutter/flutter/issues/124723
This is a part of the broader macOS platform view support effort: https://github.com/flutter/flutter/issues/41722
[C++, Objective-C, Java style guides]: https://github.com/flutter/engine/blob/main/CONTRIBUTING.md#style
Normally when platform view is clipped to a simple rect we rely on
clipping to container layer bounds. However when the clip rect is
rotated the container layer is expanded accordingly and clipping to path
must be used instead.
Fixes https://github.com/flutter/flutter/issues/128175
For consistency with Flutter (and all other platforms), Flutter views in the macOS embedder set `isFlipped` to ensure a co-ordinate system with the origin in the top-left, with y co-ordinates increasing towards the bottom edge of the view.
Previously, we were using a stock NSView as the container, which uses a bottom-left origin by default. Instead we extract the PlatformView container view as its own class with `isFlipped` true.
This doesn't correct the issue of the view anchorpoint/position but does correct rotation direction.
This also applies the transform back to origin prior to other transforms when adjusting the platformview position rather than after.
Issue: https://github.com/flutter/flutter/issues/124490
[C++, Objective-C, Java style guides]: https://github.com/flutter/engine/blob/main/CONTRIBUTING.md#style
On some Samsung devices Flutter with spell-check enabled will crash when typing/moving near the ">" character.
Stack trace is
```
Fatal Exception: java.lang.NullPointerException: Attempt to invoke virtual method 'int android.view.textservice.SentenceSuggestionsInfo.getSuggestionsCount()' on a null object reference
at io.flutter.plugin.editing.SpellCheckPlugin.onGetSentenceSuggestions(SpellCheckPlugin.java:26)
at android.view.textservice.SpellCheckerSession.lambda$handleOnGetSentenceSuggestionsMultiple$1$android-view-textservice-SpellCheckerSession(SpellCheckerSession.java:224)
at android.view.textservice.SpellCheckerSession$$ExternalSyntheticLambda0.run(:4)
at android.os.Handler.handleCallback(Handler.java:942)
at android.os.Handler.dispatchMessage(Handler.java:99)
at android.os.Looper.loopOnce(Looper.java:226)
at android.os.Looper.loop(Looper.java:313)
at android.app.ActivityThread.main(ActivityThread.java:8747)
at java.lang.reflect.Method.invoke(Method.java)
at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:571)
at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:1067)
```
This test sometimes caused assertion failures when running on Windows with ANGLE. The process may be unable to safely clean up global objects if GL commands are pending when the test exits.
The original PR caused some golden test failures down the line, likely due to bad analysis of when the combined BlendMode and color would result in a NOP situation.
This PR adds tests that go through every BlandMode and pair it with a variety of colors and Color/ImageFilters to verify that the operations are only omitted when they actually produce no change in the output. It also checks the validity of the "modifies_transparent_black" property of DisplayLists which can be used in place of the current CanvasSpy/DlOpSpy classes.
The description from the [previous PR](https://github.com/flutter/engine/pull/41463) updated with the new name of the DL property:
---------------------------------
This optimization avoids recording unnecessary render operations that will not affect the output and also eliminates the need for "draw detection" mechanisms like `DlOpSpy` and `CanvasSpy` by remembering if any non-transparent operations were included. The `DlOpSpy` unit tests were updated to check if the results from that object match the new `DisplayList::modifies_transparent_black()` method.
Fixes https://github.com/flutter/flutter/issues/125338
In addition, this change will unblock some other Issues:
- https://github.com/flutter/flutter/issues/125318
- https://github.com/flutter/flutter/issues/125403
Empty SkRect needs to be created with `SkRect::MakeEmpty()` otherwise the default value is undefined (and possibly garbage).
*If you had to change anything in the [flutter/tests] repo, include a link to the migration guide as per the [breaking change policy].*
[C++, Objective-C, Java style guides]: https://github.com/flutter/engine/blob/main/CONTRIBUTING.md#style
## Description
This reverts commit 879917b to re-land #40542 to fix the lint warning which caused the original PR to be reverted (the lint warnings were turned on after the original was landed, which is why it wasn't caught earlier).
This extracts an extra method from `[FlutterMutatorView applyFlutterLayer:]` to perform the final PlatformViewContainer and PlatformView re-nesting in the view stack, then apply transforms, and finally apply the outer axis-aligned master clip rect around the whole thing.
This is a refactoring for readability (primarily clearer parameterisation) and applies no semantic changes to the code. This is the last in a series of refactoring applied to this method.
Issue: https://github.com/flutter/flutter/issues/124490
[C++, Objective-C, Java style guides]: https://github.com/flutter/engine/blob/main/CONTRIBUTING.md#style
Fixes https://github.com/flutter/flutter/issues/126878
This disables device private upload on iOS when backgrounded, and disables mipmap generation when backgrounded.
We don't have a good way to test the core problem in this repo because it only reproduces on physical iOS hardware - simulators don't really care if you do this stuff in the background.
I'll write a devicelab test in the framework to capture this. In the mean time it can be reviewed.
We could consider making the IOManager a shared_ptr instead of having an fml::WeakPtr and that'd clean up some of the extra arguments to engine construction - or we could consider vending the sync switch from impeller::Context unconditionally, but it's pretty iOS specific...
Wide gamut is not supported on the iOS simulator, and the iOS platform's FlutterView.isWideGamutSupported checked for this. However, the ImageDecoder decides whether to enable wide gamut based on the engine's settings, so the flag in the settings needs to be modified for the simulator.
