* Fixing invalid state bug for text editing. Flutter Framework was sending editing state selection base and extent as -1. Since -1 is an invalid value for a dom element selection it was not applied to the last editing state. Now if the base or offset is sent as negative value, web engine will set 0 to the selection range.
* Addressing PR comments.
This adds an isFocusable to SemanticsFlag so that the framework can tell the engine what semantics nodes are allowed to be focused, which will affect what platform flags are applied to the semantics information.
This flag is not yet in use by the frame
This reverts commit 639cc113f0b2ccf9fcf69ded7960d41d0b611f80.
Fixes https://github.com/flutter/flutter/issues/41394 and other
related correctness issues.
TBR: @arbreng @jason-simmons @mehmetf
* add PersistedPlatformView attribute update handling to enable resizing
* update size of root element created by a PlatformView
* improved comments
* enforce effective size of PlatformView surface
* updated formating of platform_view.dart
* stop storing root element of PlatformView in its Surface
When the PlatformViemSurface adopts the DOM elements from a previous PlatformViewSurface the the stored value will lost.
* move setting overflow property to createElement
* reflect selection changes in Firefox. With this change if the keyboard arrow keys to move the cursor the selection change is synced to Flutter Framework
* Addresing PR comments. Renaming. Adding the domelement to SelectionChangeDetection constructor.
* add initial value to selection start/end
add initial value to selection start/end
On Fuchsia, add a build flag for compositing OpacityLayers using the system
compositor vs Skia, which exposes a fastpath for opacity via Scenic.
This will only work under certain circumstances, in particular nested
OpacityLayers will not render correctly!
On Fuchsia, add a build flag for compositing PhysicalShapeLayers using
the system compositor vs Skia. Set to off by default, which restores
performant shadows on Fuchsia.
Remove the opacity exposed from ChildView, as that was added mistakenly.
Finally, we centralize the logic for switching between the
system-composited and in-process-composited paths inside of
ContainerLayer. We also centralize the logic for computing elevation
there. This allows the removal of many OS_FUCHSIA-specific code-paths.
Test: Ran workstation on Fuchsia; benchmarked before and after
Bug: 23711
Bug: 24163
* Fix broken tests
* Interpret negative radii as 0 in recording_canvas drawDRRect
This allows drawing DRRects that may have some negative values on
its corners (caused by deflating a RRect with some non-round corners,
for example)
See added unit test for an example of the above.
Fixes https://github.com/flutter/flutter/issues/40728
* enabling spellcheck for text editing
* Finalize location changes in IOS. IOS is ready for spellcheck now. Disable spellcheck since we realized some tag/autocomplete transfer from Framework is necessary for complete spellcheck implementation.
* Remove todo since this PR fixes the updated location issue for IOS too. All browsers are supported now.
* Adding boolean flags to make conditions more readable. Fixing typos.
* Fixing unit tests. Addressing github PR comments.
* Import all golden files and tests from internal repo.
* Adapt test files to new screenshot API, and tweak some settings.
(Check PR to see individual changes to each file)
Fixes https://github.com/flutter/flutter/issues/40975
New features for golden tests (for web):
* Height/width of headless browser instance extracted to constants
* Remove border from the host iframe
* Added 'region' to matchGoldenFile, so screenshots can capture only
a subset of the viewport
* Added image-format-awareness (png) to screenshot comparison, so it
doesn't only look at raw bytes
* When a test fails, output a diff image alongside the unexpected
output.
The earlier design speculated that embedders could affect the same
transformations on the layers post engine compositor presentation but before
final composition.
However, the linked issue points out that this design is not suitable for use
with hardware overlay planes. When rendering to the same, to affect the
transformation before composition, embedders would have to render to an
off-screen render target and then apply the transformation before presentation.
This patch negates the need for that off-screen render pass.
To be clear, the previous architecture is still fully viable. Embedders still
have full control over layer transformations before composition. This is an
optimization for the hardware overlay planes use-case.
Fixes b/139758641
* Improve the CanvasKit backend for Flutter Web
- Improve font handling by trying to load a "normal" font face
instead of using the first face matching the family.
- Implement Vertices and drawVertices
* Add license header to vertices.dart
* Remove unused 'encodedPositions'
* Delete commented old code. Don't use Skia by default
* Add `vertices.dart` to licenses file
1. Various functionalities offered by this tool are now organized into commands (e.g. `felt test`, `felt check-licenses`).
2. The felt tool can now be invoked from anywhere, not necessarily from the web_ui directory.
3. This new structure helps us scale better as we add more commands (e.g. soon a `build/watch` command is coming).
* enabling spellcheck for text editing
* Fixing errors in the input box, which arrises when the fontweight is null.
* Rollback the spellcheck change. Still not tested in safari mobile.
* Carrying the assignment of fontweight from index to EditingStyle constructor. This reduce # of null checks to one.
- Add a custom `PlatformPlugin` that spins up a server waiting for test to request a screenshot.
- When a screenshot is requested the plugin talks to Chrome via the debug port, captures a screenshot, and compares it with a golden file
- This PR also adds proper CLI for `dev/test.dart` with an `ArgParser` for future extension into a proper developer tool for the team.
- As a first couple of features, it adds `--debug` option to launch Chrome in debug mode, a `--target` option to choose a single test to run rather than all tests, and `--shard` option to choose a subset of tests to run.
Limitations:
- While screenshot test will run on Cirrus, they are configured to not fail on Cirrus. Need to solve Chrome version skew.
- Stack maps do not work yet (you get stacks, but they are nonsensical)
- When requesting a single test, build_runner builds all tests anyway
- Nothing but desktop Chrome is supported
- Nothing but Linux is supported
- There's no Chrome version pinning; currently assumed stable Chrome channel
