This reverts commit 92ef2b9ce1d4e54d5efb3e6a962e9e724dc7efeb.
This requires either runApp() or
WidgetsFlutterBinding.ensureInitialized() to have been called before
using any MethodChannels. Plugins broadly rely on MethodChannels and
right now there's no general requirements that they be constructed
within the runApp call, so the ecosystem breakages from this are broader
than originally thought. Reverting for now.
This relands https://github.com/flutter/flutter/pull/35297
The followings have been done to fix the broken tests:
1. Add `didSendFirstFrameRasterizedEvent` extension and its tests
2. Wait for `didSendFirstFrameRasterizedEvent` instead of
`didSendFirstFrameEvent` during start up tests
3. Mark missed (probably newly added) start up tests as flaky
* Deprecates `BinaryMessages` in favor of a default instance of `BinaryMessenger`, called `defaultBinaryMessenger`
* Platform channels use the `defaultBinaryMessenger` for their binaryMessenger default argument.
* some space formattings
* always use blocks in if-else if a block is used
* format spaces in for and while
* allow multiline if conditions
* fix missing space
* add trailing commas on list/map/parameters
* add trailing commas on Invocation with nb of arg>1
* add commas for widget containing widgets
* add trailing commas if instantiation contains trailing comma
* revert bad change
Service extensions can only be activated in debug or profile mode, their code should never be included in release mode. This PR adds guards around all service extension registration calls that enable Dart's tree shaker to remove the extension's code in release mode, which reduces our binary size:
Android Snapshot (uncompressed): minus 127,384 Bytes (-124.40KB)
APK (compressed): minus 38,136 Bytes (-37.24KB)
iOS Snapshot (App.framework, uncompressed): 264,304 Bytes(-258.10KB)
For details: https://docs.google.com/document/d/13JlgvliCn5sWwT2K2SfDwD1NhEfxpJH9DCf22gZZru8/edit
**Benchmark Regressions:** This PR may cause benchmarks to regress because it may change the timing of GC. If you notice a benchmark regression **please note down the exact set of benchmarks that regressed on this PR** and then feel free to revert. I will follow-up with a PR that forces a GC before the effected benchmarks run to get a clean baseline before re-applying this PR.
Add back screenshot extension that was temporarily disabled.
Add matchesReferenceImage matcher to test that the screenshot extension
generates equivalent images to InspectorService.instance.screenshot.
Enable taking screenshots of arbitrary RenderObjects from a running application from within the inspector.
Key functionality is in the added _ScreenshotPaintingContext class.
* Upgrade everything except matcher.
* Roll matcher (and test)
* Adjust tests that depend on flutter:test directly to depend on a shim
* Require use of package:test shim and remove other references to package:test
New a11y traversal:
- sort direct sibling SemanticsNodes only
- use new sorting algorithm
- implement RTL
- test semantics in traversal order by default
- add AppBar traversal test
- breaking: remove nextNodeId/previousNodeId from the framework
- breaking: remove DebugSemanticsDumpOrder.geometricOrder
* Support exposing the InspectorService over the Flutter
service extension protocol as well as the observatory protocol.
We will probably remove most of the observatory protocol support once a
couple versions of the Flutter IntelliJ plugin have shipped that use the
Flutter service extension protocol. The only reason to continue supporting
the observatory protocol is it will allow using the inspector when paused
at a breakpoint.
This adds an API for defining the semantic node traversal order.
It adds a sortOrder argument to the Semantics widget, which is a class that can define a list of sort keys to sort on. The keys are sorted globally so that an order that doesn't have to do with the current widget hierarchy may be defined.
It also adds a shortcut sortKey argument to the Semantics widget that simply sets the sortOrder to just contain that key.
The platform side (flutter/engine#4540) gets an additional member in the SemanticsData object that is an integer describing where in the overall order each semantics node belongs. There is an associated engine-side change that takes this integer and uses it to order widgets for the platform's accessibility services.
This allows the scheduler library to depend on the services library
and the painting library to depend on the scheduler library without
the services library having to depend on the scheduler library.
While I was at it I also cleaned up some of the binding logic: the
licenses logic can now be overridden (and the test library does so),
and the image cache can now be overridden as well.
Bare bones widget inspector support.
Toggle the widget inspector from the flutter tool by pressing 'i'.
When the widget inspector is select mode:
Pointer down to to inspect a widget.
Pointer click to finalize selection of a widget. You can now interact
with the application as you normally would but with the inspected widget
highlighted.
Click the inspect icon in bottom left corner of screen to reactivate
select mode.
It was previously possible for event dispatch to occurr during the
brief window where the tree had been marked dirty but before it had
been relaid out by reassemble, which would cause assertions to fire if
someone did a hot reload while touching the device.
* Revert "Revert "Change all ocurrences of '$runtimeType#$hashCode' to use the idAndType method. (#10871)" (#10880)"
This reverts commit ceb814aa647ecd310794a72a7a5c28820ec57a25.
* Change all instances of '$runtimeType#$hashCode' to use the describeIdentity method.
The describeIdentity method generates a shorter description with a consistent length
consisting of the runtime type and the a 5 hex character long truncated
version of the hash code.
This splits the frame pipeline into two, beginFrame and drawFrame.
As part of making this change I added some debugging hooks that helped
debug the issues that came up:
* I added debugPrintScheduleFrameStacks which prints a stack whenever
a frame is actually scheduled, so you can see why frames are being
scheduled.
* I added some toString output to EditableText and RawKeyboardListener.
* I added a scheduler_tester.dart library for scheduler library tests.
* I changed the test framework to flush microtasks before pumping.
* Some asserts that had the old string literal form were replaced by
asserts with messages.
I also fixed a few subtle bugs that this uncovered:
* setState() now calls `ensureVisualUpdate`, rather than
`scheduleFrame`. This means that calling it from an
AnimationController callback does not actually schedule an extra
redundant frame as it used to.
* I corrected some documentation.
