Bump lower Dart SDK constraints to 3.0 & add class modifiers

[goderbauer]

Part of #118837.

This will allow us to use the latest Dart 3.0 features.

This PR also adds class modifiers where necessary:

• to all classes identified as mixin-able as part of the analysis in http://flutter.dev/go/mixin-modifier
• subclasses of LinkedListEntry (a newly marked base class) to make the analyzer warnings below happy
• slight adjustments to how TapRegionRegistry is used to fix analyzer warnings

  error • The type '_OverlayEntryLocation' must be 'base', 'final' or 'sealed' because the supertype 'LinkedListEntry' is 'base' • packages/flutter/lib/src/widgets/overlay.dart:1641:7 • subtype_of_base_or_final_is_not_base_final_or_sealed
  error • The type '_RenderDeferredLayoutBox' must be 'base', 'final' or 'sealed' because the supertype 'LinkedListEntry' is 'base' • packages/flutter/lib/src/widgets/overlay.dart:1957:7 • subtype_of_base_or_final_is_not_base_final_or_sealed
  error • The type '_ListenerEntry' must be 'base', 'final' or 'sealed' because the supertype 'LinkedListEntry' is 'base' • packages/flutter/lib/src/widgets/scroll_notification_observer.dart:38:7 • subtype_of_base_or_final_is_not_base_final_or_sealed

Blocked on waiting for flutter/engine#40178 to roll into the framework.

[goderbauer]

FYI @jakemac53 @chingjun

[goderbauer]

@LongCatIsLooong Do the changes to overlay.dart look good to you?

@gspencergoog Do the changes in tap_region.dart look good to you to keep TapRegionRegistry as an interface instead of making it mixin-able?

[LongCatIsLooong]

YES! The overlay.dart changes LGTM. Pattern matching and record are still behind the experimental flag right?

[goderbauer]

Pattern matching and record are still behind the experimental flag right?

Once this PR lands we're good to use them in the framework. This PR is still blocked on waiting for Dart v3.0.0-325.0.dev to roll into the engine & framework, though. That Dart version flips the flag on these features.

[jakemac53]

cc @johnniwinther it sounds like we should hold off on landing this until exhaustiveness checking is finished? What is the status and is there an issue we can track?

[goderbauer]

I am already getting a bunch of exhaustive_cases warnings on this PR (see failing checks).

[johnniwinther]

@goderbauer Note that the errors you see currently are from a temporary exhaustiveness algorithm based on flow-analysis. The real algorithm, which is in the process of being enabled, will yield different results.

[johnniwinther]

The CL for enabling the full exhaustiveness checking is https://dart-review.googlesource.com/c/sdk/+/288703

[gspencergoog]

TapRegion changes LGTM.

[goderbauer]

Full exhaustiveness checking is rolling into the framework with #123059.

[christopherfujino]

LGTM

[goderbauer]

To my surprise, with the bump to 3.0 the analyzer complained about a handful of additional places where const should be used, I fixed those up in the last commit.

Edit: Looks like at runtime these new const get flagged as invalid: https://logs.chromium.org/logs/flutter/buildbucket/cr-buildbucket/8786064751393689137/+/u/run_test.dart_for_web_canvaskit_tests_shard_and_subshard_3/test_stdout

[Piinks]

ScrollBehavior changed re: dart-lang/sdk#51802 LGTM. Thank you!

[gspencergoog]

Are these going to show up in any customer code that uses prefer_const_constructors? This lint is turned on as part of flutter_lints, so it might affect a lot of people if so.

[goderbauer]

Yeah, they will see this once they bump the sdk constraints to 3.0 in their pubspecs. I am pushing hard on dart-lang/sdk#51800 to get this fixed before Dart 3 is released to stable.

[gspencergoog]

Yeah, that seems important.

[gspencergoog]