Some people call me the space cowboy
Some people call me the gangster of love
Some people call me Maurice

Ensures we don't regress #6577 on Windows.

Ensures we don't regress https://github.com/flutter/flutter/issues/6577.

Eliminates the assumption that the flutter SDK is installed in a
directory named "flutter" from ticker_expectation.txt.

In getFlutterRoot(), scripts loaded via data: URIs are URI encoded.
getFlutterRoot() scans the contents of the data for the file:// URI path
of the Flutter SDK, which itself is URI-encoded. The end result is that
if the SDK path contains a space, the embedded file:// URI will contain
a %20. When this is encoded in a data: URI, the contents are
URI-encoded, resulting in %2520, since the % is encoded to %25.

This patch decodes the data: URI before extracting the SDK file:// URI.

The Flutter engine now supports package: and file: imports that resolve
to paths on disk that include spaces and other URI-escaped characters.
This patch eliminates the restriction that Dart source paths not include
%20 (or other URI-escaped characters) in their paths in flutter_tool
tests.

Includes:
* Roll topaz to 531e2778d86b213bc663ea51bb0e0dde33d8ac8c (#4657)

This supports SDK/project paths including URI-escaped UTF-8 sequences
(e.g., including spaces, accented characters, etc.).

* More restricted versioning on iOS

* Escape the key base64 and don't echo it back on travis

* Have match not print back the cert url

When uploading, gsutil is guessing wrong about our desired MIME types. This makes it explicit.

Includes:
* Exclude frontend_server from the license crawl (https://github.com/flutter/engine/pull/4645)
* Make native wrapper classes non-abstract (https://github.com/flutter/engine/pull/4607)
* Restrict clang-format to C, C++, Obj-C, Obj-C++ (https://github.com/flutter/engine/pull/4654)

This reverts commit 74ddda51.

* Reduce noise in xcodebuild stdout

* Reduce output from cleaning, from non failing xcode outputs and from script

* Check xcodebuild clean's exit code

* Revert "Reduce noise in xcodebuild stdout"

This reverts commit 222a26f55f2a3fc7b115e8a65e55c9be36f241bd.

Includes:
* Roll topaz to 08a3394395036a2bb9b556f5b0eb8f365d2c0fa5 (flutter/engine#4650)
* [fuchsia] Add missing AOT entry points for FromFileResult and MapResult. (flutter/engine#4649)
* [fuchsia] Fix compile-time error introduced in 58e6c23. (flutter/engine#4651)
* Roll topaz to 046105efd225ed61c94e65dbd4a528256a3a34a9 (flutter/engine#4652)

Topaz rolls enable:
* Support for Flutter SDK, project paths including spaces (flutter/engine#4650)
* Support for Flutter SDK, project paths including arbitrary UTF-8 sequences (flutter/engine#4652)

* Roll engine to pre-dart roll

* Roll engine to pick up updated dart

* Apply Map changes

* Move to dev.22

* Fix some analysis issues

* Silent analyzer

* More consts

* More const massaging

* Yet more const massaging

* Yet more const massaging

* Use nonconst()

We need to pull in Dart 2 compliant mockito

Fixes #14532

This allows const GlobalObjectKey(0) to be concisely formatted as
[GlobalObjectKey int#0000] in both Dart 2 and Dart 1 modes.

Without this change it would be formatted as
[GlobalObjectKey<State<StatefulWidget>> int#0000] because in Dart 2
types are instantiated to bounds.

In addition to retaining general readability this also fixes few
tests that rely on this short string representation (see
test/widgets/global_keys_duplicated_test.dart).

The current ARM64 back end generates code that crashes on some devices,
including Pixel phones.  With this change, the android-arm64 target will
not be used by "flutter run" unless explicitly requested.
This is intended as a workaround until we can roll out an engine with the
required Dart VM fix.

See https://github.com/flutter/flutter/issues/14454

By default flutter run will build a 64-bit APK if the attached Android device
is 64-bit.  Specifying --target-platform=android-arm will deploy a 32-bit APK
to a 64-bit device.

Fixes https://github.com/flutter/flutter/issues/14526

Channels don't preserve Map and List type arguments: Map<String, Object>
arrives as Map<dynamic, dynamic> to the receiver.

In Dart 2 type system dynamic no longer serves as bottom type so
Map<dynamic, dynamic> can't be assign to a variable of type
Map<String, dynamic>.

Issue #14556

In Dart 2 mode we throw different exception: compiler generates
throw of a compilation error instead of runtime throwing
noSuchMethod.

--preview-dart-2.

Sometimes fails writing to stdout as flutter terminates. Marking as
flaky until resolved.

Use mockito in the Dart 2 compliant way:

* use typed(...) in appropriate places to wrap matchers like any;
* don't pass matcher directly into the calls - use `argThat(matcher)`.

Also rename paintColorMatcher to hasColor to match how other similar
functions are named and how the code reads.

* Use explicit types for onClick handler in diagnostics_test.

This test formats it to string and expects exact match.

Before this change onClick had type () => dynamic in Dart 1 and () => Null in
Dart 2.

* Fix strong mode issues in packages/flutter/test/material/dropdown_test.dart

Pass explicit type argument to renderObjectList(...) in expressions like

     List<RenderBox> l = tester.renderObjectList(...).toList();

Dart 2 mode inference is not strong enough to infer RenderBox as a type argument
fto renderObjectList and passes RenderObject instead, which later leads to
runtime check failures (because List<RenderObject> is not List<RenderBox>).

* Fix strong mode issues in packages/flutter/test/material/theme_test.dart.

Pass type argument explicity to widgetList. Dart 2 type inference can't infer
it by itself.

* Fix strong mode issue packages/flutter/test/widgets/unique_widget_test.dart

Pass correct type argument to GlobalKey.

* Fix type annotation in packages/flutter/test/material/app_test.dart.

pushNamed returns Future<Object> not Future<String>.

This reverts commit 1c236d52.

Fixes https://github.com/flutter/flutter/issues/14492

I wanted to run a "real command" as part of the test, but in the interest of fixing the build, I'm just removing the test.

There were some problems I introduced with the last PR for this. It passed the test, but failed in practice.

This adds tests for those failure cases, adds a "--help" and fixes the test so that it doesn't try and actually download MinGit as part of the Windows test.

I added package:platform as a dependency, so I did a force upgrade on the packages.

Also, re-enabling 'create package' in the cache warming code, now that #14448 is fixed.