Sets up rules to create an APK that is comprised of solely native code. Existing executable targets (like GTests) can then use this to run on Android devices while having access to activities, windows, etc.. This allows for broader test coverage. Basically, anything that needed an ANativeWindow could only be tested in an integration test.
Executables that need access to the native activity must provide an implementation of `NativeActivityMain` that returns a custom subclass of `flutter::NativeActivity`. The `native_activity_apk` reads like an `executable` or `shared_library` target. Just one that packages that executable in an APK.
The APK is built using the Android Tools and does not use Gradle. Creating a new APK after invalidating some code takes ~200ms on my machine. The edit, compile, run cycle for only a tiny bit worse than testing on the host.
Builds on top of this new infrastructure to create a `GTestActivity` that runs an existing test suites. This works really well except the GTest suite logs to `STDOUT` whereas the engine logs to `logcat`. To quickly work around this, a custom test status listener has been wired up. This only displays the test results to logcat today but a similar mechanism can be used to talk to the test runner in the host. I will wire this up in an upcoming patch as there is no hooks into this from CI right now.
Creates an APK variant of the `impeller_toolkit_android_unittests` harness.
This reverts commit 037aa6b4caa6a3a726e67e519324b2c0a1ec274a.
The original cause of the revert was a flake introduced because the unit-test could request a frame from the Choreographer if it ran long enough. The availability checks in the choreographer were inaccurate after we intentionally backed out of using the Callback32 variant on 32 bit platforms.
The new tests didn't catch it because of an unrelated issue. In the first version of the patch for review, the proc table was only supposed to run on API levels 29 and above. When @dnfield requested we also get rid of the NDK helpers, the choreographer and additional utilities were added. But the API level gate in the new test harness wasn't removed. This made the tests be skipped. That gate has been removed entirely now. The error that cause the revert because of flakiness will now be a reliable failure.
Work towards https://github.com/flutter/flutter/issues/145219.
Previously all logging would be silent in the case that `test_foo` uploads a digest that is considered "untriaged", and we'd be entirely reliant on the `flutter-gold` check to pick this up asynchronously.
As part of debugging https://github.com/flutter/flutter/issues/145219 (but probably to keep this code indefinitely, it's not harmful), we now unconditionally log the swallowed failures to `stderr` so they will show up in our LUCI logs.
/cc @gaaclarke @jonahwilliams
Only available on Android device API levels >= 29. Proc table is setup has versioning checks. All handles are type safe. Collection of handles takes into account cleanup tasks (like reparenting surface controls). The proc table contains code duplicated in ndk_helpers and I will remove that in favor of this in a subsequent patch.
Part of https://github.com/flutter/engine/pull/51213 being chopped up.
Previously the DisplayListBuilder would only pass along bounds for a saveLayer when they were supplied by the caller that was building the DisplayList. This would require Impeller to use post-processing of the EntityPass lists to compute them on its own.
DisplayList can now compute those bounds as it builds the DisplayList to save dispatch clients from having to do so on their own. It will also provide an indicator in the case when the caller supplied bounds that ended up being too small to capture all of the content, causing clipping by the layer render target.
Closes https://github.com/flutter/flutter/issues/144365.
Wasn't able to delete as much as I wanted (it's used by the video-rendering code path), but it should fix the problem we're seeing. I expect this change to remove the bottom (Android-rendered) image from our golden files for `ExternalTexturesTests`.
Reverts: flutter/engine#51429
Initiated by: bdero
Reason for reverting: Engine tree breakage
```
+ /b/s/w/ir/cache/builder/src/out/host_release/display_list_region_benchmarks --benchmark_format=json
/b/s/w/ir/cache/builder/src/flutter/testing/benchmark/generate_metrics.sh: line 17: /b/s/w/ir/cache/builder/src/out/host_release/display_list_region_benchmarks: No such file or directory
```
Original PR Author: flar
Reviewed By: {godofredoc}
This change reverts the following previous change:
These benchmark results aren't shown in Skia perf because they were never added to the CI tasks to run.
- Replaced manual `StringBuffer()..writeln('stdout: ...')` with a single
`SkiaGoldProcessError` constructor.
- Updated tests to make sure it's working.
_/cc @dnfield @jonahwilliams FYI only._
Updates the linting script to ban the use of `VERSION_CODES`.
We currently have a mish-mash of using the integers, using `VERSION_CODES`, and even how we import the version codes. This makes it more confusing when doing things like #51070 - I think it is clearer to see `22` than `LOLLIPOP_MR1`.
I'd like to get LGTM (or at least no opinion) from all the requested reviewers here.
This improves performance by avoiding the 2-pass Gaussian blur in more shadow drawing situations.
The new golden also serves as a good reference for how mask blurs are supposed to work alongside various other paint properties such as color filters, color opacity, image filters, and blending.
The top 5 shapes are various RRect cases and are rendering correctly via the new blur style implementation in this patch.
The two bottom rows (the triangles and arcs) are non-rrect paths, so they're falling back to rendering using the 2-pass Gaussian blur. Rendering errors are circled in red below:
* Cases 1, 2, 7, and 9 all appear to rendering fine.
* Cases 3, 4, 5, and 6 all have mask blur styles set to `BlurStyle::kSolid`. After the first clipped overlay has been drawn, subsequent clipped overlays aren't drawing.
* Case 6 is also has the blend mode set to `BlendMode::kExclusion`.
* Cases 8 and 10 are rendering with `BlurStyle::kInner` and `BlurStyle::kOuter` respectfully, but with a blur ImageFilter also set on the paint state. The ImageFilter needs to be applied to the rasterized mask blurred content.
Part of https://github.com/flutter/flutter/issues/144640.
I'll be moving the rest of the pipeline-dependent tests and packaging a shader bundle for flutter_tester once Flutter GPU works against Vulkan/swiftshader.
Testing if `skia_gold_client` and pre-submit/post-submit is playing nice by triggering an update.
Generated via:
```sh
dart ./testing/skia_gold_client/tool/generate.dart
```
Related: https://github.com/flutter/flutter/issues/144352.
Allow users to update dependencies. Examples:
* `et fetch` Fetch dependencies
In the future, `et build` will update dependencies if it detects that they have changed.
Also:
* Updates the status in the README
* Adds instructions on how to run tests
* Fixes `et run`'s description
* Makes the `--verbose` flag global
## Description
This makes the GTK dependency only active if unit tests are enabled, so that it won't create an undesired dependency on embedded platforms.
## Related Issues
- Fixes https://github.com/flutter/flutter/issues/144421
Adds a flag, `--prefix-logs-per-run`, that defaults to `true` on CI
envrionments.
If present, instead of writing the file `${LOGS_DIR}/logcat.txt`, we
write `${LOGS_DIR}/{prefix}.logcat.txt`. I don't really like this
solution, but alternatives are using zip files (requires downloading to
view) or waiting on generic log improvements.
Closes https://github.com/flutter/flutter/issues/144402.
This is a prototype of the [PlatformIsolate
API](https://github.com/flutter/flutter/issues/136314).
**UPDATE (Jan 25):** The PR is ready for review. PTAL.
The `PlatformIsolate` creation flow is:
1. `PlatformIsolate.spawn` running on parent isolate
(platform_isolate.dart)
a. Create `isolateReadyPort`
b. `PlatformIsolateNativeApi::Spawn` (platform_isolate.cc)
c. `DartIsolate::CreatePlatformIsolate` (dart_isolate.cc)
d. Isolate created. Entry point invocation task dispatched to platform
thread
e. `PlatformIsolate.spawn` returns a `Future<Isolate>`
2. On the platform thread, `_platformIsolateMain` is invoked in the
platform isolate
a. Create `entryPointPort`
b. Send `Isolate.current` metadata and `entryPointPort` back to the
parent isolate via `isolateReadyPort`
3. Back in the parent isolate, `isolateReadyPort.handler` is invoked
a. Send the user's `entryPoint` and `message` to the platform isolate
via `entryPointPort`
b. Use received isolate metadata to create a new `Isolate` representing
the platform isolate and complete the `Future<Isolate>`
4. In the platform isolate, `entryPointPort.handler` is invoked
a. Run the user's `entryPoint(message)`
The engine shutdown flow is handled by `PlatformIsolateManager`, which
maintains a set of running platform isolates.
Basically, I take blank images of red/blue/green squares, like this:
... and use Image Magick to annotate it with the current git hash:
```sh
$ drt ./testing/skia_gold_client/tool/generate.dart
Writing annotation "8069cb4ca1" on images in testing/skia_gold_client/tool/source_images and saving them in testing/skia_gold_client/tool/e2e_fixtures.
Writing to testing/skia_gold_client/tool/e2e_fixtures/solid_red_square.png
Writing to testing/skia_gold_client/tool/e2e_fixtures/solid_green_square.png
Writing to testing/skia_gold_client/tool/e2e_fixtures/solid_blue_square.png
Done: wrote 3 image.
```
... then, I upload the digests very similar to how we do it in `scenario_app`.
---
The idea here is to have a way for me to know if Skia gold is working the way I/we expect, independent of a more complicated test suite with various race-y or flake-y conditions. We could also augment it with tests of "dimensions" properties.
The `run` command builds both the host and target engines and then invokes `flutter run` so that it runs the app using the custom engine builds.
It is expected that 'et run' be used inside the directory of a flutter application.
Command line flags passed after `--` will be forwarded to `flutter run`.
Some examples:
- `et run` Build the debug variant and runs the app in that mode.
- `et run -- --profile` Build the profile variant and runs the app in that mode.
- `et run -- --release` Build the release variant and runs the app in that mode.
Also:
- Start a local_engine.json builder definition (it is missing a lot)
- Tweak the test fixture.
- Add a new Matcher to litetest.
- Tweaked the build config linter to only care about duplicate build names within the same builder.
This reverts commit 98485b3f82fec45a6a0ac932c3c20285a9a14e02.
Reverted in https://github.com/flutter/engine/pull/51056 because it
caused failures in the framework. No changes, but framework test has
been updated to use an actual robolectric implementation to avoid those
failures.
This should not land until
https://github.com/flutter/flutter/pull/144348 has landed in the
framework.
@gaaclarke @johnmccutchan fyi
