There are two issues in the previous implementation:
1. `_populateDevices` will return the devices from `deviceNotifier` if it had been initialized, assuming that once it's initialized, it has been properly populated. That assumption is not true because calling getters like `onAdded` would initialize `deviceNotifier` without populating it.
2. `deviceNotifier` instance might be replaced in some cases, causing `onAdded` subscribers to lose any future updates.
To fix (1), this commit added the `isPopulated` field in `deviceNotifier` as a more accurate flag to determine if we need to populate it.
To fix (2), this commit made `deviceNotifier` a final member in `PolingDeviceDiscovery`.
* Adds support for `flutter test --wasm`.
* The test compilation flow is a bit different now, so that it supports compilers other than DDC. Specifically, when we run a set of unit tests, we generate a "switchboard" main function that imports each unit test and runs the main function for a specific one based off of a value set by the JS bootstrapping code. This way, there is one compile step and the same compile output is invoked for each unit test file.
* Also, removes all references to `dart:html` from flutter/flutter.
* Adds CI steps for running the framework unit tests with dart2wasm+skwasm
* These steps are marked as `bringup: true`, so we don't know what kind of failures they will result in. Any failures they have will not block the tree at all yet while we're still in `bringup: true`. Once this PR is merged, I plan on looking at any failures and either fixing them or disabling them so we can get these CI steps running on presubmit.
This fixes https://github.com/flutter/flutter/issues/126692
Pre work for https://github.com/flutter/engine/pull/51229. Removes a lot of code referencing v1 of the android embedding, though not necessarily all of it (I may have missed some, it is hard to know).
Will hopefully make landing that PR less painful (or maybe painless?)
A ProxiedDevice has two port forwarder, one to forward remote host port and another to forward remote device port. During an attach workflow, the port was forwarded by the latter. Update ProxiedDDS to make sure that it works in such case.
This manually rolls pub packages and updates some calls to use updated APIs that use `Uri` instead of file paths (since macro-generated sources don't exist as files on disk).
This makes several changes to flutter web app bootstrapping.
* The build now produces a `flutter_bootstrap.js` file.
* By default, this file does the basic streamlined startup of a flutter app with the service worker settings and no user configuration.
* The user can also put a `flutter_bootstrap.js` file in the `web` subdirectory in the project directory which can have whatever custom bootstrapping logic they'd like to write instead. This file is also templated, and can use any of the tokens that can be used with the `index.html` (with the exception of `{{flutter_bootstrap_js}}`, see below).
* Introduced a few new templating tokens for `index.html`:
* `{{flutter_js}}` => inlines the entirety of `flutter.js`
* `{{flutter_service_worker_version}}` => replaced directly by the service worker version. This can be used instead of the script that sets the `serviceWorkerVersion` local variable that we used to have by default.
* `{{flutter_bootstrap_js}}` => inlines the entirety of `flutter_bootstrap.js` (this token obviously doesn't apply to `flutter_bootstrap.js` itself).
* Changed `IndexHtml` to be called `WebTemplate` instead, since it is used for more than just the index.html now.
* We now emit warnings at build time for certain deprecated flows:
* Warn on the old service worker version pattern (i.e.`(const|var) serviceWorkerVersion = null`) and recommends using `{{flutter_service_worker_version}}` token instead
* Warn on use of `FlutterLoader.loadEntrypoint` and recommend using `FlutterLoader.load` instead
* Warn on manual loading of `flutter_service_worker.js`.
* The default `index.html` on `flutter create` now uses an async script tag with `flutter_bootstrap.js`.
Some tests are assuming the flutter sdk code is being checked out to flutter and checking the code to a different repository makes them fail.
Bug: https://github.com/flutter/flutter/issues/144487
Reverts: flutter/flutter#144752
Initiated by: andrewkolos
Reason for reverting: compilation issue has turned the tree red
Original PR Author: andrewkolos
Reviewed By: {christopherfujino}
This change reverts the following previous change:
In service of https://github.com/flutter/flutter/issues/143348
When invoking a package to transform an asset, we set `FLUTTER_BUILD_MODE` to the CLI name of the build mode being used. Inspired by https://github.com/flutter/flutter/issues/101077#issuecomment-1890379501:
> Do transformers know whether they get executed in debug or release mode? I kinda imagine that being useful. Ex: There's a transformer that optimizes the file size of images. Depending on the amount and size of the images, that could take a significant amount of time. Therefore, I might want to only execute it in release builds.
Note for the reviewer: the interesting part of this change can be found in the commit [set environment variable to build mode when running asset transformer…](579912d470). The rest of the change is updating call sites with a new argument.
In service of https://github.com/flutter/flutter/issues/143348
When invoking a package to transform an asset, we set `FLUTTER_BUILD_MODE` to the CLI name of the build mode being used. Inspired by https://github.com/flutter/flutter/issues/101077#issuecomment-1890379501:
> Do transformers know whether they get executed in debug or release mode? I kinda imagine that being useful. Ex: There's a transformer that optimizes the file size of images. Depending on the amount and size of the images, that could take a significant amount of time. Therefore, I might want to only execute it in release builds.
Note for the reviewer: the interesting part of this change can be found in the commit [set environment variable to build mode when running asset transformerâ¦](579912d470). The rest of the change is updating call sites with a new argument.
This fixes https://github.com/flutter/flutter/issues/143703
We need to make sure that when source maps are enabled for the dart2js target, it advertises the sourcemap file as one of its outputs so that the web release bundle can copy it over.
The flutter engine & framework can opt out of this optimization for
individual classes / class hierarchies via:
* `@pragma(flutter:keep-to-string)`
* `@pragma(flutter:keep-to-string-in-subtypes)`
Or by using the convenience constant `@keepToString` from `dart:ui`.
=> This aligns the build process more with g3 (which already does this)
Closes https://github.com/flutter/flutter/issues/52759
Part of #137040 and #80374
The flag `throwOnPluginPubspecError` never actually was enabled during production in #79669, but only in some dart plugin tests. And in the tests the case of the error when enabling the flag was not explicitly tested. The only thing tested was, that it is not thrown when disabled.
As explained [here](https://github.com/flutter/flutter/pull/142035#discussion_r1484237904) the only case, where this error could be thrown is, when a dart implementation and a native inline implementation are provided simultaneously. But throwing an exception there is a wrong behavior, as both can coexist in a plugin package, thus in the pubspec file.
Disabling the flag means, that the error is not thrown and not shown to the user. This is the case in production (contrary to the dart plugin tests), which acts like these plugin cases of implementations are just skipped. And this is what actually should be done.
In conclusion, I think the case of coexisting dart and native implementation in pubspec was just overlooked and therefore this error validation was introduced, which is not necessary or even valid.
For more discussion, see: https://discord.com/channels/608014603317936148/608022056616853515/1200194937791205436
- This is tricky: I already added a test in #142035, which finally complies with the other tests, by removing the flag. So I think it falls in the category of "remove dead code".
- Theoretically this is a breaking change, as removing / altering some tests. But the flag actually was never valid or used, so IDK. But this may not does fall in the category of "contributed tests".
*Replace this paragraph with a description of what this PR is changing or adding, and why. Consider including before/after screenshots.*
*List which issues are fixed by this PR. You must list at least one issue. An issue is not required if the PR fixes something trivial like a typo.*
*If you had to change anything in the [flutter/tests] repo, include a link to the migration guide as per the [breaking change policy].*
`DevFSBytesContent` (and it's descendant `DevFSStringContent`) have setters that change the underlying content. These are unused outside of tests, so this PR removes them. Amongst other things, this could help me refactor https://github.com/flutter/flutter/pull/144660 into something that has fewer pitfalls.
This is purely a refactoring.
Partial implementation of https://github.com/flutter/flutter/issues/143348
This enables asset transformation during hot reload (except for web, because that has its own implementation of `DevFS` ð). Asset transformers will be reapplied after changing any asset and performing a hot reload during `flutter run`.
Explicitly handle the case where the iOS device is not paired. On `flutter run` show an error and bail instead of trying and failing to launch on the device.
On this PR:
```
$ flutter run -d 00008110-0009588C2651401E
'iPhone' is not paired. Open Xcode and trust this computer when prompted.
$
```
Fixes https://github.com/flutter/flutter/issues/144447
Closes https://github.com/flutter/flutter/pull/144095
Originally, my aim was just to refactor (as per usual), but while messing around with the `TableBorder.symmetric` constructor, I realized that `borderRadius` was missing!
This pull request makes a few class constructors more efficient, and it fixes#144277 by adding the missing parameter.
<br>
Reverts flutter/flutter#143244
Initiated by: vashworth
Reason for reverting: Increased `flutter_framework_uncompressed_bytes` - see https://github.com/flutter/flutter/issues/144251
Original PR Author: vashworth
Reviewed By: {jmagman}
This change reverts the following previous change:
Original Description:
Replace `FlutterMacOS.framework` cached artifact with `FlutterMacOS.xcframework`. Also, update usage of `FlutterMacOS.framework` to use `FlutterMacOS.xcframework`.
Part of https://github.com/flutter/flutter/issues/126016.
So far `flutter build web --wasm` was always stripping wasm symbols
except if `--no-strip-wasm` is passed.
=> Ensure that in profile mode we also keep the symbols
Replace `FlutterMacOS.framework` cached artifact with `FlutterMacOS.xcframework`. Also, update usage of `FlutterMacOS.framework` to use `FlutterMacOS.xcframework`.
Part of https://github.com/flutter/flutter/issues/126016.
When the daemon throws an exception, the receiving client is unable to surface stack traces from the daemon.
This is because it is sent with the `trace` key here:
1e8dd1e4d6/packages/flutter_tools/lib/src/daemon.dart (L308)
But the client tries to read it with the `stackTrace` key here:
1e8dd1e4d6/packages/flutter_tools/lib/src/daemon.dart (L343)
Thanks to @mraleph for spotting this!
*List which issues are fixed by this PR. You must list at least one issue. An issue is not required if the PR fixes something trivial like a typo.*
b/326825892
Part of #137040 and #80374
- Differentiate pubspec and resolution errors
- Rename platform to platformKey
- Add TODO for rework logic of flag [throwOnPluginPubspecError]
- Swap for loop: handle by platform and then by plugin
### Context:
DDC modules are abstractions over how libraries are loaded/updated. The entirety of google3 uses the DDC/legacy module system due to its flexibility extensibility over the other two (ES6 and AMD/RequireJS). Unifying DDC's module system saves us from duplicating work and will allow us to have finer grained control over how JS modules are loaded. This is a a prerequisite to features such as hot reload.
### Overview:
This change plumbs a boolean flag through flutter_tools that switches between DDC (new) and AMD (current) modules. This mode is automatically applied when `--extra-front-end-options=--dartdevc-module-format=ddc` is specified alongside `flutter run`. Other important additions include:
* Splitting Flutter artifacts between DDC and AMD modules
* Adding unit tests for the DDC module system
* Additional bootstrapper logic for the DDC module system
We don't expect to see any user-visible behavior or performance differences.
This is dependent on [incoming module system support in DWDS](https://github.com/dart-lang/webdev/pull/2295) and [additional artifacts in the engine](https://github.com/flutter/engine/pull/47783).
This is part of a greater effort to deprecate the AMD module system: https://github.com/dart-lang/sdk/issues/52361
This pull request fixes#143803 by taking advantage of Dart's null-aware operators.
And unlike `switch` expressions ([9 PRs](https://github.com/flutter/flutter/pull/143634) and counting), the Flutter codebase is already fantastic when it comes to null-aware coding. After refactoring the entire repo, all the changes involving `?.` and `??` can fit into a single pull request.
This PR implements the functionality described above and hides it behind
the `--experimental-faster-testing` flag of `flutter test`.
### The following are some performance measurements from test runs
conducted on GitHub Actions
run 1 logs:
https://github.com/derekxu16/flutter_test_ci/actions/runs/8008029772/attempts/1
run 2 logs:
https://github.com/derekxu16/flutter_test_ci/actions/runs/8008029772/attempts/2
run 3 logs:
https://github.com/derekxu16/flutter_test_ci/actions/runs/8008029772/attempts/3
**length of `flutter test --reporter=expanded test/animation
test/foundation` step**
run 1: 54s
run 2: 52s
run 3: 56s
average: 54s
**length of `flutter test --experimental-faster-testing
--reporter=expanded test/animation test/foundation` step**
run 1: 27s
run 2: 27s
run 3: 29s
average: 27.67s (~48.77% shorter than 54s)
**length of `flutter test --reporter=expanded test/animation
test/foundation test/gestures test/painting test/physics test/rendering
test/scheduler test/semantics test/services` step**
run 1: 260s
run 2: 270s
run 3: 305s
average: 278.33s
**length of `flutter test --experimental-faster-testing
--reporter=expanded test/animation test/foundation test/gestures
test/painting test/physics test/rendering test/scheduler test/semantics
test/services` step**
from a clean build (right after deleting the build folder):
run 1: 215s
run 2: 227s
run 3: 245s
average: 229s (~17.72% shorter than 278.33s)
Note that in reality, `test/material` was not passed to `flutter test`
in the trials below. All of the test files under `test/material` except
for `test/material/icons_test.dart` were listed out individually
**length of `flutter test --reporter=expanded test/material` step**
run 1: 408s
run 2: 421s
run 3: 451s
average: 426.67s
**length of `flutter test --experimental-faster-testing
--reporter=expanded test/material` step**
run 1: 382s
run 2: 373s
run 3: 400s
average: 385s (~9.77% shorter than 426.67s)
---------
Co-authored-by: Dan Field <dnfield@google.com>
Update: Accidentally use `--O4` instead of `-O4` in `dev/devicelab/lib/tasks/web_benchmarks.dart` update.
Original description:
* Make `flutter build web` have one option that determins the
optimization level: `-O<level>` / `--optimization-level=<level>` =>
Defaulting to -O4 => Will apply to both dart2js and dart2wasm
* Deprecate `--dart2js-optimization=O<level>`
* Disentagle concept of optimization from concept of static symbols =>
Add a `--strip-wasm` / `--no-strip-wasm` flag that determins whether
static symbols are kept in the resulting wasm file.
* Remove copy&past'ed code in the tests for wasm build tests
* Cleanup some artifacts code, now that we no longer use `wasm-opt`
inside flutter tools
Reverts flutter/flutter#143517
Initiated by: dnfield
Reason for reverting: broke CI, see https://ci.chromium.org/ui/p/flutter/builders/prod/Linux%20web_benchmarks_skwasm/3446/overview
Original PR Author: mkustermann
Reviewed By: {eyebrowsoffire}
This change reverts the following previous change:
Original Description:
* Make `flutter build web` have one option that determins the optimization level: `-O<level>` / `--optimization-level=<level>` => Defaulting to -O4 => Will apply to both dart2js and dart2wasm
* Deprecate `--dart2js-optimization=O<level>`
* Disentagle concept of optimization from concept of static symbols => Add a `--strip-wasm` / `--no-strip-wasm` flag that determins whether static symbols are kept in the resulting wasm file.
* Remove copy&past'ed code in the tests for wasm build tests
* Cleanup some artifacts code, now that we no longer use `wasm-opt` inside flutter tools
* Make `flutter build web` have one option that determins the
optimization level: `-O<level>` / `--optimization-level=<level>` =>
Defaulting to -O4 => Will apply to both dart2js and dart2wasm
* Deprecate `--dart2js-optimization=O<level>`
* Disentagle concept of optimization from concept of static symbols =>
Add a `--strip-wasm` / `--no-strip-wasm` flag that determins whether
static symbols are kept in the resulting wasm file.
* Remove copy&past'ed code in the tests for wasm build tests
* Cleanup some artifacts code, now that we no longer use `wasm-opt`
inside flutter tools
Add column information as table header for `flutter emulators` command.
**Before:**
```
2 available emulators:
Pixel_3_API_30 ⢠Pixel 3 API 30 ⢠Google ⢠android
Resizable_API_33 ⢠Resizable API 33 ⢠Google ⢠android
To run an emulator, run 'flutter emulators --launch <emulator id>'.
To create a new emulator, run 'flutter emulators --create [--name xyz]'.
You can find more information on managing emulators at the links below:
https://developer.android.com/studio/run/managing-avdshttps://developer.android.com/studio/command-line/avdmanager
```
**After:**
```
2 available emulators:
Id ⢠Name ⢠Manufacturer ⢠Platform
Pixel_3_API_30 ⢠Pixel 3 API 30 ⢠Google ⢠android
Resizable_API_33 ⢠Resizable API 33 ⢠Google ⢠android
To run an emulator, run 'flutter emulators --launch <emulator id>'.
To create a new emulator, run 'flutter emulators --create [--name xyz]'.
You can find more information on managing emulators at the links below:
https://developer.android.com/studio/run/managing-avdshttps://developer.android.com/studio/command-line/avdmanager
```
fixes#140656
* Flags to `dart compile wasm`
Some options are not relevant to a standalone user of `dart compile
wasm` (e.g. specyfing dart-sdk, platform file etc). => Those aren't
offered by the `dart compile wasm` tool directly. => We use the
`--extra-compiler-option=` instead which passes through arbitrary
options to the dart2wasm compiler. => We don't maintain compatibility of
those options, if we update them we'll ensure to also update flutter
tools
* Binaryen optimization passes
This change will mean we use the binaryen flags from Dart SDK which are
slightly different from the ones in flutter.
* Optimization configuration
This change will also start using the more standardized `-O` flag for
determining optimization levels. The meaning of those flags have been
mostly aligned with dart2js (with some differences remaining).
* Minimization
Using the new optimization flags, namely `-O4` for `--wasm-opt=full`,
will automatically enable the new `--minify` support. Minification is
Dart semantics preserving but changes the `<obj>.runtimeType.toString()`
to use minified names (just as in dart2js).
* Code size changes
Overall this change will reduce wonderous code size by around 10%.
Issue https://github.com/dart-lang/sdk/issues/54675
In service of https://github.com/flutter/flutter/issues/143348.
**Issue.** The `equals` implementation of `AssetsEntry` is incorrect. It compares `flavors` lists using reference equality. This PR addresses this.
This also adds a test to make sure valid asset `flavors` declarations are parsed correctly.
While we are here, this PR also includes a couple of refactorings:
* `flutter_manifest_test.dart` is a bit large. To better match our style guide, I've factored out some related tests into their own file.
* A couple of changes to the `_validateListType` function in `flutter_manifest.dart`:
* The function now returns a list of errors instead of accepting a list to append onto. This is more readable and also allows callers to know which errors were found by the call.
* The function is renamed to `_validateList` and now accepts an `Object?` instead of an `YamlList`. If the argument is null, an appropriate error message is contained in the output. This saves callers that are only interested in validation from having to write their own null-check, which they all did before.
* Some error strings were tweaked for increased readability and/or grammatical correctness.
This is an attempt at a reland of https://github.com/flutter/flutter/pull/141396
The main changes here that are different than the original PR is fixes to wire up the `flutter test` command properly with the web renderer.
This is in service of https://github.com/flutter/flutter/pull/141194
This will make it easier to get the `flutter run -d <browser>` and `flutter build fuschia` cases easier to get under test.
This is a direct revert of (the revert of (the reland of (the policy pr))): https://github.com/flutter/flutter/pull/143132.
The only change is:
1. to put a conditional all on one line, because the packages repository has a test that uses an old flutter project to make sure nothing regresses. The old project uses an old gradle version, and the old gradle version bundles an old groovy version, and the old groovy version has a bug where lines that start with `&&` don't always work: https://issues.apache.org/jira/browse/GROOVY-7218 (I enjoy that the revert reason ends up providing another strong justification to go forward with the policy). Also thanks to @reidbaker for pointing out this bug.
2. I also made a slight formatting change to the messages that print when out of the support bounds, which I think looks slightly better.
I tested this with on a branch that included a revert of https://github.com/flutter/flutter/pull/142008, and was able to recreate the failure and verify that it was resolved by 1).
Re land of https://github.com/flutter/flutter/pull/142000.
Differences:
1. Fixed the test that was failing in postsubmit. The reason was that the Flutter Gradle Plugin was being applied after KGP in that test, so we couldn't find the KGP version. This caused a log, and the test expects no logs. I moved FGP to after KGP
2. Added to the logs for when we can't find AGP. Change is from
> "Warning: unable to detect project AGP version. Skipping version checking."
to
> ~"Warning: unable to detect project AGP version. Skipping version checking. \nThis may be because you have applied the Flutter Gradle Plugin after AGP."~
update: the above is wrong, changed to
> "Warning: unable to detect project KGP version. Skipping version checking. \nThis may be because you have applied AGP after the Flutter Gradle Plugin."
3. Added a note to the app-level build.gradle templates that FGP must go last
> // The Flutter Gradle Plugin must be applied after the Android and Kotlin Gradle plugin.
Dual Web Compile has had some issues where `flutter test` is not respecting the `--web-renderer` flag for some reason. I haven't gotten entirely to the bottom of the issue, but for now we need to rever these changes while I investigate. This reverts the following PRs:
https://github.com/flutter/flutter/pull/143128https://github.com/flutter/flutter/pull/141396
While doing this revert, I had a few merge conflicts with https://github.com/flutter/flutter/pull/142760, and I tried to resolve the merge conflicts within the spirit of that PR's change, but @chingjun I might need your input on whether the imports I have modified are okay with regards to the change you were making.
Reland of https://github.com/flutter/flutter/pull/142709.
The revert of the revert is in the first commit, the fix in the commit on top.
The move of the fakes for packages/flutter_tools/test/general.shard/resident_runner_test.dart was erroneous before, as it was trying to use setters instead of a private field. This PR changes the private `_devFS` field in the fake to be a public `fakeDevFS` in line with other fakes.
## Original PR description
Native assets in other build systems are not built with `package:native_assets_builder` invoking `build.dart` scripts. Instead all packages have their own blaze rules. Therefore we'd like to not depend on `package:native_assets_builder` from flutter tools in g3 at all.
This PR aims to move the imports of `native_assets_builder` and `native_assets_cli` into the `isolated/` directory and into the files with a `main` function that are not used in with other build systems.
In order to be able to remove all imports in files used by other build systems, two new interfaces are added `HotRunnerNativeAssetsBuilder` and `TestCompilerNativeAssetsBuilder`. New parameters are then piped all the way through from the entry points:
* bin/fuchsia_tester.dart
* lib/executable.dart
The build_system/targets dir is already excluded in other build systems.
So, after this PR only the two above files and build_system/targets import from `isolated/native_assets/` and only `isolated/native_assets/` import `package:native_assets_cli` and `package:native_assets_builder`.
Context:
* https://github.com/flutter/flutter/issues/142041
Policy per https://flutter.dev/go/android-dependency-versions.
https://github.com/flutter/flutter/issues/140913
~Still a WIP while I clean up some error handling, remove some prints, and figure out a Java test (more difficult than the others because I believe we can only install one java version per ci shard).~
~Also it looks like there are errors that I need to fix when this checking is applied to a project that uses the old way of applying AGP/KGP using the top-level `build.gradle` file (instead of the new template way of applying them in the `settings.gradle` file).~ Done, this is why [these lines exist](9af6bae6b9/packages/flutter_tools/gradle/src/main/groovy/flutter.groovy (L72-L88)) in `flutter.groovy`. They just needed to be added
Since `WebReleaseBundle` is responsible for copying over the outputs from the subtargets, so it needs to be reflected in inputs and outputs so that things will be recopied when something changes.
Re-sets two jvmargs that were getting cleared because we set a value for `-Xmx`. Could help with https://github.com/flutter/flutter/issues/142957. Copied from comment here https://github.com/flutter/flutter/issues/142957:
>Two random things I ran into while looking into this that might help:
>
>1. Gradle has defaults for a couple of the jvmargs, and setting any one of them clears those defaults for the others (bug here https://github.com/gradle/gradle/issues/19750). This can cause the "Gradle daemon to consume more and more native memory until it crashes", though the bug typically has a different associated error. It seems worth it to re-set those defaults.
>2. There is a property we can set that will give us a heap dump on OOM ([-XX:HeapDumpOnOutOfMemoryError](https://docs.oracle.com/javase/8/docs/technotes/guides/troubleshoot/clopts001.html))
Mostly just a find and replace from `find . -name gradle.properties -exec sed -i '' 's/\-Xmx4G/-Xmx4G\ \-XX:MaxMetaspaceSize=2G\ \-XX:+HeapDumpOnOutOfMemoryError/g' {} \;`, with the templates and the one test that writes from a string replaced by hand. I didn't set a value for `MaxMetaspaceSize` in the template files because I want to make sure this value doesn't cause problems in ci first (changes to the templates are essentially un-revertable for those who `flutter create` while the changes exist).
Reverts flutter/flutter#142709
Initiated by: vashworth
Reason for reverting: `Mac tool_tests_general` started failing on this commit: https://ci.chromium.org/ui/p/flutter/builders/prod/Mac%20tool_tests_general/15552/overview
Original PR Author: dcharkes
Reviewed By: {christopherfujino, chingjun, reidbaker}
This change reverts the following previous change:
Original Description:
Native assets in other build systems are not built with `package:native_assets_builder` invoking `build.dart` scripts. Instead all packages have their own blaze rules. Therefore we'd like to not depend on `package:native_assets_builder` from flutter tools in g3 at all.
This PR aims to move the imports of `native_assets_builder` and `native_assets_cli` into the `isolated/` directory and into the files with a `main` function that are not used in with other build systems.
In order to be able to remove all imports in files used by other build systems, two new interfaces are added `HotRunnerNativeAssetsBuilder` and `TestCompilerNativeAssetsBuilder`. New parameters are then piped all the way through from the entry points:
* bin/fuchsia_tester.dart
* lib/executable.dart
The build_system/targets dir is already excluded in other build systems.
So, after this PR only the two above files and build_system/targets import from `isolated/native_assets/` and only `isolated/native_assets/` import `package:native_assets_cli` and `package:native_assets_builder`.
Context:
* https://github.com/flutter/flutter/issues/142041
Native assets in other build systems are not built with `package:native_assets_builder` invoking `build.dart` scripts. Instead all packages have their own blaze rules. Therefore we'd like to not depend on `package:native_assets_builder` from flutter tools in g3 at all.
This PR aims to move the imports of `native_assets_builder` and `native_assets_cli` into the `isolated/` directory and into the files with a `main` function that are not used in with other build systems.
In order to be able to remove all imports in files used by other build systems, two new interfaces are added `HotRunnerNativeAssetsBuilder` and `TestCompilerNativeAssetsBuilder`. New parameters are then piped all the way through from the entry points:
* bin/fuchsia_tester.dart
* lib/executable.dart
The build_system/targets dir is already excluded in other build systems.
So, after this PR only the two above files and build_system/targets import from `isolated/native_assets/` and only `isolated/native_assets/` import `package:native_assets_cli` and `package:native_assets_builder`.
Context:
* https://github.com/flutter/flutter/issues/142041
â¦and removed the no-shuffle tag.
This PR fixes#142376 by fixing the flaky test in language_version_test.dart and removes the no-shuffle tag.
Â
## The Problem
The test expected the language version that is set at the top of the test file ('2.13' set in language_version_test.dart â line 14) but defaulted to the language version set in the file it is testing ('2.12' is set in language_version.dart).
This problem was hidden when some other test ran before this test and set up the language version correctly.
Â
## The Fix
Make the test itself load the default language version we are testing against.
Add a new `BuildTargets` class that provides commonly used build targets. And avoid importing files from `build_system/targets` except from the top level entrypoints or from top level commands.
Also move `scene_importer.dart` and `shader_compiler.dart` into `build_system/tools` because they are not `Target` classes, but wrapper for certain tools.
With this change, we can ignore all files in `build_system/targets` internally and make PR #142709 easier to land internally. See cl/603434066 for the corresponding internal change.
Related to:
https://github.com/flutter/flutter/pull/142709https://github.com/flutter/flutter/issues/142041
Also note that I have opted to add a new variable in `globals.dart` for `BuildTargets` in this PR, but I know that we are trying to get rid of globals. Several alternatives that I was considering:
1. Add a new field in `BuildSystem` that returns a `BuildTargets` instance. Since `BuildSystem` is already in `globals`, we can access build targets using `globals.buildSystem.buildTargets` without adding a new global variable.
2. Properly inject the `BuildTargetsImpl` instance from the top level `executable.dart` and top level commands.
Let me know if you want me to do one of the above instead. Thanks!
This implements dual compile via the newly available flutter.js bootstrapping APIs for intelligent build fallback.
* Users can now use the `FlutterLoader.load` API from flutter.js
* Flutter tool injects build info into the `index.html` of the user so that the bootstrapper knows which build variants are available to bootstrap
* The semantics of the `--wasm` flag for `flutter build web` have changed:
- Instead of producing a separate `build/web_wasm` directory, the output goes to the `build/web` directory like a normal web build
- Produces a dual build that contains two build variants: dart2wasm+skwasm and dart2js+CanvasKit. The dart2wasm+skwasm will only work on Chrome in a cross-origin isolated context, all other environments will fall back to dart2js+CanvasKit.
- `--wasm` and `--web-renderer` are now mutually exclusive. Since there are multiple build variants with `--wasm`, the web renderer cannot be expressed via a single command-line flag. For now, we are hard coding what build variants are produced with the `--wasm` flag, but I plan on making this more customizable in the future.
* Build targets now can optionally provide a "build key" which can uniquely identify any specific parameterization of that build target. This way, the build target can invalidate itself by changing its build key. This works a bit better than just stuffing everything into the environment defines because (a) it doesn't invalidate the entire build, just the targets which are affected and (b) settings for multiple build variants don't translate well to the flat map of environment defines.
When the Dart VM is not found within 10 minutes in CI on CoreDevices (iOS 17+), stop the app and upload the logs from DerivedData. The app has to be stopped first since the logs are not put in DerivedData until it's stopped.
Also, rearranged some logic to have CoreDevice have its own function for Dart VM url discovery.
Debugging for https://github.com/flutter/flutter/issues/142448.
Refactors `ShaderTarget` to make it opaque as to whether it's using Impeller or SkSL and instead has it focus on the target platform it's generating for.
ImpellerC includes SkSL right now whether you ask for it or not.
The tester target also might need SkSL or Vulkan depending on whether `--enable-impeller` is passed.
This PR increases Android's `minSdkVersion` to 21.
There are two changes in this PR aside from simply increasing the number
from 19 to 21 everywhere.
First, tests using `flutter_gallery` fail without updating the
lockfiles. The changes in the PR are the results of running
`dev/tools/bin/generate_gradle_lockfiles.dart` on that app.
Second, from
[here](https://developer.android.com/build/multidex#mdex-pre-l):
> if your minSdkVersion is 21 or higher, multidex is enabled by default
and you don't need the multidex library.
As a result, the `multidex` option everywhere is obsolete. This PR
removes all logic and tests related to that option that I could find.
`Google testing` and `customer_tests` pass on this PR, so it seems like
this won't be too breaking if it is at all. If needed I'll give this
some time to bake in the framework before landing the flutter/engine
PRs.
Context: https://github.com/flutter/flutter/issues/138117,
https://github.com/flutter/flutter/issues/141277, b/319373605
It's now possible to natively compile a flutter app for windows-arm64. Cross-compilation is not yet implemented.
Uses arm64 artifacts now available for Dart/Flutter. Platform detection is based on Abi class, provided by Dart. Depending if Dart is an arm64 or x64 binary, the Abi is set accordingly. Initial bootstrap of dart artifacts (update_dart_sdk.ps1) is checking PROCESSOR_ARCHITECTURE environment variable, which is the way to detect host architecture on Windows.
This is available only for master channel (on other channels, it fallbacks to windows-x64).
On windows-x64, it produces an x64 app. On windows-arm64, it produces an arm64 app.
The legacy welcome message would be printed even if `CI=true` confusing
parsers of the output.
This fixes: https://github.com/flutter/flutter/issues/139737
---------
Co-authored-by: eliasyishak <42216813+eliasyishak@users.noreply.github.com>
Part of work on [#101077](https://github.com/flutter/flutter/pull/141194). This is done as a separate PR to avoid a massive diff.
## Context
1. The `FakeCommand` class accepts a list of patterns that's used to match a command given to its `FakeProcessManager`. Since `FakeCommand` can match a list of patterns, not just specifically strings, it can be used to match commands where the exact value of some arguments can't (easily) known ahead of time. For example, a part of the tool may invoke a command with an argument that is the path of a temporarily file that has a randomly-generated basename.
2. The `FakeCommand` class provides on `onRun` parameter, which is a callback that is run when the `FakeProcessManager` runs a command that matches the `FakeCommand` in question.
## Issue
In the event that a `FakeCommand` is constructed using patterns, the test code can't know the exact values used for arguments in the command. This PR proposes changing the type of `onRun` from `VoidCallback?` to `void Function(List<String>)?`. When run, the value `List<String>` parameter will be the full command that the `FakeCommand` matched.
Example:
```dart
FakeCommand(
command: <Pattern>[
artifacts.getArtifactPath(Artifact.engineDartBinary),
'run',
'vector_graphics_compiler',
RegExp(r'--input=/.*\.temp'),
RegExp(r'--output=/.*\.temp'),
],
onRun: (List<String> command) {
final outputPath = (() {
// code to parse `--output` from `command`
})();
testFileSystem.file(outputPath).createSync(recursive: true);
},
)
```
This PR adds a test that reproduces the problem described in the linked issue: hot restart on the web seems to not update if the app being run is `const`.
The new test is expected to fail, until the `const` issue with hot restart in the web is resolved.
Expected failure mode is a 15s timeout in the following test:
```
02:31 +3 ~1 -1: Hot reload (index.html: Default) (with `const MyApp()`)): newly added code executes during hot restart [E]
TimeoutException after 0:00:15.000000: Future not completed
dart:async _startMicrotaskLoop
...
```
(And then a bunch of output that I'm not 100% sure is intended :))
## Issues
* #141588
Part of work on https://github.com/flutter/flutter/pull/141194
The [`AssetBundle`](0833929c99/packages/flutter_tools/lib/src/asset.dart (L80)) class contains two members, `entries` and `entryKinds`. `entries` contains asset data indexed by asset key. `entryKinds` contains the "kinds" of these assets, again indexed by asset key.
**Change.** Rather than have two separate maps, this PR proposes combining these maps into one by wrapping the asset data and kind into a single data type `AssetBundleEntry`.
**Purpose.** In https://github.com/flutter/flutter/pull/141194, I am considering associating more information with an asset. In particular, what transformers are meant to be applied to it when copying it to the build output. Rather than adding another map member onto `AssetBundle` (e.g. `entryTransformers`), I decided to make things neater by introducing the `AssetBundleEntry` type.
Fixes https://github.com/flutter/flutter/issues/141827
Reland: https://dart-review.googlesource.com/c/sdk/+/346960 has rolled into g3, so the imports should now resolve in g3 as well.
> [!CAUTION]
> _Do NOT merge if "Google Testing" bot didn't run!_
Rolls the packages from https://github.com/dart-lang/native in the native assets implementation.
Most notable we're refactoring `package:native_assets_cli` for `build.dart` use.
Therefore, all imports to that package for Flutter/Dart should be to the implementation internals that are no longer visible for `build.dart` writers. Hence all the import updates.
No behavior in Flutter apps should change.
This PR also updates the template to use the latests version of `package:native_assets_cli` which no longer exposes all the implementation details.
Packages the native assets for iOS and MacOS in frameworks.
Issue:
* https://github.com/flutter/flutter/issues/140544
* https://github.com/flutter/flutter/issues/129757
## Details
* [x] This packages dylibs from the native assets feature in frameworks. It packages every dylib in a separate framework.
* [x] The dylib name is updated to use `@rpath` instead of `@executable_path`.
* [x] The dylibs for flutter-tester are no longer modified to change the install name. (Previously it was wrongly updating the install name to the location the dylib would have once deployed in an app.)
* [x] Use symlinking on MacOS.
Includes the Engine roll from
https://github.com/flutter/flutter/pull/141841
A new version of Dart is having trouble with the tool integration test
test `passing one file with errors are detected`:
https://ci.chromium.org/ui/p/flutter/builders/try/Mac%20tool_integration_tests_2_4/31851/overview.
However the analysis server emits the expected errors when we give it
both the file without issues and the file with issues.
My guess is that the analysis server has changed it's behavior slightly
when supplied with a single malformed file.
Since the Dart roll is >20 dev versions behind, and this is the only
failing presubmit test, and it's testing something a bit weird, I
suggest we investigate the right way to test the thing that test was
attempting to cover as a follow-up.
Reverts flutter/flutter#137618
Initiated by: Jasguerrero
This change reverts the following previous change:
Original Description:
It's now possible to natively compile a flutter app for
windows-arm64. Cross-compilation is not yet implemented.
Uses arm64 artifacts now available for Dart/Flutter.
Platform detection is based on Abi class, provided by Dart. Depending if
Dart is an arm64 or x64 binary, the Abi is set accordingly.
Initial bootstrap of dart artifacts (update_dart_sdk.ps1) is checking
PROCESSOR_ARCHITECTURE environment variable, which is the way to detect
host architecture on Windows.
This is available only for master channel (on other channels, it
fallbacks to windows-x64).
On windows-x64, it produces an x64 app. On windows-arm64, it produces an
arm64 app.
It's now possible to natively compile a flutter app for
windows-arm64. Cross-compilation is not yet implemented.
Uses arm64 artifacts now available for Dart/Flutter.
Platform detection is based on Abi class, provided by Dart. Depending if
Dart is an arm64 or x64 binary, the Abi is set accordingly.
Initial bootstrap of dart artifacts (update_dart_sdk.ps1) is checking
PROCESSOR_ARCHITECTURE environment variable, which is the way to detect
host architecture on Windows.
This is available only for master channel (on other channels, it
fallbacks to windows-x64).
On windows-x64, it produces an x64 app. On windows-arm64, it produces an
arm64 app.
Rolls the packages from https://github.com/dart-lang/native in the native assets implementation.
Most notable we're refactoring `package:native_assets_cli` for `build.dart` use.
Therefore, all imports to that package for Flutter/Dart should be to the implementation internals that are no longer visible for `build.dart` writers. Hence all the import updates.
No behavior in Flutter apps should change.
This PR also updates the template to use the latests version of `package:native_assets_cli` which no longer exposes all the implementation details.
## Description
This adds a check to make sure that the `--empty` flag isn't applied to non-app templates.
## Related Issues
- Fixes https://github.com/flutter/flutter/issues/141592
## Tests
- Added a test.
I continued [my mission](https://github.com/flutter/flutter/pull/141431) to find as many typos as I could. This time it's a smaller set than before.
There is no need for issues since it's a typo fix.
This should wait for some upstream work, just don't want to lose it locally for now. I'll switch this from draft and update the description when it's ready.
Support for FFI calls with @Native external functions through Native assets on Android add to app. This enables bundling native code without any build-system boilerplate code.
For more info see:
* https://github.com/flutter/flutter/issues/129757
## Implementation details for Android add2app
The `.so` files are bundled with the same mechanism that bundles `libapp.so`.
Change the following in the `flutter create` templates. I didn't make any auto-migrations for existing apps because none seem that critical:
1. Turn on `ASSETCATALOG_COMPILER_GENERATE_SWIFT_ASSET_SYMBOL_EXTENSIONS` in iOS and macOS.
1. Turn on `BuildIndependentTargetsInParallel` in macOS template. https://github.com/flutter/flutter/pull/125827/files#r1181817619
1. Turn on `DEAD_CODE_STRIPPING` in macOS template.
1. Set `ENABLE_USER_SCRIPT_SANDBOXING=NO` in iOS and macOS template. `flutter` scripts don't work with this on. This might require a migration in the future to explicitly turn this one off. However at least for now if the setting isn't present it defaults to `NO`.
Add migration for `LastUpgradeVersion` so users won't see these validation issues in Xcode.
Run migrator on all the example apps. A few aren't Flutter apps so I edited them in Xcode.
Fixes https://github.com/flutter/flutter/issues/140253
See also https://github.com/flutter/flutter/issues/125817 and https://github.com/flutter/flutter/pull/90304.
Fixes https://github.com/flutter/flutter/issues/140665 by replacing use of `setUpAll` and `teardownAll` with `setUp` and `teardown` respectively.
My theory has to how this issue happened (and how this PR fixes it) is that `setUpAll` is not guaranteed to run after any `setUp` in any parent test groups. However, `setUp` is guaranteed to run after any set-up callbacks in parent groups. From the [documentation](https://api.flutter.dev/flutter/flutter_test/setUp.html):
> If this is called within a test group, it applies only to tests in that group. The body will be run after any set-up callbacks in parent groups or at the top level.
Meanwhile, [`setUpAll`](https://api.flutter.dev/flutter/flutter_test/setUpAll.html) has a weaker documented guarantee that applies only to other `setUpAll` calls:
> If this is called within a test group, The body will run before all tests in that group. It will be run after any [setUpAll](https://api.flutter.dev/flutter/flutter_test/setUpAll.html) callbacks in parent groups or at the top level. It won't be run if none of the tests in the group are run.
