By default, the browser fuzzes the timer APIs such that they have a granularity of approximately 100 microseconds (this is due to Spectre mitigation techniques). However, many of the thing we are trying to measure actually have a much finer granularity than 100 microseconds. As a result, many of our benchmarks are extremely noisy and don't provide accurate data.
By serving the initial script files with the `Cross-Origin-Opener-Policy: same-origin` and `Cross-Origin-Embedder-Policy: require-corp` HTTP headers, the browser runs the benchmarks in a `crossOriginIsolated` context, which restores the fine granularity of APIs such as `performance.now()` to microsecond precision.
Also, we were considering anything an outlier that was more than one standard deviation away from the mean. In a normal distribution, that means we are only capturing 68% of the data and the rest are considered outliers. This is not ideal. Doing two standard deviations away captures 95% of the data, and the outliers are in the remaining 5%, which seems much more reasonable.
I think the flake is due to setclipboard or semantics update race condition. I migrated the test to use integration test package which relies less on timing
fixes https://github.com/flutter/flutter/issues/124636
* Add Linux unit tests to plugin template
Adds an example native unit test to the plugin template for Linux,
matching the structure we use for our 1P plugin unit tests. Once these
have been added for all platforms+languages, they will be documented on
a new plugin development page to explain their use.
While ideally we would adjust the engine APIs first to allow for testing
the method call handler directly, it's unclear when we will have time
for that work, and for a complex plugin most of the testing wouldn't be
at that layer anyway, so having the structure in place with the
limitations documented is still a significant improvement over having
nothing in the template.
Part of https://github.com/flutter/flutter/issues/82458
* Add creation test
* Add integration tests
* Missing newlines
* test owner
* Typo
* Added integration test for wide gamut support.
* cleaned up
* deleted files that can be generated
* switched back to standard flutter in the shell script
* added devicelab task
* removed analysis options
* analyzer
* Fixed task
* made local_run.sh not executable
* analyzer
* removed the logo asset
* added task to ci
* updated pubspec
* analysis errors fixed
* updated pubspec
* Add Windows unit tests to plugin template
Adds an example native unit test to the plugin template for Windows,
matching the format we use for our 1P plugin example app unit tests.
Once these have been added for all platforms+languages, they will be
documented on a new plugin development page to explain their use.
Since we don't appear to be running our current plugin e2e tests for
Windows, this adds a new configuration to run them. I haven't
`led`-tested this, so it may not work, but this will give a starting
point for getting them running.
Part of https://github.com/flutter/flutter/issues/82458
* Minor fix
* Add test owner
* Fix typo
* Fix test feature flag
* wip
* add track entire web build output dir size
* add more fields
* migrate metrics to use bytes rather than kb
* update keys
* use -9 on tar cz
* delete tempDir and tar first before measuring size
