Internal act: Flush timers at end of scope

[acdlite]

If there are any suspended fallbacks at the end of the act scope, force them to display by running the pending timers (i.e. setTimeout).

The public implementation of act achieves the same behavior with an extra check in the work loop (shouldForceFlushFallbacks). Since our internal act needs to work in both development and production, without additional runtime checks, we instead rely on Jest's mock timers.

This doesn't not affect refresh transitions, which are meant to delay indefinitely, because in that case we exit the work loop without posting a timer.

Also doesn't affect the public act, only the internal one used by our tests.

[codesandbox-ci]

This pull request is automatically built and testable in CodeSandbox.

To see build info of the built libraries, click here or the icon next to each commit SHA.

Latest deployment of this branch, based on commit a3a5877:

Sandbox	Source
React	Configuration

[sizebot]

No significant bundle size changes to report.

Size changes (experimental)

Generated by 🚫 dangerJS against a3a5877

[sizebot]

No significant bundle size changes to report.

Size changes (stable)

Generated by 🚫 dangerJS against a3a5877

[rickhanlonii]

I'm concerned about using timers for this because you'll trigger side effects that may not actually happen in a real environment. There are some internal tests in React Native that this will break that expose this concern, and I needed to add a flag to get the tests to pass: D22803246

Here was an attempt to fix it without timers: #19471

[acdlite]

@rickhanlonii This does not affect public tests, only the internal ones used to test React itself.

The public implementation of act achieves the same behavior with an extra check in the work loop (shouldForceFlushFallbacks).

[acdlite]

See #19745 for full context

[acdlite]

Need to fix ESlint failure. (jest global not defined.)

[rickhanlonii]

My points still apply though right?

[acdlite]

React Native uses the public version of act. This PR is only additive; it doesn't change any existing behavior.

I have a separate plan to fix timers not being flushed for the public act.

[rickhanlonii]

Approving to unblock, but the React Native test was a Relay test so I'm still concerned about triggering timer side-effects that are not related to fallback.

[acdlite]

Ah I see, I didn't realize you meant a Relay test.

Does it use partial rendering? (i.e. toFlushAndYieldThrough, as opposed to toFlushAndYield or toHaveYielded). If not, then the public act should suffice (i.e. flush the fallbacks correctly without using fake timers) once #19789 lands.

[rickhanlonii]

My fault, I used "React Native" as an alias for "xplat/js" out of habit. That sounds good, and would allow us to remove this hack to get around using timers to flush fallbacks.