* Add html_all_collection type to correct typeof document.all
* process HTMLAllCollection like HTMLElement + fix flow issue
* fix lint
* move flow fix comment
* Make it work with iframes too
* optimize how we get html_all_collection type
* use once Object.prototype.toString.call
* Use Retry lane for resuming CPU suspended work
* Use a global render timeout for CPU suspense heuristics
* Fix profiler test since we're now reading time more often
* Sync to new reconciler
* Test synchronously rerendering should not render more rows
* Add flow to SyntheticEvent
* Minimal implementation of known and unknown synthetic events
* less casting
* Update EnterLeaveEventPlugin.js
Co-authored-by: Dan Abramov <dan.abramov@gmail.com>
We use safelyCallDestroy for commitUnmount and passive effects unmounts but we call destroy directly in commitHookEffectListUnmount (AKA layout effects unmounts because we don't use this anywhere else). This PR changes the direct destroy call to safelyCallDestroy for consistency
The behavior of error boundaries for passive effects that throw during cleanup was recently changed so that React ignores boundaries which are also unmounting in favor of still-mounted boundaries. This commit implements that same behavior for layout effects (useLayoutEffect, componentWillUnmount, and ref-detachment).
The new, skip-unmounting-boundaries behavior is behind a feature flag (`skipUnmountedBoundaries`).
A passive effect's cleanup function may throw after an unmount. Prior to this commit, such an error would be ignored. (React would not notify any error boundaries.) After this commit, React's behavior varies depending on which reconciler fork is being used.
For the old reconciler, React will call componentDidCatch for the nearest unmounted error boundary (if there is one). If there are no unmounted error boundaries, React will still swallow the error because the return pointer has been disconnected, so the normal error handling logic does not know how to traverse the tree to find the nearest still-mounted ancestor.
For the new reconciler, React will skip any unmounted boundaries and look for a still-mounted boundary. If one is found, it will call getDerivedStateFromError and/or componentDidCatch (depending on the type of boundary).
Tests have been added for both reconciler variants for now.
These stacks improve the profiler data but they're expensive to generate and generating them can also cause runtime errors in larger applications (although an exact repro has been hard to nail down). Removing them for now. We can revisit adding them after this profiler has been integrated into the DevTools extension and we can generate them lazily.
No longer need this, since we have starvation protection in userspace.
This will also allow us to remove the concept from the Scheduler
package, which is nice because `postTask` doesn't currently support it.
`shouldYield` will currently return `true` if there's a higher priority
task in the Scheduler queue.
Since we yield every 5ms anyway, this doesn't really have any practical
benefit. On the contrary, the extra checks on every `shouldYield` call
are wasteful.
* [eslint-plugin-react-cooks] Report constant constructions
The dependency array passed to a React hook can be thought of as a list of cache keys. On each render, if any dependency is not `===` its previous value, the hook will be rerun. Constructing a new object/array/function/etc directly within your render function means that the value will be referentially unique on each render. If you then use that value as a hook dependency, that hook will get a "cache miss" on every render, making the dependency array useless.
This can be especially dangerous since it can cascade. If a hook such as `useMemo` is rerun on each render, not only are we bypassing the option to avoid potentially expensive work, but the value _returned_ by `useMemo` may end up being referentially unique on each render causing other downstream hooks or memoized components to become deoptimized.
* Fix/remove existing tests
* Don't give an autofix of wrapping object declarations
It may not be safe to just wrap the declaration of an object, since the object may get mutated.
Only offer this autofix for functions which are unlikely to get mutated.
Also, update the message to clarify that the entire construction of the value should get wrapped.
* Handle the long tail of nodes that will be referentially unique
* Catch let/var constant constructions on initial assignment
* Trim trailing whitespace
* Address feedback from @gaearon
* Rename "assignment" to "initialization"
* Add test for a constant construction used in multiple dependency arrays
Because `postTask` returns a promise, errors inside a `postTask`
callback result in the promise being rejected.
If we don't catch those errors, then the browser will report an
"Unhandled promise rejection" error. This is a confusing message to see
in the console, because the fact that `postTask` is a promise-based API
is an implementation detail from the perspective of the developer.
"Promise rejection" is a red herring.
On the other hand, if we do catch those errors, then we need to report
the error to the user in some other way.
What we really want is the default error reporting behavior that a
normal, non-Promise browser event gets.
So, we'll re-throw inside `setTimeout`.
This updates the experimental Scheduler postTask build to call postTask
directly, instead of managing our own custom queue and work loop.
We still use a deadline 5ms mechanism to implement `shouldYield`.
The main thing that postTask is currently missing is the continuation
feature — when yielding to the main thread, the yielding task is sent
to the back of the queue, instead of maintaining its position.
While this would be nice to have, even without it, postTask may be good
enough to replace our userspace implementation.
We'll run some tests to see.
* Add ReactVersion to SchedulingProfiler render scheduled marks
* Move ReactVersion to a new --react-init-* mark
Co-authored-by: E-Liang Tan <eliang@eliangtan.com>
* Support inner component _debugOwner in memo
* test with devtool context
* remove memo test
* Merged master; tweaked test and snapshot
* Pass owner to createFiber fn when creating a memo component.
Co-authored-by: Theodore Han <tqhan317@gmail.com>
* test: Simulate mouseover in browser
* Fix duplicate onMouseEnter event when relatedTarget is a root
* Test leave as well
Co-authored-by: Sebastian Silbermann <silbermann.sebastian@gmail.com>
* test: Add current behavior for event types of onFocus/onBlur
* fix: onFocus/onBlur have a matching event type
* fix useFocus
* fix: don't compare native event types with react event types
* Add FocusIn/FocusOutEventInterface
* A simpler alternative fix
* Add regression tests
* Always pass React event type and fix beforeinput
Co-authored-by: Dan Abramov <dan.abramov@me.com>
