* Store list of contexts on the fiber
Currently, context can only be read by a special type of component,
ContextConsumer. We want to add support to all fibers, including
classes and functional components.
Each fiber may read from one or more contexts. To enable quick, mono-
morphic access of this list, we'll store them on a fiber property.
* Context.unstable_read
unstable_read can be called anywhere within the render phase. That
includes the render method, getDerivedStateFromProps, constructors,
functional components, and context consumer render props.
If it's called outside the render phase, an error is thrown.
* Remove vestigial context cursor
Wasn't being used.
* Split fiber.expirationTime into two separate fields
Currently, the `expirationTime` field represents the pending work of
both the fiber itself — including new props, state, and context — and of
any updates in that fiber's subtree.
This commit adds a second field called `childExpirationTime`. Now
`expirationTime` only represents the pending work of the fiber itself.
The subtree's pending work is represented by `childExpirationTime`.
The biggest advantage is it requires fewer checks to bailout on already
finished work. For most types of work, if the `expirationTime` does not
match the render expiration time, we can bailout immediately without
any further checks. This won't work for fibers that have
`shouldComponentUpdate` semantics (class components), for which we still
need to check for props and state changes explicitly.
* Performance nits
Optimize `readContext` for most common case
* Add a test that StrictMode shows up in the component stack
The SSR test passes. The client one doesn't.
* Include Modes in component stack
* Update other tests to include modes
Adds custom Jest matchers that help with writing async tests:
- `toFlushThrough`
- `toFlushAll`
- `toFlushAndThrow`
- `toClearYields`
Each one accepts an array of expected yielded values, to prevent
false negatives.
Eventually I imagine we'll want to publish this on npm.
* Add a repro case for profiler unwinding
This currently fails the tests due to an unexpected warning.
* Add a regression test for context stack
* Simplify the first test case
* Update nextUnitOfWork inside completeUnitOfWork()
The bug was caused by a structure like this:
</Provider>
</div>
</errorInCompletePhase>
We forgot to update nextUnitOfWork so it was still pointing at Provider when errorInCompletePhase threw. As a result, we would try to unwind from Provider (rather than from errorInCompletePhase), and thus pop the Provider twice.
The `yield` method isn't tied to any specific root. Putting this
on the main export enables test components that are not within scope
to yield even if they don't have access to the currently rendering
root instance. This follows the pattern established by ReactNoop.
Added a `clearYields` method, too, for reading values that were yielded
out of band. This is also based on ReactNoop.
* Revert #5947 and disable the test
* Fix isDefaultPrevented and isPropagationStopped to not get nulled
This was a bug introduced by #5947. It's very confusing that they become nulled while stopPropagation/preventDefault don't.
* Add a comment
* Run Prettier
* Fix grammar
* Fix Portal unmount
Before that change, currentParent is not set as a container even if it should so it break on react-native and probably other custom renderers
* Assert that *ToContainer() methods receive containers
* Add regression tests
* Add comments
This is a leftover from #13161 that I forgot to include.
It ensures we don't accidentally write code in the old way and end up passing the stack twice.
* Use %s in the console calls
* Add shared/warningWithStack
* Convert some warning callsites to warningWithStack
* Use warningInStack in shared utilities and remove unnecessary checks
* Replace more warning() calls with warningWithStack()
* Fixes after rebase + use warningWithStack in react
* Make warning have stack by default; warningWithoutStack opts out
* Forbid builds that may not use internals
* Revert newly added stacks
I changed my mind and want to keep this PR without functional changes. So we won't "fix" any warnings that are already missing stacks. We'll do it in follow-ups instead.
* Fix silly find/replace mistake
* Reorder imports
* Add protection against warning argument count mismatches
* Address review
* Suspending inside a constructor outside of strict mode
Outside of strict mode, suspended components commit in an incomplete
state, then are synchronously deleted in a subsequent commit. If a
component suspends inside the constructor, it mounts without
an instance.
This breaks at least one invariant: during deletion, we assume that
every mounted component has an instance, and check the instance for
the existence of `componentWillUnmount`.
Rather than add a redundant check to the deletion of every class
component, components that suspend inside their constructor and outside
of strict mode are turned into empty functional components before they
are mounted. This is a bit weird, but it's an edge case, and the empty
component will be synchronously unmounted regardless.
* Do not fire lifecycles of a suspended component
In non-strict mode, suspended components commit, but their lifecycles
should not fire.
Instead of wrapping ReactDebugCurrentFrame.getStackAddendum() call into a custom wrapper inside ReactElementValidator, "teach" the main ReactDebugCurrentFrame.getStackAddendum() to take currently validating element into account.
* Fix getComponentName() for types with nested $$typeof
* Temporarily remove Profiler ID from messages
* Change getComponentName() signature to take just type
It doesn't actually need the whole Fiber.
* Remove getComponentName() forks in isomorphic and SSR
* Remove unnecessary .type access where we already have a type
* Remove unused type
* Remove event simulation of onChange events
It’s time to get rid of even more `ReactTestUtils.Simulate`s. In this PR
we remove the event simulation from all onChange tests. To do this, we
have to get a setter to the untracked value/checked props.
All remaining `ReactTestUtils.Simulate` calls are either testing
ReactTestUtils or assert that they do/don't throw.
* Use input instead of change event for all but checkbox, radio, and select
`React$ElementRef<T>` is the type of the ref _instance_ for a component of type T, whereas `React$Ref<T>` is the type of the ref _prop_ for a component of type T, which seems to be the intended type here.
* Refactor ReactDebugCurrentFiber to use named exports
This makes the difference between it and ReactFiberCurrentFrame a bit clearer.
ReactDebugCurrentFiber is Fiber's own implementation.
ReactFiberCurrentFrame is the thing that holds a reference to the current implementation and delegates to it.
* Unify ReactFiberComponentTreeHook and ReactDebugCurrentFiber
Conceptually they're very related.
ReactFiberComponentTreeHook contains implementation details of reading Fiber's stack (both in DEV and PROD).
ReactDebugCurrentFiber contained a reference to the current fiber, and used the above utility.
It was confusing when to use which one. Colocating them makes it clearer what you could do with each method.
In the future, the plan is to stop using these methods explicitly in most places, and instead delegate to a warning system that includes stacks automatically. This change makes future refactorings simpler by colocating related logic.
* Rename methods to better reflect their meanings
Clarify which are DEV or PROD-only.
Clarify which can return null.
I believe the "work in progress only" was a mistake. I introduced it because I wasn't sure what guarantees we have around .return. But we know for sure that following a .return chain gives us an accurate stack even if we get into WIP trees because we don't have reparenting. So it's fine to relax that naming.
* Rename ReactDebugCurrentFiber -> ReactCurrentFiber
It's not completely DEV-only anymore.
Individual methods already specify whether they work in DEV or PROD in their names.
An infinite update loop can occur when an update is scheduled inside a
lifecycle method, which causes a re-render, which schedules another
update, and so on. Before the Fiber rewrite, this scenario resulted in a
stack overflow.
Because Fiber does not use the JavaScript stack, we maintain our own
counter to track the number of nested, synchronous updates. We throw an
error if the limit is exceeded.
The nested update limit is currently 1000. I chose this number
arbitrarily, certain that there was no valid reason for a component to
schedule so many synchronous re-renders.
I think we can go much lower. This commit decreases the limit to 50. I
believe this is still comfortably above the reasonable number of
synchronous re-renders a component may perform.
This will make it easier for developers to debug infinite update bugs
when they occur.
A recent change to the scheduler caused a regression when scheduling
many updates within a single batch. Added a test case that would
have caught this.
This is an unobservable change to all but the (under development) DevTools Profiler plugin. It is being done so that the plugin can safely feature detect a version of React that supports it. The profiler API has existed since the 16.4.0 release, but it did not support the DevTools plugin prior to PR #13058.
Side note: I am not a big fan of the term "base duration". Both it and "actual duration" are kind of awkward and vague. If anyone has suggestions for better names– this is the best time to bikeshed about them.
* Reset ReactProfilerTimer's DEV-only Fiber stack after an error
* Added ReactNoop functionality to error during "complete" phase
* Added failing profiler stack unwinding test
* Potential fix for unwinding time bug
* Renamed test
* Don't record time until complete phase succeeds. Simplifies unwinding.
* Expanded ReactProfilerDevToolsIntegration-test coverage a bit
* Added unstable_flushWithoutCommitting method to noop renderer
* Added failing multi-root/batch test to ReactProfiler-test
* Beefed up tests a bit and added some TODOs
* Profiler timer differentiates between batched commits and in-progress async work
This was a two-part change:
1) Don't count time spent working on a batched commit against yielded async work.
2) Don't assert an empty stack after processing a batched commit (because there may be yielded async work)
This is kind of a hacky solution, and may have problems that I haven't thought of yet. I need to commit this so I can mentally clock out for a bit without worrying about it. I will think about it more when I'm back from PTO. In the meanwhile, input is welcome.
* Removed TODO
* Replaced FiberRoot map with boolean
* Removed unnecessary whitespace edit
Andrew Clark
Dan Abramov
Mateusz Burzyński
Nicole Levy
Fumiya Shibusawa
Johan Henriksson
jddxf
Thibault Malbranche
Brian Vaughn
Philipp Spieß
Sen Yang
Moti Zilberman
dongyuwei
Brandon Dail
Chang Yan