Per title. This gives us much more granular memoization when the source used optional member expressions. Note that we only infer optional deps when the source used optionals: we don't (yet) infer optional dependencies from conditionals.
ghstack-source-id: 104d0b712d
Pull Request resolved: https://github.com/facebook/react/pull/30838
Handles an additional case as part of testing combinations of the same path being accessed in different places with different segments as optional/unconditional.
ghstack-source-id: ace777fcbb
Pull Request resolved: https://github.com/facebook/react/pull/30836
Updates PropagateScopeDeps and DeriveMinimalDeps to understand optional dependency paths (`a?.b`). There a few key pieces to this:
In PropagateScopeDeps we jump through some hoops to work around the awkward structure of nested OptionalExpressions. This is much easier in HIR form, but I managed to get this pretty close and i think it will be landable with further cleanup. A good chunk of this is avoiding prematurely registering a value as a dependency - there are a bunch of indirections in the ReactiveFunction structure:
```
t0 = OptionalExpression
SequenceExpression
t0 = Sequence
...
LoadLocal t0
```
Where if at any point we call `visitOperand()` we'll prematurely register a dependency instead of declareProperty(). The other bit is that optionals can be optional=false for nested member expressions where not all the parts are actually optional (`foo.bar?.bar.call()`). And of course, parts of an optional chain can still be conditional even when optional=true (for example the `x` in `foo.bar?.[x]?.baz`). Not all of this is tested yet so there are likely bugs still.
The other bit is DeriveMinimalDeps, which is thankfully easier. We add OptionalAccess and OptionalDep and update the merge and reducing logic for these cases. There is probably still more to update though, for things like merging subtrees. There are a lot of ternaries that assume a result can be exactly one of two states (conditional/unconditional, dependency/access) and these assumptions don't hold anymore. I'd like to refactor to dependency/access separate from conditional/optional/unconditional. Also, the reducing logic isn't quite right: once a child is optional we keep inferring all the parents as optional too, losing some precision. I need to adjust the reducing logic to let children decide whether their path token is optional or not.
ghstack-source-id: 207842ac64
Pull Request resolved: https://github.com/facebook/react/pull/30819
If the inferred deps are more precise (non-optional) than the manual deps (optional) it should pass validation.
The other direction also seems like it would be fine - inferring optional deps when the original was non-optional - but for now let's keep the "at least as precise" rule.
ghstack-source-id: 9f7a99ee5f
Pull Request resolved: https://github.com/facebook/react/pull/30816
Branch terminals didn't have a fallthrough because they correspond to an outer terminal (optional, logical, etc) that has the "real" fallthrough. But understanding how branch terminals correspond to these outer terminals requires knowing the branch fallthrough. For example, `foo?.bar?.baz` creates terminals along the lines of:
```
bb0:
optional fallthrough=bb4
bb1:
optional fallthrough=bb3
bb2:
...
branch ... (fallthrough=bb3)
...
bb3:
...
branch ... (fallthrough=bb4)
...
bb4:
...
```
Without a fallthrough on `branch` terminals, it's unclear that the optional from bb0 has its branch node in bb3. With the fallthroughs, we can see look for a branch with the same fallthrough as the outer optional terminal to match them up.
ghstack-source-id: d48c623289
Pull Request resolved: https://github.com/facebook/react/pull/30814
Adds an `optional: boolean` property to each token in a DependencyPath, currently always set to false. Also updates the equality and printing logic for paths to account for this field.
Subsequent PRs will update our logic to determine which manual dependencies were optional, then we can start inferring optional deps as well.
ghstack-source-id: 66c2da2cfa
Pull Request resolved: https://github.com/facebook/react/pull/30813
Previously the path of a ReactiveScopeDependency was `Array<string>`. We need to track whether each property access is optional or not, so as a first step we change this to `Array<{property: string}>`, making space for an additional property in a subsequent PR.
ghstack-source-id: c5d38d72f6
Pull Request resolved: https://github.com/facebook/react/pull/30812
AnalyzeFunctions was reusing the `ReactiveScopeDependency` type since it happened to have a convenient shape, but we need to change this type to represent optionality. We now use a locally defined type instead.
ghstack-source-id: e305c6ede4
Pull Request resolved: https://github.com/facebook/react/pull/30811
Summary:
This addresses the issue of the compiler being overly restrictive about refs escaping into object expressions. Rather than erroring whenever a ref flows into an object, we will now treat the object itself as a ref, and apply the same escape rules to it. Whenever we look up a property from a ref value, we now don't know whether that value is itself a ref or a ref value, so we assume it's both.
The same logic applies to ref-accessing functions--if such a function is stored in an object, we'll propagate that property to the object itself and any properties looked up from it.
ghstack-source-id: 5c6fcb895d
Pull Request resolved: https://github.com/facebook/react/pull/30821
If we see the "Maximum call stack size exceeded" error we know we've hit
stack overflow. We can recover from this by spawning a new task and
trying again. Effectively a zero-cost trampoline in the normal case. The
new task will have a clean stack. If you have a lot of siblings at the
same depth that hits the limit you can end up hitting this once for each
sibling but within that new sibling you're unlikely to hit this again.
So it's not too expensive.
If it errors again in the retryTask pass, the other error handling takes
over which causes this to be able to still not infinitely stall. E.g.
when the component itself throws an error like this.
It's still better to increase the stack limit for performance if you
have a really deep tree but it doesn't really hurt to be able to recover
since it's zero cost when it doesn't happen.
We could do the same thing for Flight. Those trees don't tend to be as
deep but could happen.
This loops over the remainingReconcilingChildren to find existing
FiberInstances that match the updated Fiber. This is the same thing we
already do for virtual instances. This avoids the need for a
`fiberToFiberInstanceMap`.
This loop is fast but there is a downside when the children set is very
large and gets reordered with keys since we might have to loop over the
set multiple times to get to the instances in the bottom. If that
becomes a problem we can optimize it the same way ReactChildFiber does
which is to create a temporary Map only when the children don't line up
properly. That way everything except the first pass can use the Map but
there's no need to create it eagerly.
Now that we have the loop we don't need the previousSibling field so we
can save some memory there.
These don't have their own time since they don't take up any time to
render but they show up in the tree for context. However they never
render themselves. Their base tree time is the base time of their
children. This way they take up the same space as their combined
children in the Profiler tree. (Instead of leaving a blank line which
they did before this PR.)
The frontend doesn't track the difference between a virtual instance and
a Fiber that didn't render this update. This might be a bit confusing as
to why it didn't render. I add the word "client" to make it a bit
clearer and works for both. We should probably have different verbiage
here based on it is a Server Component or something else.
<img width="1103" alt="Screenshot 2024-08-26 at 5 00 47 PM"
src="https://github.com/user-attachments/assets/87b811d4-7024-466a-845d-542493ed3ca2">
I also took the opportunity to remove idToTreeBaseDurationMap and
idToRootMap maps. Cloning the Map isn't really all that super fast
anyway and it means we have to maintain the map continuously as we
render. Instead, we can track it on the instances and then walk the
instances to create a snapshot when starting to profile. This isn't as
fast but really fast too and requires less bookkeeping while rendering
instead which is more sensitive than that one snapshot in the beginning.
## Summary
suspenseCallback feature has proved to be useful for FB Web. Let's look
at enabling the feature for the React Native build.
## How did you test this change?
Will sync the react changes with a React Native build and will verify
that performance logging is correctly notified of suspense promises
during the suspense callback.
We don't have the source location of Server Components on the client
because we don't want to eagerly do the throw trick for all Server
Components just in case. Unfortunately Node.js doesn't expose V8's API
to get a source location of a function.
We do have the owner stacks of the JSX that created it though and at
some point we'll also show that location in DevTools.
However, the realization is that if a Server Component is the owner of
any child. The owner stack of that child will have the owner component's
source location as its bottom stack frame.
The technique I'm implementing here is to track whenever a child mounts
we already have its owner. We track the first discovered owned child's
stack on the owner. Then when we ask for a Source location of the owner
do we parse that stack and extract the location of the bottom frame.
This doesn't give us a location necessarily in the top of the function
but somewhere in the function.
In this case the first owned child is the Container:
<img width="1107" alt="Screenshot 2024-08-22 at 10 24 42 PM"
src="https://github.com/user-attachments/assets/95f32850-24a5-4151-8ce6-b7b89db68aee">
<img width="648" alt="Screenshot 2024-08-22 at 10 24 20 PM"
src="https://github.com/user-attachments/assets/4bcba033-866f-4684-9beb-de09d189deff">
We can even use this technique for Fibers too. Currently I use this as a
fallback in case the error technique didn't work. This covers a case
where nothing errors but you still render a child. This case is actually
quite common:
```
function Foo() {
return <Bar />;
}
```
However, for Fibers we could really just use the `inspect(function)`
technique which works for all cases. At least in Chrome.
Unfortunately, this technique doesn't work if a Component doesn't create
any new JSX but just renders its children. It also currently doesn't
work if the child is filtered since I only look up the owner if an
instance is not filtered. This means that the container in the fixture
can't view source by default since the host component is filtered:
```
export default function Container({children}) {
return <div>{children}</div>;
}
```
<img width="1107" alt="Screenshot 2024-08-22 at 10 24 35 PM"
src="https://github.com/user-attachments/assets/c3f8f9c5-5add-4d35-9290-3a5079e82adc">
I noticed that there is a delay due to the inspection being split into
one part that gets the attribute and another eval that does the
inspection. This is a bit hacky and uses temporary global names that are
leaky. The timeout was presumably to ensure that the first step had
fully propagated but it's slow. As we've learned, it can be throttled,
and it isn't a guarantee either way.
Instead, we can just consolidate these into a single operation that
by-passes the bridge and goes straight to the renderer interface from
the eval.
I did the same for the viewElementSource helper even though that's not
currently in use since #28471 but I think we probably should return to
that technique when it's available since it's more reliable than the
throw - at least in Chrome. I'm not sure about the status of React
Native here. In Firefox, inspecting a function with source maps doesn't
seem to work. It doesn't jump to original code.
This allows us to handle common operations such as `useFragment(...).edges.nodes ?? []` where we have a `Phi(MixedReadonly, Array)`. The underlying pattern remains general-purpose and not Relay-specific, and any API that returns transitively "mixed" data (primitives, arrays, plain objects) can benefit from the same type refinement.
ghstack-source-id: 5128310894
Pull Request resolved: https://github.com/facebook/react/pull/30797
Redo of an earlier (pre-OSS) PR to infer types of phi nodes. There are a few pieces to this:
1. Update InferTypes to infer the type of `phi.id.type`, not the unused `phi.type`.
2. Update the algorithm to verify that all the phi types are actually equal, not just have the same kind.
3. Handle circular types by removing the cycle.
However, that reveals another issue: InferMutableRanges currently infers the results of `Store` effects _after_ its fixpoint loop. That was fine when a Store could never occur on a phi (since they wouldn't have a type to get a function signature from). Now though, we can have Store effects occur on phis, and we need to ensure that this correctly updates the mutable range of the phi operands - recursively. See new test that fails without the fixpoint loop.
ghstack-source-id: 2e1b02844d
Pull Request resolved: https://github.com/facebook/react/pull/30796
This is a complex case: we not only need phi type inference but also need to be able infer the union of `MixedReadonly | Array`.
ghstack-source-id: 935088910d
Pull Request resolved: https://github.com/facebook/react/pull/30793
This is a refactor of the fix in #27505.
When a transition update is scheduled by a popstate event, (i.e. a back/
forward navigation) we attempt to render it synchronously even though
it's a transition, since it's likely the previous page's data is cached.
In #27505, I changed the implementation so that it only "upgrades" the
priority of the transition for a single attempt. If the attempt
suspends, say because the data is not cached after all, from then on it
should be treated as a normal transition.
But it turns out #27505 did not work as intended, because it relied on
marking the root with pending synchronous work (root.pendingLanes),
which was never cleared until the popstate update completed.
The test scenarios I wrote accidentally worked for a different reason
related to suspending the work loop, which I'm currently in the middle
of refactoring.
This reverts commit b34b750729.
This hack doesn't play well internally so I'm reverting this for now
(but keeping the compilationMode override). I'll audit the locations we
report later and try to make them more accurate so we won't need this
workaround.
ghstack-source-id: b6be29c11d
Pull Request resolved: https://github.com/facebook/react/pull/30792
Currently you can jump to definition of a function by right clicking
through the context menu. However, it's pretty difficult to discover.
This makes the functions clickable to jump to definition - like links.
This uses the same styling as we do for links (which are btw only
clickable if they're not editable). Including cursor: pointer.
I added a background on hover which follows the same pattern as the
owners list.
I also dropped the ƒ prefix when displaying functions. This is a cute
short cut and there's precedence in how Chrome prints functions in the
console *if* the function's toString would've had a function prefix like
if it was a function declaration or expression. It does not do this for
arrow functions or object methods.
Elsewhere in the JS ecosystem this isn't really used anywhere. It
invites more questions than it answers.
The parenthesis and curlies are enough. There's no ambiguity here since
strings have quotations. It looks better with just its object method
form. Keeping it simple seems best. To my eyes this flows better because
I'm used to looking at function syntax but not weird "f"s.
Before:
<img width="433" alt="Screenshot 2024-08-20 at 11 55 09 PM"
src="https://github.com/user-attachments/assets/9dd50da6-598f-4291-9e24-1cdc7200dc9e">
After:
<img width="388" alt="Screenshot 2024-08-20 at 11 46 01 PM"
src="https://github.com/user-attachments/assets/dd988e14-412e-4deb-8c8c-26a54be8331f">
After (Hover):
<img width="389" alt="Screenshot 2024-08-20 at 11 46 31 PM"
src="https://github.com/user-attachments/assets/6fb4ebed-5dc1-448a-8e4d-b6d4f3903329">
Stacked on #30758 and #30755.
This is copy paste from #30755 into the ESM package. We use the
`webpack-sources` package for the source map utility but it's not
actually dependent on Webpack itself. Could probably inline it in the
build.
Uses the returnIdentifier added in the previous PR to provide a stable identifier for which we can infer a return type for functions, then wires up the equations in InferTypes to infer the type.
ghstack-source-id: 22c0a9ea09
Pull Request resolved: https://github.com/facebook/react/pull/30785
If you have a function expression which _captures_ a mutable value (but does not mutate it), and that function is invoked during render, we infer the invocation as a mutation of the captured value. But in some circumstances we can prove that the captured value cannot have been mutated, and could in theory avoid inferring a mutation.
ghstack-source-id: 47664e48ce
Pull Request resolved: https://github.com/facebook/react/pull/30783
Adds a new Environment config option which allows specifying a function that is called to resolve types of imported modules. The function is passed the name of the imported module (the RHS of the import stmt) and can return a TypeConfig, which is a recursive type of the following form:
* Object of valid identifier keys (or "*" for wildcard) and values that are TypeConfigs
* Function with various properties, whose return type is a TypeConfig
* or a reference to a builtin type using one of a small list (currently Ref, Array, MixedReadonly, Primitive)
Rather than have to eagerly supply all known types (most of which may not be used) when creating the config, this function can do so lazily. During InferTypes we call `getGlobalDeclaration()` to resolve global types. Originally this was just for known react modules, but if the new config option is passed we also call it to see if it can resolve a type. For `import {name} from 'module'` syntax, we first resolve the module type and then call `getPropertyType(moduleType, 'name')` to attempt to retrieve the property of the module (the module would obviously have to be typed as an object type for this to have a chance of yielding a result). If the module type is returned as null, or the property doesn't exist, we fall through to the original checking of whether the name was hook-like.
TODO:
* testing
* cache the results of modules so we don't have to re-parse/install their types on each LoadGlobal of the same module
* decide what to do if the module types are invalid. probably better to fatal rather than bail out, since this would indicate an invalid configuration.
ghstack-source-id: bfdbf67e3d
Pull Request resolved: https://github.com/facebook/react/pull/30771
The fixture from the previous PR was getting inconsistent behavior because of the following:
1. Create an object in a useMemo
2. Create a callback in a useCallback, where the callback captures the object from (1) into a local object, then passes that local object into a logging method. We have to assume the logging method could modify the local object, and transitively, the object from (1).
3. Call the callback during render.
4. Pass the callback to JSX.
We correctly infer that the object from (1) is captured and modified in (2). However, in (4) we transitively freeze the callback. When transitively freezing functions we were previously doing two things: updating our internal abstract model of the program values to reflect the values as being frozen *and* also updating function operands to change their effects to freeze.
As the case above demonstrates, that can clobber over information about real potential mutability. The potential fix here is to only walk our abstract value model to mark values as frozen, but _not_ override operand effects. Conceptually, this is a forward data flow propagation — but walking backward to update effects is pushing information backwards in the algorithm. An alternative would be to mark that data was propagated backwards, and trigger another loop over the CFG to propagate information forward again given the updated effects. But the fix in this PR is more correct.
ghstack-source-id: c05e716f37
Pull Request resolved: https://github.com/facebook/react/pull/30766
This fixture bails out on ValidatePreserveExistingMemo but would ideally memoize since the original memoization is safe. It's trivial to make it pass by commenting out the commented line (`LogEvent.log(() => object)`). I would expect the compiler to infer this as possible mutation of `logData`, since `object` captures a reference to `logData`. But somehow `logData` is getting memoized successfully, but we still infer the callback, `setCurrentIndex`, as having a mutable range that extends to the `setCurrentIndex()` call after the useCallback.
ghstack-source-id: 4f82e34510
Pull Request resolved: https://github.com/facebook/react/pull/30764
DevTools shouldn't use react-is since that's versioned to one version of
React. We don't need to since we use all the symbols from
shared/ReactSymbols anyway and have a fork of typeOf that can cover
both.
Now JSX of old React versions show up with proper JSX formatting when
inspecting.
These are only needed internally so I'm opting to just do it in the
commit artifacts job instead of amending the build config.
ghstack-source-id: 6a5382b028
Pull Request resolved: https://github.com/facebook/react/pull/30775
Similar to https://github.com/facebook/react/pull/30768 we want to
schedule work during prerendering in microtasks both for the root task
and pings. We continue to schedule flushes as Tasks to allow as much
work to be batched up as possible.
The unbundled form is just a way to show case a prototype for how an
unbundled version of RSC can work. It's not really intended for every
bundler combination to provide such a configuration.
There's no configuration of Turbopack that supports this mode atm and
possibly never will be since it's more of an integrated server/client
experience.
This removes the unbundled form and node register/loaders from the
turbopack build.
Follow up to #30741.
This is just for the reference Webpack implementation.
If there is a source map associated with a Node ESM loader, we generate
new source map entries for every `registerServerReference` call.
To avoid messing too much with it, this doesn't rewrite the original
mappings. It just reads them while finding each of the exports in the
original mappings. We need to read all since whatever we append at the
end is relative. Then we just generate new appended entries at the end.
For the location I picked the location of the local name identifier.
Since that's the name of the function and that gives us a source map
name index. It means it jumps to the name rather than the beginning of
the function declaration. It could be made more clever like finding a
local function definition if it is reexported. We could also point to
the line/column of the function declaration rather than the identifier
but point to the name index of the identifier name.
Now jumping to definition works in the fixture.
<img width="574" alt="Screenshot 2024-08-20 at 2 49 07 PM"
src="https://github.com/user-attachments/assets/7710f0e6-2cee-4aad-8d4c-ae985f8289eb">
Unfortunately this technique doesn't seem to work in Firefox nor Safari.
They don't apply the source map for jumping to the definition.
In https://github.com/facebook/react/pull/29491 I updated the work
scheduler for Flight to use microtasks to perform work when something
pings. This is useful but it does have some downsides in terms of our
ability to do task prioritization. Additionally the initial work is not
instantiated using a microtask which is inconsistent with how pings
work.
In this change I update the scheduling logic to use microtasks
consistently for prerenders and use regular tasks for renders both for
the initial work and pings.
Shortcut for the common case where only a single flag is checked. Same
as `gate(flags => flags.enableFeatureFlag)`.
Normally I don't care about these types of conveniences but I'm about to
add a lot more inline flag checks these all over our tests and it gets
noisy. This helps a bit.
When we introduced prerendering for flight we modeled an abort of a
flight prerender as having unfinished rows. This is similar to how
postpone was already implemented when you postponed from "within" a
prerender using React.unstable_postpone. However when aborting with a
postponed instance every boundary would be eagerly marked for client
rendering which is more akin to prerendering and then resuming with an
aborted signal.
The insight with the flight work was that it's not so much the postpone
that describes the intended semantics but the abort combined with a
prerender. So like in flight when you abort a prerender and enableHalt
is enabled boundaries and the shell won't error for any reason. Fizz
will still call onPostpone and onError according to the abort reason but
the consuemr of the prerender should expect to resume it before trying
to use it.
When aborting a prerender we should leave references unfulfilled, not
share a common unfullfilled reference. functionally today this doesn't
matter because we don't have resuming but the semantic is that the row
was not available when the abort happened and in a resume the row should
fill in. But by pointing each task to a common unfulfilled chunk we lose
the ability for these references to resolves to distinct values on
resume.
When aborting with a postpone value boundaries are put into client
rendered mode even during prerenders. This doesn't follow the postpoen
semantics of the rest of fizz where during a prerender a postpone is
tracked and it will leave holes in tracked postpone state that can be
resumed. This change updates this behavior to match the postpones
semantics between aborts and imperative postpones.
Noticed this from #30707. This was vestigial from from circleci and now
that we're on GH actions I think we should be able to remove this option
altogether.
ghstack-source-id: 78e8b0243b
Pull Request resolved: https://github.com/facebook/react/pull/30753
Joe Savona
Mike Vitousek
Sebastian Markbåge
Benoit Girard
Lauren Tan
Andrew Clark
Josh Story
Sebastian "Sebbie" Silbermann