When we made stylesheets suspend even during high priority updates we
exposed a bug in the loading tracking of stylesheets that are loaded as
part of the preamble. This allowed these stylesheets to put suspense
boundaries into fallback mode more often than expected because cases
where a stylesheet was server rendered could now cause a fallback to
trigger which was never intended to happen.
This fix updates resource construction to evaluate whether the instance
exists in the DOM prior to construction and if so marks the resource as
loaded and inserted.
One ambiguity that needed to be solved still is how to tell whether a
stylesheet rendered as part of a late Suspense boundary reveal is
already loaded. I updated the instruction to clear out the loading
promise after successfully loading. This is useful because later if we
encounter this same resource again we can avoid the microtask if it is
already loaded. It also means that we can concretely understand that if
a stylesheet is in the DOM without this marker then it must have loaded
(or errored) already.
Stacked on #29551
Flight pings much more often than Fizz because async function components
will always take at least a microtask to resolve . Rather than
scheduling this work as a new macrotask Flight now schedules pings in a
microtask. This allows more microtasks to ping before actually doing a
work flush but doesn't force the vm to spin up a new task which is quite
common give n the nature of Server Components
Host Components can exist as four semantic types
1. regular Components (Vanilla obv)
2. singleton Components
2. hoistable components
3. resources
Each of these component types have their own rules related to mounting
and reconciliation however they are not direclty modeled as their own
unique fiber type. This is partly for code size but also because
reconciling the inner type of these components would be in a very hot
path in fiber creation and reconciliation and it's just not practical to
do this logic check here.
Right now we have three Fiber types used to implement these 4 concepts
but we probably need to reconsider the model and think of Host
Components as a single fiber type with an inner implementation. Once we
do this we can regularize things like transitioning between a resource
and a regular component or a singleton and a hoistable instance. The
cases where these transitions happen today aren't particularly common
but they can be observed and currently the handling of these transitions
is incomplete at best and buggy at worst. The most egregious case is the
link type. This can be a regular component (stylesheet without
precedence) a hoistable component (non stylesheet link tags) or a
resource (stylesheet with a precedence) and if you have a single jsx
slot that tries to reconcile transitions between these types it just
doesn't work well.
This commit adds an error for when a Hoistable goes from Instance to
Resource. Currently this is only possible for `<link>` elements going to
and from stylesheets with precedence. Hopefully we'll be able to remove
this error and implement as an inner type before we encounter new
categories for the Hoistable types
detecting type shifting to and from regular components is harder to do
efficiently because we don't want to reevaluate the type on every update
for host components which is currently not required and would add
overhead to a very hot path
singletons can't really type shift in their one practical implementation
(DOM) so they are only a problem in theroy not practice
This one should be fully behind the `enableOwnerStacks` flag.
Instead of printing the parent Component stack all the way to the root,
this now prints the owner stack of every JSX callsite. It also includes
intermediate callsites between the Component and the JSX call so it has
potentially more frames. Mainly it provides the line number of the JSX
callsite. In terms of the number of components is a subset of the parent
component stack so it's less information in that regard. This is usually
better since it's more focused on components that might affect the
output but if it's contextual based on rendering it's still good to have
parent stack. Therefore, I still use the parent stack when printing DOM
nesting warnings but I plan on switching that format to a diff view
format instead (Next.js already reformats the parent stack like this).
__Follow ups__
- Server Components show up in the owner stack for client logs but logs
done by Server Components don't yet get their owner stack printed as
they're replayed. They're also not yet printed in the server logs of the
RSC server.
- Server Component stack frames are formatted as the server and added to
the end but this might be a different format than the browser. E.g. if
server is running V8 and browser is running JSC or vice versa. Ideally
we can reformat them in terms of the client formatting.
- This doesn't yet update Fizz or DevTools. Those will be follow ups.
Fizz still prints parent stacks in the server side logs. The stacks
added to user space `console.error` calls by DevTools still get the
parent stacks instead.
- It also doesn't yet expose these to user space so there's no way to
get them inside `onCaughtError` for example or inside a custom
`console.error` override.
- In another follow up I'll use `console.createTask` instead and
completely remove these stacks if it's available.
Updates Suspensey instances and resources to preload even during urgent
updates and to potentially suspend.
The current implementation is unchanged for transitions but for sync
updates if there is a suspense boundary above the resource/instance it
will be rendered in fallback mode instead.
Note: This behavior is not what we want for images once we make them
suspense enabled. We will need to have forked behavior here to
distinguish between stylesheets which should never commit when not
loaded and images which should commit after a small delay
Before this change, `useFormStatus` is only activated if a form is
submitted by an action function (either `<form action={actionFn}>` or
`<button formAction={actionFn}>`).
After this change, `useFormStatus` will also be activated if you call
`startTransition(actionFn)` inside a submit event handler that is
`preventDefault`-ed.
This is the last missing piece for implementing a custom `action` prop
that is progressively enhanced using `onSubmit` while maintaining the
same behavior as built-in form actions.
Here's the basic recipe for implementing a progressively-enhanced form
action. This would typically be implemented in your UI component
library, not regular application code:
```js
import {requestFormReset} from 'react-dom';
// To implement progressive enhancement, pass both a form action *and* a
// submit event handler. The action is used for submissions that happen
// before hydration, and the submit handler is used for submissions that
// happen after.
<form
action={action}
onSubmit={(event) => {
// After hydration, we upgrade the form with additional client-
// only behavior.
event.preventDefault();
// Manually dispatch the action.
startTransition(async () => {
// (Optional) Reset any uncontrolled inputs once the action is
// complete, like built-in form actions do.
requestFormReset(event.target);
// ...Do extra action-y stuff in here, like setting a custom
// optimistic state...
// Call the user-provided action
const formData = new FormData(event.target);
await action(formData);
});
}}
/>
```
This follows the same principle as in #28611.
We cannot serialize Blobs of a form data into HTML because you can't
initialize a file input to some value. However the serialization of
state in an Action can contain blobs. In this case we do error but
outside the try/catch that recovers to error to client replaying instead
of MPA mode. This errors earlier to ensure that this works.
Testing this is a bit annoying because JSDOM doesn't have any of the
Blob methods but the Blob needs to be compatible with FormData and the
FormData needs to be compatible with `<form>` nodes in these tests. So I
polyfilled those in JSDOM with some hacks.
A possible future enhancement would be to encode these blobs in a base64
mode instead and have some way to receive them on the server. It's just
a matter of layering this. I think the RSC layer's `FORM_DATA`
implementation can pass some flag to encode as base64 and then have
decodeAction include some way to parse them. That way this case would
work in MPA mode too.
This PR reorganizes the `react-dom` entrypoint to only pull in code that
is environment agnostic. Previously if you required anything from this
entrypoint in any environment the entire client reconciler was loaded.
In a prior release we added a server rendering stub which you could
alias in server environments to omit this unecessary code. After landing
this change this entrypoint should not load any environment specific
code.
While a few APIs are truly client (browser) only such as createRoot and
hydrateRoot many of the APIs you import from this package are only
useful in the browser but could concievably be imported in shared code
(components running in Fizz or shared components as part of an RSC app).
To avoid making these require opting into the client bundle we are
keeping them in the `react-dom` entrypoint and changing their
implementation so that in environments where they are not particularly
useful they do something benign and expected.
#### Removed APIs
The following APIs are being removed in the next major. Largely they
have all been deprecated already and are part of legacy rendering modes
where concurrent features of React are not available
* `render`
* `hydrate`
* `findDOMNode`
* `unmountComponentAtNode`
* `unstable_createEventHandle`
* `unstable_renderSubtreeIntoContainer`
* `unstable_runWithPrioirty`
#### moved Client APIs
These APIs were available on both `react-dom` (with a warning) and
`react-dom/client`. After this change they are only available on
`react-dom/client`
* `createRoot`
* `hydrateRoot`
#### retained APIs
These APIs still exist on the `react-dom` entrypoint but have normalized
behavior depending on which renderers are currently in scope
* `flushSync`: will execute the function (if provided) inside the
flushSync implemention of FlightServer, Fizz, and Fiber DOM renderers.
* `unstable_batchedUpdates`: This is a noop in concurrent mode because
it is now the only supported behavior because there is no legacy
rendering mode
* `createPortal`: This just produces an object. It can be called from
anywhere but since you will probably not have a handle on a DOM node to
pass to it it will likely warn in environments other than the browser
* preloading APIS such as `preload`: These methods will execute the
preload across all renderers currently in scope. Since we resolve the
Request object on the server using AsyncLocalStorage or the current
function stack in practice only one renderer should act upon the
preload.
In addition to these changes the server rendering stub now just rexports
everything from `react-dom`. In a future minor we will add a warning
when using the stub and in the next major we will remove the stub
altogether
stacked on #28870
inline script children have been encoded as HTML for a while now but
this can easily break script parsing so practically if you were
rendering inline scripts you were using dangerouslySetInnerHTML. This is
not great because now there is no escaping at all so you have to be even
more careful. While care should always be taken when rendering untrusted
script content driving users to use dangerous APIs is not the right
approach and in this PR the escaping functionality used for
bootstrapScripts and importMaps is being extended to any inline script.
the approach is to escape 's' or 'S" with the appropriate unicode code
point if it is inside a <script or </script sequence. This has the nice
benefit of minimally escaping the text for readability while still
preserving full js parsing capabilities. As articulated when we
introduced this escaping for prior use cases this is only safe because
we are escaping the entire script content. It would be unsafe if we were
not escaping the entirety of the script because we would no longer be
able to ensure there are no earlier or later <script sequences that put
the parser in unexpected states.
style text content has historically been escaped as HTML which is
non-sensical and often leads users to using dangerouslySetInnerHTML as a
matter of course. While rendering untrusted style rules is a security
risk React doesn't really provide any special protection here and
forcing users to use a completely unescaped API is if anything worse. So
this PR updates the style escaping rules for Fizz to only escape the
text content to ensure the tag scope cannot be closed early. This is
accomplished by encoding "s" and "S" as hexadecimal unicode
representation "\73 " and "\53 " respectively when found within a
sequence like </style>. We have to be careful to support casing here
just like with the script closing tag regex for bootstrap scripts.
Based on:
- #28804
---
This sets adds a new ReactDOM export called requestFormReset, including
setting up the export and creating a method on the internal ReactDOM
dispatcher. It does not yet add any implementation.
Doing this in its own commit for review purposes.
The API itself will be explained in the next PR.
This updates the behavior of form actions to automatically reset the
form's uncontrolled inputs after the action finishes.
This is a frequent feature request for people using actions and it
aligns the behavior of client-side form submissions more closely with
MPA form submissions.
It has no impact on controlled form inputs. It's the same as if you
called `form.reset()` manually, except React handles the timing of when
the reset happens, which is tricky/impossible to get exactly right in
userspace.
The reset shouldn't happen until the UI has updated with the result of
the action. So, resetting inside the action is too early.
Resetting in `useEffect` is better, but it's later than ideal because
any effects that run before it will observe the state of the form before
it's been reset.
It needs to happen in the mutation phase of the transition. More
specifically, after all the DOM mutations caused by the transition have
been applied. That way the `defaultValue` of the inputs are updated
before the values are reset. The idea is that the `defaultValue`
represents the current, canonical value sent by the server.
Note: this change has no effect on form submissions that aren't
triggered by an action.
`<noscript>` scopes should be considered inert from the perspective of
Fizz since we assume they'll only be used in rare and adverse
circumstances. When we added preload support for img tags we did not
include the noscript scope check in the opt-out for preloading. This
change adds it in
fixes: #27910
This is similar to #28771 but for isomorphic. We need a make over for
these dispatchers anyway so this is the first step. Also helps flush out
some internals usage that will break anyway.
It flattens the inner mutable objects onto the ReactSharedInternals.
Treat async (boolean prop) consistently with Float. Previously float
checked if `props.async === true` (or not true) but the rest of
react-dom considers anything truthy that isn't a function or symbol as
`true`. This PR normalizes the Float behavior.
Stacked on #28751
Historically explicit hydration scheduling used the reconciler's update
priority to schedule the hydration. There was a lingering todo to switch
to using event priority in the absence of an explicit update priority.
This change updates the hydration priority by referring to the event
priority if no update priority is set
Stacked on #28771
ReactDOMCurrentDispatcher has longer property names for various methods.
These methods are only ever called internally and don't need to be
represented with as many characters. This change shortens the names and
aligns them with the hint codes we use in Flight. This alignment is
passive since not all dispatcher methods will exist as flight
instructions but where they can line up it seems reasonable to make them
do so
Stacked on #28751
ReactDOMSharedInternals uses properties of considerable length to model
mutuable state. These properties are not mangled during minification and
contribute a not insigificant amount to the uncompressed bundle size and
to a lesser degree compressed bundle size.
This change rewrites the DOMInternals in a way that shortens property
names so we can have smaller builds.
It also treats the entire object as a mutable container rather than
having different mutable sub objects.
The same treatment should be given to ReactSharedInternals
This PR moves `flushSync` out of the reconciler. there is still an
internal implementation that is used when these semantics are needed for
React methods such as `unmount` on roots.
This new isomorphic `flushSync` is only used in builds that no longer
support legacy mode.
Additionally all the internal uses of flushSync in the reconciler have
been replaced with more direct methods. There is a new
`updateContainerSync` method which updates a container but forces it to
the Sync lane and flushes passive effects if necessary. This combined
with flushSyncWork can be used to replace flushSync for all instances of
internal usage.
We still maintain the original flushSync implementation as
`flushSyncFromReconciler` because it will be used as the flushSync
implementation for FB builds. This is because it has special legacy mode
handling that the new isomorphic implementation does not need to
consider. It will be removed from production OSS builds by closure
though
Currently updatePriority is tracked in the reconciler. `flushSync` is
going to be implemented reconciler agnostic soon and we need to move the
tracking of this state to the renderer and out of reconciler. This
change implements new renderer bin dings for getCurrentUpdatePriority
and setCurrentUpdatePriority.
I was originally going to have the getter also do the event priority
defaulting using window.event so we eliminate getCur rentEventPriority
but this makes all the callsites where we store the true current
updatePriority on the stack harder to work with so for now they remain
separate.
I also moved runWithPriority to the renderer since it really belongs
whereever the state is being managed and it is only currently exposed in
the DOM renderer.
Additionally the current update priority is not stored on
ReactDOMSharedInternals. While not particularly meaningful in this
change it opens the door to implementing `flushSync` outside of the
reconciler
In React DOM, in general, we don't differentiate between `null` and
`undefined` because we expect to target DOM APIs. When we're setting a
property on a Custom Element, in the new heuristic, the goal is to allow
passing whatever data type instead of normalizing it. Switching between
`undefined` and `null` as an explicit value should therefore be
respected.
However, in this mode if `undefined` is used for the initial value, we
don't actually set the property at all. If passing `null` we will now
initialize it to the value `null`. Meaning `undefined` kind of
represents the default.
### Removing Properties
There is a pretty complex edge case which is what should happen when a
prop used to exist but was removed from the props object. This doesn't
have any kind of defined semantics. It really should mean - return to
"default". Because in the declarative world it means the same as if it
was just created - i.e. we can't just leave it as it was.
The closest might be `delete object.property` but that's not really the
intended way that properties on custom elements / classes are supposed
to operate. Additionally, for a property to even hit our heuristic it
must pass the `in` test and must exist to being with so the default must
have a value.
Since the point of these properties is to contain any kind of type,
there isn't really a conceptual default value. E.g. a numeric default
value might be zero `0` while a default string might be empty `""` and
default object might `null`. Additionally, the conceptual default can
really be initialized to anything. There's also varied precedence in the
ecosystem here and really no consensus. Anything we pick would be kind
of wrong, so we used to just pick `null`.
_The safest way to consume a Custom Element is to always pass the same
set of props._
JS does have a concept of a "default value" though and that is described
as the value `undefined`. That's why default argument / object property
initializers are initialized if the value is `undefined`.
The problem with using `undefined` as value is that [you shouldn't
actually ever set the value of a class property to
`undefined`](https://twitter.com/sebmarkbage/status/1774082540296388752).
A property should always be initialized to some value. It can't be left
missing and shouldn't be initialized to the value `undefined` for hidden
class optimizations. If we just mutate it to be `undefined` it would be
potentially bad for perf and shouldn't really be the value after
removing property - it should be returned to default.
Every property should really have a setter to be useful since it is what
is used to trigger reactivity when it changes. Sometimes you can just
use the properties passively when something else happens but most of the
time it should be a setter but to reach parity with DOM it should really
be always so that the active value can be normalized.
Those setters can have default argument initializers to represent what
the default value should be. Therefore Custom Element properties should
be used like this:
```js
class CustomElement extends HTMLElement {
_textLabel = '';
_price = 0;
_items = null;
constructor() {
super();
}
set textLabel(value = '') {
this._textLabel = value;
}
get textLabel() {
return this._textLabel;
}
set price(value = 0) {
this._price = value;
}
get price() {
return this._price;
}
set items(value = null) {
this._items = value;
}
get items() {
return this._items;
}
}
```
The default initializer can be used to initialize a value back to its
original default when `undefined` is passed to it. Therefore, we pass
`undefined`, not because we expect that to be the value of a property
but because that's the value that represents "return to default".
This fixes#28203 but not really for the reason specified in the issue.
We don't expect you to actually store the `undefined` value but to use a
setter to set the property to something else that represents the
default. When we initialize the element the first time, we won't set
anything if it's the value `undefined` so we assume that the property
initializers running in the constructor is going to set the same default
value as if we set the property to `undefined`.
cc @josepharhar
We previously only included the component stack.
Cleaned up the fields in Fizz server that wasn't using consistent hidden
classes in dev vs prod.
Added a prefix to errors serialized from server rendering. It can be a
bit confusing to see where this error came from otherwise since it
didn't come from elsewhere on the client. It's really kind of confusing
with other recoverable errors that happen on the client too.
We've rolled out this flag internally on WWW. This PR removed flag
`enableCustomElementPropertySupport`
Test plan:
-- `yarn test`
Co-authored-by: Ricky Hanlon <rickhanlonii@gmail.com>
## Overview
This has landed, so we can remove the flag
## Changelog
This change blocks using javascript URLs such as:
```html
<a href="javascript:notfine">p0wned</a>
```
We previously announced dropping support for this via a warning:
> A future version of React will block javascript: URLs as a security
precaution. Use event handlers instead if you can. If you need to
generate unsafe HTML try using dangerouslySetInnerHTML instead.
Stacked on top of #28498 for test fixes.
### Don't Rethrow
When we started React it was 1:1 setState calls a series of renders and
if they error, it errors where the setState was called. Simple. However,
then batching came and the error actually got thrown somewhere else.
With concurrent mode, it's not even possible to get setState itself to
throw anymore.
In fact, all APIs that can rethrow out of React are executed either at
the root of the scheduler or inside a DOM event handler.
If you throw inside a React.startTransition callback that's sync, then
that will bubble out of the startTransition but if you throw inside an
async callback or a useTransition we now need to handle it at the hook
site. So in 19 we need to make all React.startTransition swallow the
error (and report them to reportError).
The only one remaining that can throw is flushSync but it doesn't really
make sense for it to throw at the callsite neither because batching.
Just because something rendered in this flush doesn't mean it was
rendered due to what was just scheduled and doesn't mean that it should
abort any of the remaining code afterwards. setState is fire and forget.
It's send an instruction elsewhere, it's not part of the current
imperative code.
Error boundaries never rethrow. Since you should really always have
error boundaries, most of the time, it wouldn't rethrow anyway.
Rethrowing also actually currently drops errors on the floor since we
can only rethrow the first error, so to avoid that we'd need to call
reportError anyway. This happens in RN events.
The other issue with rethrowing is that it logs an extra console.error.
Since we're not sure that user code will actually log it anywhere we
still log it too just like we do with errors inside error boundaries
which leads all of these to log twice.
The goal of this PR is to never rethrow out of React instead, errors
outside of error boundaries get logged to reportError. Event system
errors too.
### Breaking Changes
The main thing this affects is testing where you want to inspect the
errors thrown. To make it easier to port, if you're inside `act` we
track the error into act in an aggregate error and then rethrow it at
the root of `act`. Unlike before though, if you flush synchronously
inside of act it'll still continue until the end of act before
rethrowing.
I expect most user code breakages would be to migrate from `flushSync`
to `act` if you assert on throwing.
However, in the React repo we also have `internalAct` and the
`waitForThrow` helpers. Since these have to use public production
implementations we track these using the global onerror or process
uncaughtException. Unlike regular act, includes both event handler
errors and onRecoverableError by default too. Not just render/commit
errors. So I had to account for that in our tests.
We restore logging an extra log for uncaught errors after the main log
with the component stack in it. We use `console.warn`. This is not yet
ignorable if you preventDefault to the main error event. To avoid
confusion if you don't end up logging the error to console I just added
`An error occurred`.
### Polyfill
All browsers we support really supports `reportError` but not all test
and server environments do, so I implemented a polyfill for browser and
node in `shared/reportGlobalError`. I don't love that this is included
in all builds and gets duplicated into isomorphic even though it's not
actually needed in production. Maybe in the future we can require a
polyfill for this.
### Follow Ups
In a follow up, I'll make caught vs uncaught error handling be
configurable too.
---------
Co-authored-by: Ricky Hanlon <rickhanlonii@gmail.com>
Stacked on #28502.
This builds on the mechanism in #28502 by adding a diff of everything
we've collected so far to the thrown error or logged error.
This isn't actually a longest common subsequence diff. This means that
there are certain cases that can appear confusing such as a node being
added/removed when it really would've appeared later in the list. In
fact once a node mismatches, we abort rendering so we don't have the
context of what would've been rendered. It's not quite right to use the
result of the recovery render because it can use client-only code paths
using useSyncExternalStore which would yield false differences. That's
why diffing the HTML isn't quite right.
I also present abstract components in the stack, these are presented
with the client props and no diff since we don't have the props that
were on the server. The lack of difference might be confusing but it's
useful for context.
The main thing that's data new here is that we're adding some siblings
and props for context.
Examples in the [snapshot
commit](https://github.com/facebook/react/pull/28512/commits/e14532fd8df858a319d3eca687d83227209a498c).
Stacked on #28458.
This doesn't actually really change the messages yet, it's just a
refactor.
Hydration warnings can be presented either as HTML or React JSX format.
If presented as HTML it makes more sense to make that a DOM specific
concept, however, I think it's actually better to present it in terms of
React JSX.
Most of the time the errors aren't going to be something messing with
them at the HTML/HTTP layer. It's because the JS code does something
different. Most of the time you're working in just React. People don't
necessarily even know what the HTML form of it looks like. So this takes
the approach that the warnings are presented in React JSX in their rich
object form.
Therefore, I'm moving the approach to yield diff data to the reconciler
but it's the reconciler that's actually printing all the warnings.
While Meta is still using legacy mode and we can't remove completely,
Meta is not using legacy hydration so we should be able to remove that.
This is just the first step. Once removed, we can vastly simplify the
DOMConfig for hydration.
This will have to be rebased when tests are upgraded.
This bumps the canary versions to v19 to communicate that the next
release will be a major. Once this lands, we can start merging breaking
changes into `main`.
This a follow up to #28564. It's alternative to #28609 which takes
#28610 into account.
It used to be possible to return JSX from an action with
`useActionState`.
```js
async function action(errors, payload) {
"use server";
try {
...
} catch (x) {
return <div>Error message</div>;
}
}
```
```js
const [errors, formAction] = useActionState(action);
return <div>{errors}</div>;
```
Returning JSX from an action is itself not anything problematic. It's
that it also has to return the previous state to the action reducer
again that's the problem. When this happens we accidentally could
serialize an Element back to the server.
I fixed this in #28564 so it's now blocked if you don't have a temporary
reference set.
However, you can't have that for the progressive enhancement case. The
reply is eagerly encoded as part of the SSR render. Typically you
wouldn't have these in the initial state so the common case is that they
show up after the first POST back that yields an error and it's only in
the no-JS case where this happens so it's hard to discover.
As noted in #28609 there's a security implication with allowing elements
to be sent across this kind of payload, so we can't just make it work.
When an error happens during SSR our general policy is to try to recover
on the client instead. After all, SSR is mainly a perf optimization in
React terms and it's not primarily intended for a no JS solution.
This PR takes the approach that if we fail to generate the progressive
enhancement payload. I.e. if the serialization of previous state /
closures throw. Then we fallback to the replaying semantics just client
actions instead which will succeed.
The effect of this is that this pattern mostly just works:
- First render in the typical case doesn't have any JSX in it so it just
renders a progressive enhanced form.
- If JS fails to hydrate or you click early we do a form POST. If that
hits an error and it tries to render it using JSX, then the new page
will render successfully - however this time with a Replaying form
instead.
- If you try to submit the form again it'll have to be using JS.
Meaning if you use JSX as the error return value of form state and you
make a first attempt that fails, then no JS won't work because either
the first or second attempt has to hydrate.
We have ideas for potentially optimizing away serializing unused
arguments like if you don't actually use previous state which would also
solve it but it wouldn't cover all cases such as if it was deeply nested
in complex state.
Another approach that I considered was to poison the prev state if you
passed an element back but let it through to the action but if you try
to render the poisoned value, it wouldn't work. The downside of this is
when to error. Because in the progressive enhancement case it wouldn't
error early but when you actually try to invoke it at which point it
would be too late to fallback to client replaying. It would probably
have to always error even on the client which is unfortunate since this
mostly just works as long as it hydrates.
A while back we implemented a heuristic that if a chunk was large it was
assumed to be produced by the render and thus was safe to stream which
results in transferring the underlying object memory. Later we ran into
an issue where a precomputed chunk grew large enough to trigger this
hueristic and it started causing renders to fail because once a second
render had occurred the precomputed chunk would not have an underlying
buffer of bytes to send and these bytes would be omitted from the
stream. We implemented a technique to detect large precomputed chunks
and we enforced that these always be cloned before writing.
Unfortunately our test coverage was not perfect and there has been for a
very long time now a usage pattern where if you complete a boundary in
one flush and then complete a boundary that has stylehsheet dependencies
in another flush you can get a large precomputed chunk that was not
being cloned to be sent twice causing streaming errors.
I've thought about why we even went with this solution in the first
place and I think it was a mistake. It relies on a dev only check to
catch paired with potentially version specific order of operations on
the streaming side. This is too unreliable. Additionally the low limit
of view size for Edge is not used in Node.js but there is not real
justification for this.
In this change I updated the view size for edge streaming to match Node
at 2048 bytes which is still relatively small and we have no data one
way or another to preference 512 over this. Then I updated the assertion
logic to error anytime a precomputed chunk exceeds the size. This
eliminates the need to clone these chunks by just making sure our view
size is always larger than the largest precomputed chunk we can possibly
write. I'm generally in favor of this for a few reasons.
First, we'll always know during testing whether we've violated the limit
as long as we exercise each stream config because the precomputed chunks
are created in module scope. Second, we can always split up large chunks
so making sure the precomptued chunk is smaller than whatever view size
we actually desire is relatively trivial.
## Overview
Adds a `pending` state to useFormState, which will be replaced by
`useActionState` in the next diff. We will keep `useFormState` around
for backwards compatibility, but functionally it will work the same as
`useActionState`, which has an `isPending` state returned.
We broke the ability to "break on uncaught exceptions" by adding a
try/catch higher up in the scheduling. We're giving up on fixing that so
we can remove the replay trick inside an event handler.
The issue with that approach is that we end up double logging a lot of
errors in DEV since they get reported to the page.
It's also a lot of complexity around this feature.
The idea here is that host dispatchers are not bound to renders so we
need to be able to dispatch to them at any time. This updates the
implementation to chain these dispatchers so that each renderer can
respond to the dispatch. Semantically we don't always want every
renderer to do this for instance if Fizz handles a float method we don't
want Fiber to as well so each dispatcher implementation can decide if it
makes sense to forward the call or not. For float methods server
disaptchers will handle the call if they can resolve a Request otherwise
they will forward. For client dispatchers they will handle the call and
always forward. The choice needs to be made for each dispatcher method
and may have implications on correct renderer import order. For now we
just live with the restriction that if you want to use server and client
together (such as renderToString in the browser) you need to import the
server renderer after the client renderer.
Josh Story
Sebastian Markbåge
Sebastian Silbermann
Andrew Clark
Juan Pinilla
Hugo Sales
Jan Kassens
Jack Pope
Sophie Alpert
Andrey Lunyov
Ricky
Justin Dhillon
