Uh oh!
There was an error while loading. Please reload this page.
[Server Components] Walk parsed JSON instead of using reviver for parsing RSC payload - #35776
Merged
Merged
Conversation
timneutkens
commented
Feb 13, 2026
Uh oh!
There was an error while loading. Please reload this page.
timneutkens
commented
Feb 13, 2026
Uh oh!
There was an error while loading. Please reload this page.
timneutkens
commented
Feb 13, 2026
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
…/react into add/rsc-parsing-perf
unstubbable
marked this pull request as ready for review
February 16, 2026 20:46
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Co-authored-by: Hendrik Liebau <mail@hendrik-liebau.de>
Co-authored-by: Hendrik Liebau <mail@hendrik-liebau.de>
unstubbable
approved these changes
Feb 19, 2026
Uh oh!
There was an error while loading. Please reload this page.
AbanoubGhadban added a commit
to AbanoubGhadban/react
that referenced
this pull request
Apr 18, 2026
…yload Backport of react#35776 to v19.0.3. Replace the JSON.parse reviver callback in parseModel with a two-step approach: plain JSON.parse() followed by a recursive reviveModel() post-process. This yields a ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings for every key-value pair. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Merged
3 tasks
AbanoubGhadban added a commit
to AbanoubGhadban/react
that referenced
this pull request
Apr 18, 2026
… payload Backport of react#35776 to v19.0.3. Replace the JSON.parse reviver callback in parseModel with a two-step approach: plain JSON.parse() followed by a recursive reviveModel() post-process. This yields a ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings for every key-value pair. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AbanoubGhadban added a commit
to AbanoubGhadban/react
that referenced
this pull request
Apr 18, 2026
… payload Backport of react#35776 to v19.0.3. Replace the JSON.parse reviver callback in parseModel with a two-step approach: plain JSON.parse() followed by a recursive reviveModel() post-process. This yields a ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings for every key-value pair. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AbanoubGhadban added a commit
to shakacode/react_on_rails_rsc
that referenced
this pull request
Apr 18, 2026
… payload Rebuild react-server-dom-webpack from React fork rsc-patches/v19.0.3 with the reviveModel optimization (backport of react/react#35776). Replaces JSON.parse reviver callback with plain JSON.parse() followed by a recursive reviveModel() walk in pure JS, yielding ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
4 tasks
AbanoubGhadban added a commit
to shakacode/react_on_rails_rsc
that referenced
this pull request
Apr 18, 2026
… payload Rebuild react-server-dom-webpack from React fork rsc-patches/v19.0.3 with the reviveModel optimization (backport of react/react#35776). Replaces JSON.parse reviver callback with plain JSON.parse() followed by a recursive reviveModel() walk in pure JS, yielding ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AbanoubGhadban added a commit
to AbanoubGhadban/react
that referenced
this pull request
Apr 18, 2026
… payload Backport of react#35776 to v19.0.3. Replace the JSON.parse reviver callback in parseModel with a two-step approach: plain JSON.parse() followed by a recursive reviveModel() post-process. This yields a ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings for every key-value pair. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
4 tasks
AbanoubGhadban added a commit
to AbanoubGhadban/react
that referenced
this pull request
May 4, 2026
… payload (#4) Backport of react#35776 to v19.0.3. Replaces the JSON.parse reviver callback in parseModel with a two-step approach: plain JSON.parse() followed by a recursive reviveModel() walk in pure JavaScript. Yields ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings for every key-value pair. Also adds a no-op setter on the $Y (omitted object) getter-only property to prevent TypeError under strict mode when reviveModel reassigns the property after parseModelString returns. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AbanoubGhadban added a commit
to shakacode/react_on_rails_rsc
that referenced
this pull request
May 4, 2026
…yload (#33) Rebuild react-server-dom-webpack from React fork rsc-patches/v19.0.3 with the reviveModel optimization (backport of react/react#35776). Replaces JSON.parse reviver callback with plain JSON.parse() followed by a recursive reviveModel() walk in pure JS, yielding ~75% speedup in RSC chunk deserialization by eliminating C++→JS boundary crossings. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
justin808 added a commit
to shakacode/react_on_rails_rsc
that referenced
this pull request
Jun 12, 2026
* Document Option 5 stock npm runtime go decision in eliminate-react-fork doc Records the #55 spike outcome: GO on depending on stock npm react-server-dom-webpack instead of the vendored fork build, gated on the #60 migration checklist, with Option 4 (patch files, 2-patch corpus) as the explicit fallback. Includes the complete vendored-vs-stock delta inventory, upstream-status evidence for both true runtime patches (JSON-walk upstreamed in react/react#35776, FOUC fix not upstream), a wrapper-layer FOUC design that needs no React patch, parse benchmark numbers, exports/conditions parity mapping, webpack-globals/manifest contract verification, and the security finding that the vendored 19.0.4 runtime predates the CVE-2026-23869/23870 fixes. Closes#55 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Address PR review feedback on Option 5 doc - Correct peer-dependency wording to React >=19.2.7 (greptile) - Link the CVE advisories (claude review) - Note linear scaling of the parse-cost regression (claude review) - Promote the unbundled ./server verification to an explicit fallback gate (claude review) - Document preinit dedupe idempotency in the Proxy design (claude review) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Address second review pass on Option 5 doc - Retitle Option 4 heading to Superseded/Fallback Only (heading inconsistency) - Reword CVE remediation tracking to reflect actual state (spike report flag) - Document benchmark methodology inline so it is reproducible after a bump - State the semver impact of the React >=19.2.7 peer floor explicitly - Add explicit prerequisites before archiving the fork (point of no return) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
5 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Follow-up to vercel/next.js#89823 with the actual changes to React.
Replaces the
JSON.parsereviver callback ininitializeModelChunkwith a two-step approach: plainJSON.parse()followed by a recursivereviveModel()post-process (same as in Flight Reply Server). This yields a ~75% speedup in RSC chunk deserialization.Problem
createFromJSONCallbackreturns a reviver function passed as the second argument toJSON.parse(). This reviver is called for every key-value pair in the parsed JSON. While the logic inside the reviver is lightweight, the dominant cost is the C++ → JavaScript boundary crossing — V8'sJSON.parseis implemented in C++, and calling back into JavaScript for every node incurs significant overhead.Even a trivial no-op reviver
(k, v) => vmakesJSON.parse~4x slower than bareJSON.parsewithout a reviver:Change
Replace the reviver with a two-step process:
JSON.parse(resolvedModel)— parse the entire payload in C++ with no callbacksreviveModel— recursively walk the resulting object in pure JavaScript to apply RSC transformationsThe
reviveModelfunction includes additional optimizations over the original reviver:parseModelStringwhen the string starts with$, skipping the vast majority of strings (class names, text content, etc.)Results
You can find the related applications in the Next.js PR as I've been testing this on Next.js applications.
Table as Server Component with 1000 items
Before:
After:
Table as Client Component with 1000 items
Before:
After:
Nested Suspense
Before:
After:
Even more nested Suspense (double-level Suspense)
Before:
After:
How did you test this change?
Ran it across many Next.js benchmark applications.
The entire Next.js test suite passes with this change.