- 30 Oct, 2025 21 commits
-
-
nextjs-bot authored
-
JJ Kasper authored
This fixes parallel routes being considered separate outputs with the same pathname in the adapters build-complete hook and instead merges them into the same output combining the necessary outputs.
-
Benjamin Woodruff authored
Noticed this was dead while I was looking through how we might integrate bincode. We only ever use `any_serialization` and not `magic_serialization`. In theory, `magic_serialization` is slightly better because it can include the type name in the error message. If we go through with the bincode stuff, I'll replace `any_serialization` and the `MagicAny` serialization with something that doesn't depend on `erased-serde`.
-
Sebastian "Sebbie" Silbermann authored
Co-authored-by:Zack Tanner <1939140+ztanner@users.noreply.github.com>
-
Sebastian "Sebbie" Silbermann authored
-
Sebastian "Sebbie" Silbermann authored
-
Sebastian "Sebbie" Silbermann authored
-
Hendrik Liebau authored
-
Tobias Koppers authored
### What? That makes a tracing look a bit better
-
Sebastian "Sebbie" Silbermann authored
-
Sebastian "Sebbie" Silbermann authored
Moves the assertions out of `act` and removes redundant waiting for elements. Similar to https://github.com/vercel/next.js/pull/85417
-
Tobias Koppers authored
### What? Fix Equal implementation on a bunch of structs. Make tasks deterministic
-
Luke Sandberg authored
[turbopack] Change where `cell`s are created in `resolve_raw` to make cell allocation order deterministic. (#85525) Constructing cells within a `try_flat_join` will lead to racy creation. Instead we just need to do a second pass after the fact Also eliminate a clone while we are there.
-
Hendrik Liebau authored
I noticed that we fixed some of the differences in error stacks and source frames between bundlers, which allows us to consolidate duplicate snapshots. To accomplish this, I asked Claude to write an AST-based script to find and deduplicate the affected snapshots.
-
Tobias Koppers authored
### What? Get rid of `impl Into` for value types. It's kind of legacy and we want cell creation to be kind of explicit instead of hidden in `into()` calls. Also reduces the generated code size of the macro, potentially reducing rust build time and binary size.
-
Luke Sandberg authored
[turbopack] Move generation of `cacheLife` types out of the webpack plugin and into the dev bundler directly (#85539) Ensure that users configuring custom `cacheLife` configs with Typescript get support with turbopack This follows the same pattern as typed-routes, so generation can happen during `next type-gen` and also via the dev bundler. Also this fixes a subtle bug in the timestamp formatting where we believed months had 30 or 31 _weeks_ instead of 30 or 31 _days_s. Beyond the unit tests, i also tried out adding a `cacheLife` config to the `with-turbopack` example and observed that the types were generated and vscode was able to autocomplete my custom config type. Fixes NAR-458
-
Sebastian "Sebbie" Silbermann authored
-
Luke Sandberg authored
* `all` is dead * `unchanged` is dead * `wrap` should be private
-
Sebastian "Sebbie" Silbermann authored
-
Swarnava Sengupta authored
Fix a small typo
-
nextjs-bot authored
-
- 29 Oct, 2025 19 commits
-
-
nextjs-bot authored
-
Zack Tanner authored
When a server action triggers a redirect, it will return the RSC payload for the destination state in a single roundtrip. This lets us update the UI without needing to waterfall when getting the target page's data. We reject the server action with a special error signaling that the form state should be reset, but otherwise the redirect itself is unnecessary, since we already are handling it in the server action reducer. However, when we introduced Activity for bfCache, this meant navigating back to the previous page would replay the redirect error that was stashed in the `RedirectBoundary`. I incorrectly patched this by just returning the current router state, rather than the redirect destination's state, but this lost the ability to return the updated state in the same roundtrip as the redirect. Instead, this marks the error that gets stashed in the `RedirectBoundary` as being "handled". Re-activating the hidden subtree will now still correctly reset the state (because the boundary handles it), but won't attempt to perform the redirect again. Closes NAR-470
-
Vercel Release Bot authored
This auto-generated PR updates the development integration test manifest used when testing Rspack.
-
Vercel Release Bot authored
This auto-generated PR updates the production integration test manifest used when testing Rspack.
-
Gus Fune authored
### What? This PR refactors the Google Tag Manager implementation on `@next/third-parties`, it rewrites the implementation of the script url, params and types to allow `gtmId` to be optional when a script URL is provided. ### Why? Google released [Google Tag Gateway for Advertisers](https://developers.google.com/tag-platform/tag-manager/gateway/setup-guide?setup=manual) which allows first-party tracking on GTM. When running GTM using the current implementation, it doesn't work because the `?id=` is hardcoded into the implementation as it doesn't consider the possibility of tag gateway or any other setup where the ID is already part of the url. ### How? - Expanded the types - Refactored the initialisation of the script URL - Updated documentation relevant to GTM Closes #81009 Co-authored-by:
JJ Kasper <jj@jjsweb.site>
-
Josh Story authored
Today the `captureOwnerStack()` function is provided to shared utilities through an AsyncLocalStorage that scopes the method from the appropriate React instance. This is so that external code like patches to sync IO methods can still generate errors with the appropriate React owner information even when the patched code itself is not bundled and can be called from etiher SSR or RSC contexts. This works but it makes plumbing the React instances around tricky. There is a simpler way. Most of the time you can just try both React's. If one gives you a non-null/undefined result then you know you are in that scope. If neither do then you're outside a React scope altogether. In this change I remove `captureOwnerStack()` from the workUnitStore types and just call it from the shared server runtime which gives even external code access to the appropriate React instances for bundled code
-
Sebastian "Sebbie" Silbermann authored
Co-authored-by:graphite-app[bot] <96075541+graphite-app[bot]@users.noreply.github.com>
-
Joseph authored
Closes: https://github.com/vercel/next.js/issues/85281
-
Hendrik Liebau authored
With `--debug-build-paths` being added in #85052, we can now use this CLI flag in tests, instead of setting the private environment variable `NEXT_PRIVATE_APP_PATHS`. Related to this change, this PR fixes how we print the dev, start and build commands during tests, by properly escaping the CLI arguments. This ensures that the command is copy-pasteable even when the args contain escaped characters.
-
Sebastian "Sebbie" Silbermann authored
`createSandbox` fully isolates the returned session from the existing Next instance by restarting the server. This is wasteful if the existing instance is already started. Some tests called `nextTestSetup()` without `skipStart` which got fixed in this PR. Some tests can just use `patchFile` nowadays instead of `createSandbox`.
-
Joseph authored
-
Joseph authored
-
Hendrik Liebau authored
When defining a `cacheLife` with an expire time below 5 minutes or a revalidate time of 0 we consider the `'use cache'` function dynamic during prerendering, i.e. it will be excluded from the static shell. Similarly, accessing runtime `params` (not defined in `generateStaticParams`) also makes the function dynamic. Those cases require adding a parent suspense boundary. If there's no suspense boundary, we should show an error during prerendering at build time, as well as in development as part of the prerender validation. The latter is currently not working as expected when the `'use cache'` directive is used in a page (or layout) directly. As the added snapshot tests show, no redbox is shown in those cases. This is a known issue that we plan to address separately. We will also follow up with improving the error messages and docs for those dynamic cache cases. This detail got lost in #85087.
-
Andrew Clark authored
Currently the head (i.e. the viewport and the metadata) is stored in the same cache entry as the route tree. This is not ideal because the head may or may not contain application data, and the response can differ depending on the prefetch strategy. We need to be able to re-fetch the head without also invalidating the route tree. This refactors the Segment Cache to treat the head data as a special kind of "segment". It doesn't appear in the route tree, so there's some logic to account for that, but for the most part it reuses all the existing implementation for fetching and caching normal segment data.
-
Sebastian "Sebbie" Silbermann authored
-
Sebastian "Sebbie" Silbermann authored
-
Joseph authored
-
Jiachi Liu authored
-
JJ Kasper authored
This updates manifest with known flakey tests x-ref: https://github.com/vercel/next.js/actions/runs/18886182152/job/53902013243 x-ref: https://github.com/vercel/next.js/actions/runs/18886182152/job/53902013283 x-ref: https://github.com/vercel/next.js/actions/runs/18886182152/job/53902013307
-