### Why?
The current wording leaves room for questions/misinterpretations like:
- What does "use `src/`" mean? Does this affect my build, like compile
`src/*.ts` to `lib/*.js`?
- Oh, I can customize import aliases to use @ for project root instead
of doing relative everywhere? I'll choose Yes!
I think the proposed wording clarifies ^
---------
Co-authored-by: Sam Ko <sam@vercel.com>
This adds a new target in the next-swc package designed as an entrypoint
for collecting memory benches. Later PRs in the stack add a script to
run the benches and finally a manual-triggered CI step
We'd like to warn users if they have particularly slow file IO, so that they can correct the problem themselves, and don't send us reports of poor performance.
- Feature request: https://vercel.slack.com/archives/C03KAR5DCKC/p1716051650641529
- Tweet about how Bun does this: https://x.com/jarredsumner/status/1637549427677364224
- Bun implementation: 06a9aa80c3/src/install/install.zig (L3038)
**Why 100ms?** Bun used to use 10ms, found it too noisy, and switched to 100ms.
This benchmark should run non-blocking in the background and should not meaningfully slow down server startup (even on slow disks).
## Simulated Testing
I looked around and found https://github.com/schoentoon/slowpokefs/. It hasn't been updated in 10 years, but still seems to build fine.
In a nextjs project directory, turn `.next` into an artifically slow mount point:
```
fusermount -uz .next; rm -rf .next .next.real && mkdir .next .next.real && ~/slowpokefs/slowpokefs -m 50 -M 50 --no-slow-read -F .next.real .next
```
<img width="695" alt="Screenshot 2024-05-21 at 4 14 58 PM" src="https://github.com/vercel/next.js/assets/180404/217d7692-33cf-42b7-bbf7-5a530b9e0df1">
Run `pnpm dev --turbo` and see that the warning is generated.
## "Real World" Testing
The following tests are from a linux VM using virtiofs to connect to a 5400 RPM USB 2.0 HDD with APFS:
Only the .next is on an HDD (bind mount):
Otherwise idle disk: `4.86ms 4.44ms 5.31ms`
Otherwise busy disk (copying files): `69.34ms 53.75ms 25.12ms`
The whole project directory is on the HDD (more realistic):
Otherwise idle disk: `20.29ms 35.61ms 48.12ms`
Otherwise busy disk (copying files): `131.40ms 21.71ms 87.33ms`
Most of the time the threshold was not reached, but the 131.40ms test did trigger the warning!
Fixes PACK-3087
This reverts commit
637033c410.
What wasn't clear originally is that we require that required packages
have their `package.json` resolvable. So either they don't have an
`exports` field, or their `package.json` is listed in the `exports`
field. The React beta types didn't have `package.json` entry in
`exports` so we accidentally thought they weren't installed.
I published [`19.0.0-beta.2` that has `package.json` in
`exports`](13879ee571 (diff-81cc573aa0c2bd0e13f9463499747741704aabccd7474f544db710befd7bcfc4R44))
so we can restore the old behavior. It's still questionable IMO to do
all that I/O work just for a nicer error message that may contain
false-positive but that's for another time.
Alternative to https://github.com/vercel/next.js/pull/65867 this
stabilizes our `swrDelta` config that allows customizing the
`stale-while-revalidate` period that is included in the `Cache-Control`
header for ISR enabled routes.
This is not a breaking change itself as no default value is provided as
is the case already.
This PR changes the `getWaitUntil` logic to always check if the platform
provided a request context with a `waitUntil` implementation (via [the
`@next/request-context`
symbol](05e6b82576/packages/next/src/server/after/wait-until-builtin.ts (L12)))
, regardless of minimalMode or runtime.
Previously, we wouldn't check the context at all unless `minimalMode`
was set, and defaulted `waitUntil` to a noop otherwise. This would make
integration more difficult for providers who don't use `minimalMode`,
because they'd have no way to inject their `waitUntil` implementation.
Ensures that app routes (like the `favicon.ico` added in the test case
in this PR which is represented as an app route) do not have a prefetch
RSC also attached.
## Background
Previously we introduced automatic caching for `fetch` based on certain
heuristics that were a bit tricky to grasp all scenarios. The scenarios
we would automatically cache were no dynamic data access before the
fetch call e.g. `headers()` or `cookies()`, the fetch call is inside of
a dynamic page e.g. `POST` method or `export const revalidate = 0` page
and the fetch is a non-`GET` request or has `Authorization` or `Cookie`
headers, or the fetch had `cache: 'no-store' | 'no-cache'` or
`revalidate: 0`.
## New Behavior
By default fetches will no longer automatically be cached. Instead they
need to be opted-in to caching via `export const fetchCache =
'default-cache' | 'force-cache',` `next: { revalidate: false or value >
0 }` or `cache: 'force-cache' | 'default-cache'`.
When the fetch call is automatically skipping the cache it won't impact
the page level ISR cacheability although if a fetch call manually
specifies `cache: 'no-store'` or `revalidate: 0` it will still bail from
the page being statically generated as it was before.
To achieve the previous behavior of automatic fetch caching all that
needs to be added is `export const fetchCache = 'default-cache'` in the
root layout(s) of your project.
When hot-reloading an edge runtime page, sometimes we randomly get this:
```
[InvariantError: Invariant: Cannot call onClose on a response that is already sent. This is a bug in Next.js.] {
name: 'InvariantError'
}
```
and since `unstable_after` relies on `onClose`, no callbacks are called.
I believe this is because `web-server` [calls `res.send()` pretty early
for streaming
responses](b9817f8351/packages/next/src/server/web-server.ts (L311)),
so `res.sent` becomes `true` even though we're still streaming, and
prevents us from calling `onClose` in the streaming render. It makes
more sense to check the CloseController's `isClosed` instead -- when
that becomes `true`, we _really_ shouldn't call onClose anymore (points
to a bug), but before that it's fine.
Follow-up to https://github.com/vercel/next.js/pull/65825 this ensures
we don't skip the static optimization specifically for metadata routes
as this most often should be static as they aren't dynamic content and
are requested very frequently.
This PR adds a flag to Next.js to enable Webpack options to improve
memory usage. See https://github.com/webpack/webpack-sources/pull/155
for a full description of the changes and impact on memory.
This PR adds a patch to `webpack-sources` temporarily that contains the
fixes as the real changes are iterated on to merge upstream in the
`webpack/webpack-sources` repository. After that is done, the patch will
be reverted and the latest `webpack-sources` version will be updated in
Next.js.
To avoid extra network hops we can compare existing cache entries we've
already fetched and see if the revalidated value matches and if it does
we can avoid sending the set request with the identical data.
As noted in (https://github.com/vercel/nextpack/pull/103), by not having
this as a default feature, when running the `build-native` script it's
not able to re-use any cargo caches that might already exist from a
default build.
Because this feature impacts the swc crate, everything dependent on it
can be forced to rebuild.
By making this a default feature, the hope/theory is that the number of
different ways we potentially build our dependency graph reduces, and
cargo cache hits will improve.
## Changes
- Add `plugin` to the default set of features for `next-swc-napi`.
- Consolidate `build-native-no-plugin-woa` with
`build-native-no-plugin`. Given the lack of any other default features,
these are equivalent.
- `i686-pc-windows-msvc`: Switch to `build-native-no-plugin-release` in
CI instead of just supplying the `--release` flag. This has the benefit
of also enabling `tracing/release_max_level_info`, which might reduce
the size of or speed up these builds.
- `x86_64-apple-darwin`: Remove a redundant `--release` flag in the CI
config (no expected changes).
## Testing
I've tested this locally on Linux (`pnpm run build-native`) and
Windows-on-ARM (`pnpm run build-native-no-plugin`) by making sure things
still build. I'll also make sure CI passes.
This parameter was previously used for a warning message, but we no
longer print that warning since
https://github.com/vercel/next.js/pull/63321 so now we can remove the
unused parameter.