This masks flight parameters from middleware so it doesn't interfere with RSC or routing.
## Bug
- [ ] Related issues linked using `fixes #number`
- [ ] Integration tests added
- [ ] Errors have helpful link attached, see `contributing.md`
## Feature
- [ ] Implements an existing feature request or RFC. Make sure the feature request has been accepted for implementation before opening a PR.
- [ ] Related issues linked using `fixes #number`
- [ ] Integration tests added
- [ ] Documentation added
- [ ] Telemetry added. In case of a feature if it's used or not.
- [ ] Errors have helpful link attached, see `contributing.md`
## Documentation / Examples
- [ ] Make sure the linting passes by running `pnpm lint`
- [ ] The examples guidelines are followed from [our contributing doc](https://github.com/vercel/next.js/blob/canary/contributing.md#adding-examples)
Co-authored-by: Tim Neutkens <6324199+timneutkens@users.noreply.github.com>
Such bots typically navigate websites using hard navigations (as they
crawl one URL at a time). Respectively, they do not benefit from
prefetches at all, while increasing the cost of both the crawl and
operating the site.
<!--
Thanks for opening a PR! Your contribution is much appreciated.
In order to make sure your PR is handled as smoothly as possible we
request that you follow the checklist sections below.
Choose the right checklist for the change that you're making:
-->
## Bug
- [ ] Related issues linked using `fixes #number`
- [x] Integration tests added
- [ ] Errors have helpful link attached, see `contributing.md`
A small refactor PR to convert some imports to type imports, as well as removing a couple of unused exports.
The Edge SSR loader is also missing the global process injection (`enhanceGlobals`).
## Bug
- [ ] Related issues linked using `fixes #number`
- [ ] Integration tests added
- [ ] Errors have helpful link attached, see `contributing.md`
## Feature
- [ ] Implements an existing feature request or RFC. Make sure the feature request has been accepted for implementation before opening a PR.
- [ ] Related issues linked using `fixes #number`
- [ ] Integration tests added
- [ ] Documentation added
- [ ] Telemetry added. In case of a feature if it's used or not.
- [ ] Errors have helpful link attached, see `contributing.md`
## Documentation / Examples
- [ ] Make sure the linting passes by running `pnpm lint`
- [ ] The examples guidelines are followed from [our contributing doc](https://github.com/vercel/next.js/blob/canary/contributing.md#adding-examples)
## Client-side router for `app` directory
This PR implements the new router that leverages React 18 concurrent features like Suspense and startTransition.
It also integrates with React Server Components and builds on top of it to allow server-centric routing that only renders the part of the page that has to change.
It's one of the pieces of the implementation of https://nextjs.org/blog/layouts-rfc.
## Details
I'm going to document the differences with the current router here (will be reworked for the upgrade guide)
### Client-side cache
In the current router we have an in-memory cache for getStaticProps data so that if you prefetch and then navigate to a route that has been prefetched it'll be near-instant. For getServerSideProps the behavior is different, any navigation to a page with getServerSideProps fetches the data again.
In the new model the cache is a fundamental piece, it's more granular than at the page level and is set up to ensure consistency across concurrent renders. It can also be invalidated at any level.
#### Push/Replace (also applies to next/link)
The new router still has a `router.push` / `router.replace` method.
There are a few differences in how it works though:
- It only takes `href` as an argument, historically you had to provide `href` (the page path) and `as` (the actual url path) to do dynamic routing. In later versions of Next.js this is no longer required and in the majority of cases `as` was no longer needed. In the new router there's no way to reason about `href` vs `as` because there is no notion of "pages" in the browser.
- Both methods now use `startTransition`, you can wrap these in your own `startTransition` to get `isPending`
- The push/replace support concurrent rendering. When a render is bailed by clicking a different link to navigate to a completely different page that still works and doesn't cause race conditions.
- Support for optimistic loading states when navigating
##### Hard/Soft push/replace
Because of the client-side cache being reworked this now allows us to cover two cases: hard push and soft push.
The main difference between the two is if the cache is reused while navigating. The default for `next/link` is a `hard` push which means that the part of the cache affected by the navigation will be invalidated, e.g. if you already navigated to `/dashboard` and you `router.push('/dashboard')` again it'll get the latest version. This is similar to the existing `getServerSideProps` handling.
In case of a soft push (API to be defined but for testing added `router.softPush('/')`) it'll reuse the existing cache and not invalidate parts that are already filled in. In practice this means it's more like the `getStaticProps` client-side navigation because it does not fetch on navigation except if a part of the page is missing.
#### Back/Forward navigation
Back and Forward navigation ([popstate](https://developer.mozilla.org/en-US/docs/Web/API/Window/popstate_event)) are always handled as a soft navigation, meaning that the cache is reused, this ensures back/forward navigation is near-instant when it's in the client-side cache. This will also allow back/forward navigation to be a high priority update instead of a transition as it is based on user interaction. Note: in this PR it still uses `startTransition` as there's no way to handle the high priority update suspending which happens in case of missing data in the cache. We're working with the React team on a solution for this particular case.
### Layouts
Note: this section assumes you've read [The layouts RFC](https://nextjs.org/blog/layouts-rfc) and [React Server Components RFC](https://reactjs.org/blog/2020/12/21/data-fetching-with-react-server-components.html)
React Server Components rendering leverages the Flight streaming mechanism in React 18, this allows sending a serializable representation of the rendered React tree on the server to the browser, the client-side React can use this serialized representation to render components client-side without the JavaScript being sent to the browser. This is one of the building blocks of Server Components. This allows a bunch of interesting features but for now I'll keep it to how it affects layouts.
When you have a `app/dashboard/layout.js` and `app/dashboard/page.js` the page will render as children of the layout, when you add another page like `app/dashboard/integrations/page.js` that page falls under the dashboard layout as well. When client-side navigating the new router automatically figures out if the page you're navigating to can be a smaller render than the whole page, in this case `app/dashboard/page.js` and `app/dashboard/integrations/page.js` share the `app/dashboard/layout.js` so instead of rendering the whole page we render below the layout component, this means the layout itself does not get re-rendered, the layout's `getServerSideProps` would not be called, and the Flight response would only hold the result of `app/dashboard/integrations/page.js`, effectively giving you the smallest patch for the UI.
---
Note: the commits in this PR were mostly work in progress to ensure it wasn't lost along the way. The implementation was reworked a bunch of times to where it is now.
Co-authored-by: Jiachi Liu <4800338+huozhi@users.noreply.github.com>
Co-authored-by: JJ Kasper <22380829+ijjk@users.noreply.github.com>
Strip next internal queries in inline flight response
## Bug
- [ ] Related issues linked using `fixes #number`
- [x] Integration tests added
- [ ] Errors have helpful link attached, see `contributing.md`
Part of #31506, this PR removes `loadEnvConfig` and `chalk` from the base server while keeping the same behavior for the node server.
## Bug
- [ ] Related issues linked using `fixes #number`
- [ ] Integration tests added
- [ ] Errors have helpful link attached, see `contributing.md`
## Feature
- [ ] Implements an existing feature request or RFC. Make sure the feature request has been accepted for implementation before opening a PR.
- [ ] Related issues linked using `fixes #number`
- [ ] Integration tests added
- [ ] Documentation added
- [ ] Telemetry added. In case of a feature if it's used or not.
- [ ] Errors have helpful link attached, see `contributing.md`
## Documentation / Examples
- [ ] Make sure the linting passes by running `yarn lint`
Updates the React 18 concurrent tests to support streaming, by using webdriver (instead of cheerio) to support dynamic updates, and being more explicit about SSR vs. hydrated content.
Also enables streaming support by setting `supportsDynamicHTML: true` in `pipe(...)` for non-bot user agents.
Our `Observable` use has gotten sufficiently complex that it makes sense to just use a 3rd party implementation and not worry about maintaining it ourselves. As a bonus, it doesn't rely on Node APIs.
Implements `renderToString` in terms of a new `renderToStream`. The former is used for legacy documents that generate the body HTML as part of `getInitialProps`. The latter will be used directly in #27794 when streaming dynamic HTML.
Since we're exposing an actual streaming response for dynamic HTML (instead of buffering with `resultFromChunks`), we use `multiplexResult` to buffer and multiplex the underlying result to multiple subscribers.
Adds `RenderResult`, replacing the `string` that `renderToHTML` used to return, with an `Observable`-like API that callers can use to subscribe and get a callback when chunks are available to flush, etc.
This is the last architectural change needed for streaming. There are, however, other things currently standing in the way of streaming. For example, it is common to mutate `res` in `getServerSideProps` to do routing work, or write headers, before fetching page data. This pattern effectively nullifies any advantages of streaming. I may do a follow-up PR that adds an experimental alternative for applications not using React 18, but the main purpose for this support is for Suspense and Server Components.
For that reason, there's no actual streaming here yet: instead we just flush a single chunk. A follow-up PR will add support for streaming suspense boundaries in React 18.