breaking: don't abort navigation when calling invalidate(All) during navigation - #16188
Merged
Conversation
…g navigation When a navigation starts, `invalidate(All)` will no longer abort that navigation - previously it indirectly did that because both invalidation and navigation shared the same navigation token; then navigation thought "oh I'm outdated I abort". Now there's a separate invalidation token so we can handle each case sensibly: - when invalidation finishes after goto finishes, it will not apply its stale data. It _will_ apply redirects though since technically it's the last navigation that happened and so it should take precedence - when invalidation finishes before goto finishes, it will apply, but it will not abort the navigation Closes #9354
🦋 Changeset detectedLatest commit: 42ae46b The changes in this PR will be included in the next version bump. This PR includes changesets to release 1 package
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
…ing a concurrent in-flight `invalidate(All)` to be silently aborted even though the navigation never happened.
This commit fixes the issue reported at packages/kit/src/runtime/client/client.js:1778
## Bug
At the top of `navigate()` in `packages/kit/src/runtime/client/client.js`:
```js
const prev_token = navigation_token;
navigation_token = invalidation_token = nav_token;
```
Both `navigation_token` and `invalidation_token` are overwritten with the new navigation's `nav_token`. However, on the block path (when `beforeNavigate` cancels the navigation and `nav` is `null`), only `navigation_token` is rolled back:
```js
if (!nav) {
block();
if (navigation_token === nav_token) navigation_token = prev_token;
return;
}
```
`invalidation_token` is left stuck at the aborted navigation's `nav_token`, and the previous value was never captured so it couldn't be restored.
### Concrete failure trigger
1. `invalidate(All)` runs `_invalidate()`: sets `token = invalidation_token = A`, captures `nav_token = navigation_token (= N0)`, then `await load_route(intent)`.
2. A navigation is triggered. `navigate()` synchronously runs `navigation_token = invalidation_token = B` (the new `nav_token`).
3. A `beforeNavigate` callback cancels it, so `nav` is `null`. `navigation_token` is restored to `N0`, but `invalidation_token` stays `B`.
4. The invalidation resumes and checks `token !== invalidation_token` → `A !== B` is true → it returns early, silently dropping the invalidation result, even though the navigation never actually happened.
This is a regression from the previous single-shared-`token` design, where the block branch restored the token so an in-flight invalidation was not aborted by a navigation that ended up blocked. The doc comment on `invalidation_token` says it is "Superseeded by both later invalidate(All)s and navigations" — but a *blocked* navigation should not supersede it.
## Fix
Capture the previous `invalidation_token` alongside `prev_token`, and restore it on the block path mirroring the `navigation_token` rollback (guarded by `=== nav_token` so a genuinely later `invalidate(All)` that superseded it is not clobbered):
```js
const prev_token = navigation_token;
const prev_invalidation_token = invalidation_token;
navigation_token = invalidation_token = nav_token;
...
if (!nav) {
block();
if (navigation_token === nav_token) navigation_token = prev_token;
if (invalidation_token === nav_token) invalidation_token = prev_invalidation_token;
return;
}
```
The `=== nav_token` guard ensures that if a newer invalidation or navigation has already taken over `invalidation_token` in the meantime, we don't roll it back.
Co-authored-by: Vercel <vercel[bot]@users.noreply.github.com>
Co-authored-by: dummdidumm <sholthausen@web.de>
Rich-Harris
approved these changes
Jul 1, 2026
Rich-Harris
pushed a commit
that referenced
this pull request
Jul 2, 2026
This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to version-3, this PR will be updated.⚠️ ⚠️ ⚠️ ⚠️ ⚠️ ⚠️ `version-3` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `version-3`.⚠️ ⚠️ ⚠️ ⚠️ ⚠️ ⚠️ # Releases ## @sveltejs/adapter-node@6.0.0-next.2 ### Major Changes - breaking: add `kit.paths.origin` config option, remove `kit.prerender.origin` and the `adapter-node` `ORIGIN` environment variable ([#16161](#16161)) ### Patch Changes - Updated dependencies [[`3c434fb`](3c434fb), [`a9284e8`](a9284e8), [`3726a7a`](3726a7a), [`f9d2240`](f9d2240), [`a9284e8`](a9284e8), [`7c040ba`](7c040ba), [`223eaad`](223eaad), [`223eaad`](223eaad), [`3b907d4`](3b907d4), [`fd628a5`](fd628a5), [`223eaad`](223eaad), [`6c1d035`](6c1d035), [`178eac0`](178eac0), [`c6562a9`](c6562a9), [`8eca2ab`](8eca2ab), [`61cf188`](61cf188)]: - @sveltejs/kit@3.0.0-next.6 ## @sveltejs/kit@3.0.0-next.6 ### Major Changes - breaking: return no content for 204 responses ([#16200](#16200)) - breaking: form action responses now use the HTTP status code returned from `fail` ([#16200](#16200)) - breaking: nested server-only directories ([#15685](#15685)) - breaking: add `kit.paths.origin` config option, remove `kit.prerender.origin` and the `adapter-node` `ORIGIN` environment variable ([#16161](#16161)) - breaking: don't abort navigation when calling `invalidate(All)` during navigation ([#16188](#16188)) - breaking: allow `handleError` to influence status code ([#16162](#16162)) - breaking: forbid external redirects by default ([#16198](#16198)) ### Minor Changes - feat: use `type: 'module'` for service worker registrations ([#16169](#16169)) - feat: add `dirty()` property to form fields ([#16208](#16208)) - feat: add `cookies.parse` method ([#16203](#16203)) ### Patch Changes - fix: drain unconsumed request bodies so keep-alive connections don't hang ([#16170](#16170)) - fix: properly handle Date objects in form.fields.set ([#16168](#16168)) - fix: skip clean fields when programmatically validating forms ([#16208](#16208)) - breaking: experimental remote form `validate({ includeUntouched })` option is now `all` ([#16208](#16208)) - fix: return `undefined` from `fields.branch.issues()` when only `fields.branch.leaf` has issues ([#16187](#16187)) - feat: add field.touched() helper to remote form fields ([#14692](#14692)) ## @sveltejs/adapter-cloudflare@8.0.0-next.1 ### Patch Changes - fix: avoid overriding user's existing `_headers` rules ([#16183](#16183)) - Updated dependencies [[`3c434fb`](3c434fb), [`a9284e8`](a9284e8), [`3726a7a`](3726a7a), [`f9d2240`](f9d2240), [`a9284e8`](a9284e8), [`7c040ba`](7c040ba), [`223eaad`](223eaad), [`223eaad`](223eaad), [`3b907d4`](3b907d4), [`fd628a5`](fd628a5), [`223eaad`](223eaad), [`6c1d035`](6c1d035), [`178eac0`](178eac0), [`c6562a9`](c6562a9), [`8eca2ab`](8eca2ab), [`61cf188`](61cf188)]: - @sveltejs/kit@3.0.0-next.6 Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Contributor
|
!!! thanks a lot |
Closed
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 free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
When a navigation starts,
invalidate(All)will no longer abort that navigation - previously it indirectly did that because both invalidation and navigation shared the same navigation token; then navigation thought "oh I'm outdated I abort".Now there's a separate invalidation token so we can handle each case sensibly:
Closes #9354