This bugfix release fixes how sorted entity adapters handle duplicate IDs, tweaks the TS types for RTKQ query state cache entries to improve how the data field is handled, and adds better cleanup for long-running listener middleware effects.

What's Changed

• fix(entityAdapter): ensure sorted addMany keeps first occurrence of duplicate ids by @ demyanm in #5097
• fix(entityAdapter): ensure sorted setMany keeps just unique IDs in state.ids by @ demyanm in #5107
• fix(types): ensure non-undefined data on isSuccess with exactOptionalPropertyTypes by @ CO0Ki3 in #5088
• Allow executing effects that have become unsubscribed to be canceled by listenerMiddleware.clearListeners by @ chris-chambers in #5102

Full Changelog: v2.9.0...v2.9.1

This feature release rewrites RTK Query's internal subscription and polling systems and the useStableQueryArgs hook for better perf, adds automatic AbortSignal handling to requests still in progress when a cache entry is removed, fixes a bug with the transformResponse option for queries, adds a new builder.addAsyncThunk method, and fixes assorted other issues.

Changelog

RTK Query Performance Improvements

We had reports that RTK Query could get very slow when there were thousands of subscriptions to the same cache entry. After investigation, we found that the internal polling logic was attempting to recalculate the minimum polling time after every new subscription was added. This was highly inefficient, as most subscriptions don't change polling settings, and it required repeated O(n) iteration over the growing list of subscriptions. We've rewritten that logic to debounce the update check and ensure a max of one polling value update per tick for the entire API instance.

Related, while working on the request abort changes, testing showed that use of plain Records to hold subscription data was inefficient because we have to iterate keys to check size. We've rewritten the subscription handling internals to use Maps instead, as well as restructuring some additional checks around in-flight requests.

These two improvements drastically improved runtime perf for the thousands-of-subscriptions-one-cache-entry repro, eliminating RTK methods as visible hotspots in the perf profiles. It likely also improves perf for general usage as well.

We've also changed the implementation of our internal useStableQueryArgs hook to avoid calling serializeQueryArgs on its value, which can avoid potential perf issues when a query takes a very large object as its cache key.

Abort Signal Handling on Cleanup

We've had numerous requests over time for various forms of "abort in-progress requests when the data is no longer needed / params change / component unmounts / some expensive request is taking too long". This is a complex topic with multiple potential use cases, and our standard answer has been that we don't want to abort those requests - after all, cache entries default to staying in memory for 1 minute after the last subscription is removed, so RTKQ's cache can still be updated when the request completes. That also means that it doesn't make sense to abort a request "on unmount".

However, it does then make sense to abort an in-progress request if the cache entry itself is removed. Given that, we've updated our cache handling to automatically call the existing resPromise.abort() method in that case, triggering the AbortSignal attached to the baseQuery. The handling at that point depends on your app - fetchBaseQuery should handle that, a custom baseQuery or queryFn would need to listen to the AbortSignal.

We do have an open issue asking for further discussions of potential abort / cancelation use cases and would appreciate further feedback.

New Options

The builder callback used in createReducer and createSlice.extraReducers now has builder.addAsyncThunk available, which allows handling specific actions from a thunk in the same way that you could define a thunk inside createSlice.reducers:

        const slice = createSlice({
          name: 'counter',
          initialState: {
            loading: false,
            errored: false,
            value: 0,
          },
          reducers: {},
          extraReducers: (builder) =>
            builder.addAsyncThunk(asyncThunk, {
              pending(state) {
                state.loading = true
              },
              fulfilled(state, action) {
                state.value = action.payload
              },
              rejected(state) {
                state.errored = true
              },
              settled(state) {
                state.loading = false
              },
            }),
        })

createApi and individual endpoint definitions now accept a skipSchemaValidation option with an array of schema types to skip, or true to skip validation entirely (in case you want to use a schema for its types, but the actual validation is expensive).

Bug Fixes

The infinite query implementation accidentally changed the query internals to always run transformResponse if provided, including if you were using upsertQueryData(), which then broke. It's been fixed to only run on an actual query request.

The internal changes to the structure of the state.api.provided structure broke our handling of extractRehydrationInfo - we've updated that to handle the changed structure.

The infinite query status fields like hasNextPage are now a looser type of boolean initially, rather than strictly false.

TS Types

We now export Immer's WritableDraft type to fix another non-portable types issue.

We've added an api.endpoints.myEndpoint.types.RawResultType types-only field to match the other available fields.

What's Changed

• Add RawResultType as a type-only property on endpoints by @ EskiMojo14 in #5037
• allow passing an array of specific schemas to skip by @ EskiMojo14 in #5042
• fix(types): re-exporting WritableDraft from immer by @ marinsokol5 in #5015
• Remove Serialisation from useStableQueryArgs by @ riqts in #4996
• add addAsyncThunk method to reducer map builder by @ EskiMojo14 in #5007
• Only run transformResponse when a query is used by @ markerikson in #5049
• Assorted bugfixes for 2.8.3 by @ markerikson in #5060
• Abort pending requests if the cache entry is removed by @ markerikson in #5061
• Update TS CI config by @ markerikson in #5065
• Rewrite subscription handling and polling calculations for better perf by @ markerikson in #5064

Full Changelog: v2.8.2...v2.9.0

This bugfix release fixes a bundle size regression in RTK Query from the build and packaging changes in v2.8.0.

If you're using v2.8.0 or v2.8.1, please upgrade to v2.8.2 right away to resolve that bundle size issue!

Changelog

RTK Query Bundle Size

In v2.8.0, we reworked our packaging setup to better support React Native. While there weren't many meaningful code changes, we did alter our bundling build config file. In the process, we lost the config options to externalize the @ reduxjs/toolkit core when building the RTK Query nested entry points. This resulted in a regression where the RTK core code also got bundled directly into the RTK Query artifacts, resulting in a significant size increase.

This release fixes the build config and restores the previous RTKQ build artifact sizes.

What's Changed

• Restructure build config to fix RTKQ externals by @ markerikson in #4985

Full Changelog: v2.8.1...v2.8.2

This bugfix release makes an additional update to the package config to fix a regression that happened with Jest and jest-environment-jsdom.

Changes

More Package Updates

After releasing v2.8.0, we got reports that Jest tests were breaking. After investigation we concluded that jest-environment-jsdom was looking at the new browser package exports condition we'd added to better support JSPM, finding an ESM file containing the export keyword, and erroring because it doesn't support ES modules correctly.

#4971 (comment) listed several viable workarounds, but this is enough of an issue we wanted to fix it directly. We've tweaked the package exports setup again, and it appears to resolve the issue with Jest.

What's Changed

• fix: browser exports condition by @ aryaemami59 in #4973

Full Changelog: v2.8.0...v2.8.1

This feature release improves React Native compatibility by updating our package exports definitions, and adds queryArg as an additional parameter to infinite query page param functions.

Changelog

Package Exports and React Native Compatibility

Expo and the Metro bundler have been adding improved support for the exports field in package.json files, but those changes started printing warnings due to how some of our package definitions were configured.

We've reworked the package definitions (again!), and this should be resolved now.

Infinite Query Page Params

The signature for the getNext/PreviousPageParam functions has been:

(
    lastPage: DataType,
    allPages: Array<DataType>,
    lastPageParam: PageParam,
    allPageParams: Array<PageParam>,
  ) => PageParam | undefined | null

This came directly from React Query's API and implementation.

We've had some requests to make the endpoint's queryArg available in page param functions. For React Query, that isn't necessary because the callbacks are defined inline when you call the useInfiniteQuery hook, so you've already got the query arg available in scope and can use it. Since RTK Query defines these callbacks as part of the endpoint definition, the query arg isn't in scope.

We've added queryArg as an additional 5th parameter to these functions in case it's needed.

Other Changes

We've made a few assorted docs updates, including replacing the search implementation to now use a local index generated on build (which should be more reliable and also has a nicer results list uI), and fixing some long-standing minor docs issues.

What's Changed

• fix: React-Native package exports by @ aryaemami59 in #4887
• Add queryArg as 5th param to page param functions by @ markerikson in #4967

Full Changelog: v2.7.0...v2.8.0

RTK has hit Stage 2.7! 🤣 This feature release adds support for Standard Schema validation in RTK Query endpoints, fixes several issues with infinite queries, improves perf when infinite queries provide tags, adds a dev-mode check for duplicate middleware, and improves reference stability in slice selectors and infinite query hooks.

Changelog

Standard Schema Validation for RTK Query

Apps often need to validate responses from the server, both to ensure the data is correct, and to help enforce that the data matches the expected TS types. This is typically done with schema libraries such as Zod, Valibot, and Arktype. Because of the similarities in usage APIs, those libraries and others now support a common API definition called Standard Schema, allowing you to plug your chosen validation library in anywhere Standard Schema is supported.

RTK Query now supports using Standard Schema to validate query args, responses, and errors. If schemas are provided, the validations will be run and errors thrown if the data is invalid. Additionally, providing a schema allows TS inference for that type as well, allowing you to omit generic types from the endpoint.

Schema usage is per-endpoint, and can look like this:

If desired, you can also configure schema error handling with the catchSchemaFailure option. You can also disable actual runtime validation with skipSchemaValidation (primarily useful for cases when payloads may be large and expensive to validate, but you still want to benefit from the TS type inference).

See the "Schema Validation" docs section in the createApi reference and the usage guide sections on queries, infinite queries, and mutations, for more details.

Infinite Query Fixes

This release fixes several reported issue with infinite queries:

• The lifecycleApi.updateCachedData method is now correctly available
• The skip option now correctly works for infinite query hooks
• Infinite query fulfilled actions now include the meta field from the base query (such as {request, response}). For cases where multiple pages are being refetched, this will be the meta from the last page fetched.
• useInfiniteQuerySubscription now returns stable references for refetch and the fetchNext/PreviousPage methods

upsertQueryEntries, Tags Performance and API State Structure

We recently published a fix to actually process per-endpoint providedTags when using upsertQueryEntries. However, this exposed a performance issue - the internal tag handling logic was doing repeated O(n) iterations over all endpoint+tag entries in order to clear out existing references to that cache key. In cases where hundreds or thousands of cache entries were being inserted, this became extremely expensive.

We've restructured the state.api.provided data structure to handle reverse-mapping between tags and cache keys, which drastically improves performance in this case. However, it's worth noting that this is a change to that state structure. This shouldn't affect apps, because the RTKQ state is intended to be treated as a black box and not generally directly accessed by user app code. However, it's possible someone may have depended on that specific state structure when writing a custom selector, in which case this would break. An actual example of this is the Redux DevTools RTKQ panel, which iterates the tags data while displaying cache entries. That did break with this change. Prior to releasing RTK 2.7,we released Redux DevTools 3.2.10, which includes support for both the old and new state.api.provided definitions.

TS Support Matrix Updates

Following with the DefinitelyTyped support matrix, we've officially dropped support for TS 5.0, and currently support TS 5.1 - 5.8. (RTK likely still works with 5.0, but we no longer test against that in CI.)

Duplicate Middleware Dev Checks

configureStore now checks the final middleware array for duplicate middleware references. This will catch cases such as accidentally adding the same RTKQ API middleware twice (such as adding baseApi.middleware and injectedApi.middlweware - these are actually the same object and same middleware).

Unlike the other dev-mode checks, this is part of configureStore itself, not getDefaultMiddleware().

This can be configured via the new duplicateMiddlewareCheck option.

Other Changes

createEntityAdapter now correctly handles adding an item and then applying multiple updates to it.

The generated combineSlices selectors will now return the same placeholder initial state reference for a given slice, rather than returning a new initial state reference every time.

useQuery hooks should now correctly refetch after dispatching resetApiState.

What's Changed

• Process multiple update for the same new entity by @ demyanm in #4890
• add infinite query support for updateCachedData by @ alexmotoc in #4905
• add infinite query skip support by @ alexmotoc in #4906
• Cache initial state in injected scenarios by @ EskiMojo14 in #4908
• Rewrite providedTags handling for better perf by @ markerikson in #4910
• Allow standard schemas to validate endpoint values by @ EskiMojo14 in #4864
• fix(rtk-query): useQuery hook does not refetch after resetApiState by @ juniusfree in #4758
• Drop TS 5.0 from compat matrix by @ markerikson in #4925
• Add duplicate middleware dev check to configureStore by @ markerikson in #4927
• Improve duplicate middleware error and save build output by @ markerikson in #4928
• Add meta handling for infinite queries by @ markerikson in #4939
• improve stability of useInfiniteQuerySubscription's return value by @ EskiMojo14 in #4937
• Add catchSchemaFailure, and docs for RTKQ schema features by @ EskiMojo14 in #4934

Full Changelog: v2.6.1...v2.7.0