feat(query-core): add MutationFunctionContext argument to mutateFn #9193
Conversation
add new argument to mutateFn enabling access to queryClient as well as meta object
github-actions
bot
added
documentation
|
we already have a PR for this: can we consolidate the work in one PR? I think we’d want 3 properties passed: |
This PR adds new mutationKey property to MutationFunctionContext that's passed to mutateFn.
|
@TkDodo updated as requested |
|
View your CI Pipeline Execution ↗ for commit 2c1a713.
☁️ Nx Cloud last updated this comment at |
| - **Required, but only if no default mutation function has been defined** | ||
| - A function that performs an asynchronous task and returns a promise. | ||
| - `variables` is an object that `mutate` will pass to your `mutationFn` | ||
| - `context` is an object that `mutate` will pass to your `mutationFn`. Contains reference to `QueryClient`, `mutationKey` and optional `meta` object. |
There was a problem hiding this comment.
there’s one more thing that’s a bit troubling, and that’s the fact that we already refer to something as “context” for mutations: The value that gets returned from onMutate, which can be seen in this example here:
and in code here:
query/packages/query-core/src/mutation.ts
Line 29 in 02c3d51
since it’s part of the public MutationState, which is available via useMutationState, we can’t just rename this, so it’s a bit unfortunate.
Any ideas ?
This PR adds new argument to
mutateFnallowing for access toqueryClientas well asmetaobject.This feature should make it easier to attach authentication headers from within
mutateFn, that would typically live inQueryCache. Currently that's made very difficult and requires either having singleton context in the codebase or requiring consumers of mutations to pass auth token each time.This PR also makes it more similar to how
queryFnbehaves, in that it acceptsQueryFunctionContext.Note: I wasn't entirely sure about the name of
MutateFunctionContext, but ultimately I decided to go ahead with that name to mirror the wayqueryFnbehaves. I realise this might be suboptimal, due to there already existing aTContextfor mutations. Happy to discuss and take direction here.