If you're using React Query you certainly know how the useQuery
hook works. Some example similar to the ones you find in the React Query documentation.
// a simple string.only key
useQuery('todos', ...)
// an array key
useQuery(['todo', id], ...)
// other, more complex keys
useQuery(['todo', id, 'comments', commentId], ...)
useQuery(['todo', {id, type: 'comments', commentId}], ...)
These keys are used to identify a specific query and is most important in combination with react query's caching mechanism. It allows react query to fetch the same query only once even if it is called multiple times in various components, and it identifies the cache to be used when fetching again or invalidating the cache.
In larger applications you'd have to make sure the keys are identical in all components or hooks using the same query or even more important if you want to invalidate the cache (aftera mutation, for example).
The react query documention does not provide a solution to this problem. My solution for this is pretty straightforward. By creating an object with a key
and query
function for each query.
const todoQuery = {
key: (id: string): ['todo', string] => ['todo', id],
query: (id: string): Promise<...> => {... fetch the todos ...},
}
export default todoQuery
Using useQuery would then look like this:
const { data, isLoading } => useQuery(todoQuery.key(id), () => todoQuery.query(id))
I think this is a simple but effective way to make sure the keys are always the same. Even when the keys need to change for some reason, you always alter them for all the places they have been used.
Photo by Joshua Aragon on Unsplash
Top comments (19)
Best thing to do is to create specific functions for your queries:
and so on.
Hooks are composable so make the most of it.
This will not help you when trying to invalidate the query. Of course you could create a invalidate function/hook for each query, too. But then the pattern starts to look not much different from what I'm proposing.
Or you can even use a combination of both: Create custom hooks PLUS use the key/query objects.
Invalidation is a whole other problem for sure. I usually have an invalidate hook that invalidates all queries within a certain domain i.e.
I've been using react query (and vue query) since day 1 and it's important we get some good/consistent patterns in place
Invalidating is easy why not just,
Works for me with normal queries and custom hooks.
Yes sure. But how would make sure that keys stay consistent in your mid-size to large React application? What if the key used by the query at another place of the application (by another person)? How do you make sure the invalidation also gets updated?
This is what my pattern tries to solve.
I don't understand what you mean by "invalidation also gets updated" what you mean by updated? Or do you mean other devs on your team will accidentally use an already existing key for a different query?
This is fine in simple cases but it doesn't scale well.
Often for a single"area" of my application I might have 15 different queries that are all interlinked and need to be invalidated together.
Having to manually invalidate every one of these in every place that updates my data, getting all of the keys correct, making sure I'm passing in the correct parameters every time, etc. It gets messy fast in a larger application.
And then if you add another query later that also needs to be invalidated, you'll be searching all across your codebase trying to find all the other invalidateQueries calls in order to add it. And I 100% guarantee you'll miss one!
Seems to me some organization with context wrappers could help here.
So, let's just assume this (slightly silly) example. In one part of your application you do have this:
Now, in a different part you have this invalidation code:
So far so good. Now let's say someone new in your team is working on a change in the first part and has to change the key for whatever reason. Maybe to this:
Now the invalidation would no longer work (well, technically it still does, because it would invalidate all 'the-key' cache entries, but it would no longer be as targeted). The new team member wouldn't know about it and nobody might notice it (well, hopefully you'd have some tests in place to catch it, but well, sometimes tests don't catch everything).
By using a unified "key generator function" as I propose in my article the keys would stay in sync everywhere they were used for a specific query function.
I'm pretty sure the invalidation still works even after adding
id
dependency no matter how manny deps you add you can always just invalidate them with just the key, I'm pretty sure I have a query in my appYes, this works.
queryClient.invalidateQueries()
would also work. But I'd rather invalidate as little as possible which means I have to be specific with the keys. Also you might want to update the cached data after the mutation (Updates from Mutation Responses) instead of just relying on. In those case you will have to use the exact key.In the end there are a lot of ways to get to the desired outcome (using prefix matching, using the
exact
option, using predicate functions...). You'd have to choose the solution that works for you. I usually tend to think of solutions that are easy to follow and prevent as many undesirable side effects as possible by design (i.E. they require lower cognitive work).I'm sure your use-case requires what you describe, for me creating custom query hooks work fine for now.
It really does
That's 138 unique queries, most of which are used dozens of times across the codebase 😅
From my experience I think you have to explicitly pass
{ exact: false }
to invalidate partial keys. I could be wrong.I prefer to create a separate key map per API namespace. It's looks like this:
In complex apps, this have several advantages:
all
keyrecentlyWatched
key, since all other recently watched keys are based on thisAlso, I don't use
useQuery
across the app. I create custom query hooks in single API packages divided to API namespace and use it in app.I really like this approach. Thanks for sharing.
I think this post explain the concept mentioned by @dikamilo tkdodo.eu/blog/effective-react-que...
this topic requires much deeper elaboration than this
Isn't that the truth for 99% of all topics in coding?