Repository navigation
🗺️ Roadmap to Query v6 #12669
Replies: 13 comments
ESM onlyWe will drop support for CJS to simply our builds and move the ecosystem towards ESM. We will also drop our status: |
React: Minimum Requirement v19We want to be able to make use of status: |
Switch default scheduler to microtaskOur scheduler currently runs in a context: status: |
Remove deprecated APIsRemove APIs that are marked as Full listquery-core
solid-query
sync-storage-persistervue-query
devtools
status: |
Observer result readerInternal Refactoring in status: |
Remove defensive checks from HydrationBoundaryCleanup. Leftover from when we had no types. context: status: |
Make QueryFn requiredit’s an oversight that context:
status: |
Add Jitter to Exponential BackoffThis is a good idea to avoid the thundering herd problem but technically a breaking change in behaviour, so we will only ship it in v6. context: status: |
Remove enabled in favour of skipTokenThis is where I would want to get most of the feedback, as there is one situation you can’t replicate with While I think this is not an ideal pattern (lazy queries solve this better), I would like to know if people are actually using this before proposing to ship this. status: |
gcTime: last should win, not highestFor some reason, we currently let the highest ever status: |
gcTime: behaviour on the serverWe currently default to Note that we generally need a way to also have context: status: |
Replace
|
Move
|
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
I’m excited to announce that we have started working on the next major version of TanStack Query - v6.
First things first:
Don’t get excited about major versions, please.
Major versions are to break thinks, NOT to add new features. Feature go in minors. I have a talk dedicated to this topic.
Now, that being said, we currently do not plan to have a huge major version where we break how everything works. We would see some wins, especially for bundle size, if we did change the public API to make it more tree-shakeable, we don’t think that this would be beneficial enough to warrant all existing consumers to re-write how they use TanStack Query today.
We’ve also learned from past major versions that breaking a lot is not received that well.
The good news is that TanStack Query is very mature and doesn’t need to break a lot. We’ll mainly use this opportunity to streamline some internals, update tooling and supported ranges and get to a more modern baseline.
How to read this discussion
I will now create a separate comment for each point that is planned in v6. Feel free to upvote, downvote or comment separately. I’ll also likely track the status here to tell you what is already implemented and if it has been released or not.
If you want to suggest new things, please do. This should be a great opportunity to change things in the Query APIs that have always bugged you :)
Note: We will likely extend this list as we find out more. Subscribe to this discussion to stay up-to-date.
Timeline
It’s done when it’s done. This is OSS - everyone works in their free time, which we might or might not have. We also work for free. GTA6 just came out too. We generally don’t do timelines or promises (pun intended). You’ll know when it’s there.
If you want to follow along, the branch where things are happening is query-v6.
The best thing you can do to help us is to provide feedback, try out beta versions and pre-releases as soon as they are ready.
DO NOT send your agents to start implementing some of the things on this list. While we appreciate contributions, please make sure that whatever you do is actually to help maintainers rather than waste their time. How do you know what helps? Ask maintainers. Best place for this is the TanStack Discord.
We will close all unsolicited, fully agentic PRs. Keep our contributing and AI policy in mind at all times.
All reactions