setQueryData should provide an option to NOT reset isInvalidated to false

[kostia1st]

setQueryData should provide an option to NOT reset isInvalidated to false.

The use case is kind of common.

I have a typical application that fetches some data, and sometimes mutates the data.
Once mutation is done on the server, sometimes I want the client data to reload (invalidate), and sometimes I just update it locally, and that's it.

The problem is when I need a combination of both.
And here I start to depend on the order of calling setQueryData and invalidateQueries. Because if I first invalidateQueries to designate some necessary update (including inactive queries, this is important), and then run setQueryData which for instance makes some kind of optimistic update, or maybe does not update anything important, but still assigns a formally new value to some old key (albeit being basically the same) - the isInvalidated flag of that inactive query gets reset to false, and no update will be happening on activation of said (eternally fresh in my setup btw) query.

Currently I just work it around with the use of getQueryState before setQueryData and then invalidateQueries with exact: true afterwards to bring isInvalidated back to true if necessary.

But would love the ability to supply some extra option into that setQueryData's 3rd parameter to prohibit resetting isInvalidated

[TkDodo]

setQueryData puts data in the cache. For the cache, it doesn't matter where data comes from. Could be from setQueryDat, but could also be from prefetchQuery, or from executing the queryFn, or via hydration, or ...

If data gets in the cache, it's considered as accurate. If your data is "older", you can pass dataUpdatedAt to setQueryData (last param), so that we know when the data was created. Hydration does this, as well as persisters.

This will make sure staleTime will be respected, but it will not ensure that data that was manually invalidated will then also be marked as "invalid".

[kostia1st]

Well, my approach to state management is that staleTime is set to Infinity, so updatedAt does not matter and does not change anything (I checked before posting this issue). I understand the motivation of why setQueryData does what it does by default. I'm just asking for a bit more flexibility which results in more convenience of use, that's it. Luckily workaround is there, although I'm not sure if it does influence CPU performance or not.