If you have two useQuery with different staleTime, would you expect isStale returned from that useQuery to: * Change individually as each staleTime passes? * Change together when shortest staleTime has passed? * Have access to both of the above as separate isGloballyStale/isLocallyStale? - ThreadSky

ephem.dev • 96 days ago

If you have two useQuery with different staleTime, would you expect isStale returned from that useQuery to:

* Change individually as each staleTime passes?
* Change together when shortest staleTime has passed?
* Have access to both of the above as separate isGloballyStale/isLocallyStale?

Comments

alvar.dev•96 days ago

I'm not answering your questions, but I'll tell you that at work we've never had to consider this. So we'd be fine either way.

ephem.dev•96 days ago

Thank you, that is another data point for what I definitely expect, that this mostly important in pretty niche cases.

Can be important if you use isStale to show a manual refetch button etc though.

alvar.dev•96 days ago

Ah I see

ephem.dev•96 days ago

To be clear, this does not have to do with when the query refetches, the query would refetch and update both when the shortest staleTime had passed and options would dictate it (mount, window focus, network reconnect).

ephem.dev•96 days ago

Actually, if a new component mounts with has a separate staleTime that is higher and has not passed yet, but an old visible component has a low staleTime that HAS passed, should the query refetch on mount? (Does today, but should it?)

ephem.dev•96 days ago

I think I’m wrong, I don’t think this would refetch today(?). I think we have some thinga to figure out. 😅

orisomething.bsky.social•96 days ago

You're asking really good questions. I have some this issues related to how Apollo Client's cache works when parts of objects are updates from other query. Apollos doesn't re-render, but it can make memory consumptions really big. Also sometimes "more memory consumption" is better than re-render 🫣

jeff.auriemma.xyz•96 days ago

Have you tried our new fragment APIs? Oftentimes colocating fragments with components ends up resulting in more predictable rendering behavior

orisomething.bsky.social•95 days ago

We actually do use Fragments in a few places. But I wonder will this make cache’s memory consumption smaller? After reading again, I’ve found @ nonreactive that might be useful for some of the cases

orisomething.bsky.social•96 days ago

(I wish I had answers. I have the 2 issues)

dige.dev•96 days ago

I would expect it to follow each useQuery instance, same as refetchInterval etc.

But Im unsure what to expect if you tried using the global queryClient to check whether or not the data is stale…

dige.dev•96 days ago

… it should declare the staleness based on the shortest stale time, but I can see how that’s contradicting 🥹

ephem.dev•95 days ago

I think this is correct. I think there are two types of staleness, global and local.

Local: Is it stale in this context?
Global: Is it stale in any context?

ephem.dev•95 days ago

An interesting quirk of this is that global staleness can go true->false without a refetch if the observer with the lowest staleTime unmounts, but I think this makes sense, no longer any need to revalidate on window focus for example.

ephem.dev•95 days ago

Local staleness should never go true->false without a refetch (unless you explicitly change the local staleTime I guess?), as this could mean hiding that manual refresh-button you were showing based on isStale.

ephem.dev•95 days ago

In 99% of situations I don’t expect this distinction to matter though.

Comments

Posting Rules

Reply