#11168 (merged today on solid-query-v6-pre, targeting @tanstack/solid-query@6.0.0-beta.x) adds a provider-owned hydration channel so QueryClientProvider primes the cache from the server-dehydrated stream and useBaseQuery waits for its own entry to be primed (whenQueryPrimed) before attaching its observer, instead of doing a mount refetch on data that's about to arrive.
useQueries doesn't go through any of that. I read the current packages/solid-query/src/useQueries.ts on solid-query-v6-pre: it builds its own QueriesObserver directly —
const observer = untrack(
() => new QueriesObserver<TCombinedResult>(client(), defaultedQueries(), ...),
)
...
createEffect(() => {
if (!isRestoring()) {
unsubscribe = observer.subscribe((result) => { ... })
}
})
— with no reference to hydrationChannel, whenQueryPrimed, or the executorRan replay-detection flag useBaseQuery added (confirmed with rg "whenQueryPrimed|executorRan|hydrat" over the file: zero matches). The observer is constructed and its optimistic result read synchronously on mount, so if this runs before the provider's channel has applied hydrate() for these query keys, useQueries sees a cold cache and behaves exactly like the pre-#11168 useBaseQuery did: no priming coordination, so a query whose data is already in flight from the SSR stream can still trigger a mount refetch even though it's well within staleTime.
This isn't a guess — the PR body says so explicitly under "Coexistence with TanStack Start": "useQueries does not go through this path (it did not serialize hydration data before either); with the channel in place its queries would be primed too as soon as they settle on the server — client attach for it could be covered in a follow-up."
Since this is still pre-release (solid-query-v6-pre, not yet on main), flagging it now while the hydration channel work is fresh in mind seems more useful than after the branch merges and beta ships. Happy to help verify a fix against the same hydration.test.tsx SSR/hydration fixture #11168 added if that's useful.
#11168 (merged today on
solid-query-v6-pre, targeting@tanstack/solid-query@6.0.0-beta.x) adds a provider-owned hydration channel soQueryClientProviderprimes the cache from the server-dehydrated stream anduseBaseQuerywaits for its own entry to be primed (whenQueryPrimed) before attaching its observer, instead of doing a mount refetch on data that's about to arrive.useQueriesdoesn't go through any of that. I read the currentpackages/solid-query/src/useQueries.tsonsolid-query-v6-pre: it builds its ownQueriesObserverdirectly —— with no reference to
hydrationChannel,whenQueryPrimed, or theexecutorRanreplay-detection flaguseBaseQueryadded (confirmed withrg "whenQueryPrimed|executorRan|hydrat"over the file: zero matches). The observer is constructed and its optimistic result read synchronously on mount, so if this runs before the provider's channel has appliedhydrate()for these query keys,useQueriessees a cold cache and behaves exactly like the pre-#11168useBaseQuerydid: no priming coordination, so a query whose data is already in flight from the SSR stream can still trigger a mount refetch even though it's well withinstaleTime.This isn't a guess — the PR body says so explicitly under "Coexistence with TanStack Start": "
useQueriesdoes not go through this path (it did not serialize hydration data before either); with the channel in place its queries would be primed too as soon as they settle on the server — client attach for it could be covered in a follow-up."Since this is still pre-release (
solid-query-v6-pre, not yet onmain), flagging it now while the hydration channel work is fresh in mind seems more useful than after the branch merges and beta ships. Happy to help verify a fix against the samehydration.test.tsxSSR/hydration fixture #11168 added if that's useful.