Invalidation

How to use

invalidate — force a refetch

import { queryAtom } from "@mongez/atomic-query";

// Prefix match: invalidates ["users"], ["users", 1], ["users", 1, "profile"], …
await queryAtom.invalidate({ queryKey: ["users"] });

// Exact match only: invalidates only ["users", 1]
await queryAtom.invalidate({ queryKey: ["users", 1], exact: true });

// Every query in the cache
await queryAtom.invalidateAll();

Invalidation refetches in the background (“silent” mode) — the UI does not flash a loading state for data that is already present.

Background variants (fire-and-forget)

Use these when you do not need to await the refetch (e.g. in an analytics event handler or a non-critical side effect).

// Fire-and-forget — scheduled via requestIdleCallback
queryAtom.invalidateBackground({ queryKey: ["users"] });
queryAtom.invalidateBackground({ queryKey: ["users", 1], exact: true });
queryAtom.invalidateBackgroundAll();

Standalone exports are available for all four functions:

import { invalidate, invalidateAll, invalidateBackground, invalidateBackgroundAll } from "@mongez/atomic-query";

Common pattern: invalidate after mutation

const createPost = useMutation<Post, { title: string }>({
  mutationFn: ({ title }, { signal }) =>
    fetch("/api/posts", { method: "POST", body: JSON.stringify({ title }), signal })
      .then(r => r.json()),

  onSuccess: (post) => {
    // Add the new record immediately (no re-fetch needed for the list itself)
    queryAtom.push(["posts"], post);
  },

  onSettled: () => {
    // Always re-sync aggregates after any outcome
    queryAtom.invalidate({ queryKey: ["posts", "stats"] });
    queryAtom.invalidate({ queryKey: ["dashboard"] });
  },
});

Manual refetch outside React

Use refetchQuery / refetchMultipleQueries when you need to await the result:

import { refetchQuery, refetchMultipleQueries } from "@mongez/atomic-query";

// Throws if the query is not in the cache
await refetchQuery(["users"]);

// Parallel
await refetchMultipleQueries([["users"], ["posts"]]);

Fire-and-forget variants (from outside components):

queryAtom.refetchQueryBackground(["users"]);
queryAtom.refetchMultipleQueriesBackground([["users"], ["posts"]]);

How key matching works

Keys are serialised to canonical JSON with sorted object keys. Prefix matching is segment-aware: a prefix matches if the target key starts with the same sequence of complete JSON elements.

Invalidating ["users", 1] matches:
  ["users", 1]                 ✓  (exact)
  ["users", 1, "profile"]      ✓  (extends the prefix)
  ["users", 1, { role: "admin" }]  ✓

Does NOT match:
  ["users", 10]                ✗  (10 ≠ 1 at that position)
  ["users", 100]               ✗
  ["posts"]                    ✗

Object key order inside a key element does not affect matching:

["users", { role: "admin", active: true }]
// is the same cache entry as
["users", { active: true, role: "admin" }]

Seeding the cache from a server loader

seedQuery writes a pre-fetched value into the cache synchronously. A useQuery consumer that mounts afterwards will skip the on-mount refetch as long as the data is still within staleTime.

// Directly (e.g. in a Remix loader boundary):
import { seedQuery } from "@mongez/atomic-query";
seedQuery({ queryKey: ["users"], data: usersFromServer });

// With a freshness window:
seedQuery({ queryKey: ["users"], data: usersFromServer, freshFor: 60_000 });

<HydrateQueries> is the React wrapper (for Next.js App Router server components):

import { HydrateQueries } from "@mongez/atomic-query";

export default async function Page() {
  const users = await db.users.findMany();
  return (
    <HydrateQueries entries={[{ queryKey: ["users"], data: users, freshFor: 60_000 }]}>
      <UserListClient />
    </HydrateQueries>
  );
}

Cache lifecycle management

// Read
queryAtom.getQuery(["users"]);    // Query | undefined
queryAtom.getData(["users"]);     // T | undefined (just the data field)
queryAtom.isStale(["users"], 60_000);  // boolean

// Remove
queryAtom.destroyQuery(["users"]);  // removes + aborts in-flight fetch
queryAtom.clearCache();             // wipe all entries, abort all in-flight fetches

// GC
queryAtom.garbageCollect(300_000);  // remove entries unobserved for > 5 min; returns count removed
queryAtom.limitCacheSize(50);       // remove least-recently-accessed until ≤ 50 entries; returns count removed

// Stats
const stats = queryAtom.getCacheStats();
// { totalQueries, loadingQueries, errorQueries, successfulQueries, totalDataSize }

Auto-GC

Auto-GC starts automatically on the first useQuery call with these defaults:

• Interval: every 60 seconds.
• Evict entries unobserved for more than 5 minutes.
• Cap cache at 100 entries.

Override the defaults (call before or after the first useQuery):

const stopGC = queryAtom.setupAutoGC(
  30_000,   // interval ms
  120_000,  // gcTime ms (unobserved threshold)
  200,      // maxQueries
);
// Stop it:
stopGC();

Standalone export:

import { setupAutoGC, garbageCollect, limitCacheSize, getCacheStats, destroyQuery, clearCache } from "@mongez/atomic-query";

Key details / Pitfalls

• invalidate refetches silently. Active consumers do not see isLoading: true during a background invalidation refetch — only isFetching: true. This preserves the current data on screen while the refresh runs.

• refetchQuery throws if the key is not in the cache. This is intentional — it signals a programming error (refetching something that was never mounted). Use queryAtom.getQuery(key) to guard if the key may not exist.

• Segment-aware matching is strict. ["users", 1] does NOT match ["users", 10] or ["users", 100]. The match requires each JSON element at every position to be identical, and the target key must start with (or equal) the full prefix array.

• freshFor in seedQuery/HydrateQueries: sets staleTime on the seeded entry. A consumer’s own staleTime option overrides this once the query is mounted. Use Infinity to completely suppress the on-mount refetch for seeded data.

• destroyQuery vs clearing from GC: destroyQuery removes the entry immediately and aborts the in-flight fetch. GC removes entries that have had no observers for longer than gcTime. Use destroyQuery when you want instant removal (e.g. logout); let GC handle routine cleanup.

• clearCache() aborts all in-flight fetches. Any components still mounted that own queries will re-enter the idle state. They will re-fetch on their next render cycle or when next observed.