إبطال التخزين المؤقت المستند إلى الوسوم للتحديث الدقيق

In Next.js 15, the Data Cache persists fetch results across requests and deployments. The challenge is busting exactly the right entries after a mutation.

• Time-based revalidation (revalidate: 60) is coarse and clock-driven.
• Path-based (revalidatePath) clears whole route segments.
• Tag-based lets you label individual fetches and invalidate just those, no matter which page rendered them.

Tags give you granular, surgical cache busting: update one product, invalidate only the fetches that read that product.

You attach tags through the next.tags option on the extended fetch. Each tag is a plain string you choose.

Here a product list fetch is labelled with two tags: a broad products tag and a specific product-42 tag.

• The broad tag invalidates the whole collection.
• The specific tag invalidates one item.

After a mutation, call revalidateTag(tag) from next/cache. Every cached fetch carrying that tag is marked stale; the next request re-fetches fresh data.

This runs inside a Server Action or a Route Handler — server-only contexts where mutations happen.

• revalidateTag does not return a promise you await for re-fetching; it just marks entries stale.
• Multiple tags? Call it once per tag.

Design a small tag hierarchy so you can choose the right blast radius:

• products — invalidate the whole catalog (e.g. after a bulk import).
• product-42 — invalidate a single item (e.g. an edit).
• product-42-reviews — invalidate a related slice without touching the product itself.

Rule of thumb: tag each fetch with both a collection tag and an entity tag. Then mutations can pick the smallest tag that covers the change.

A tag is global to the Data Cache, not tied to a route. If the home page and the detail page both fetch product-42 with the same tag, one revalidateTag('product-42') busts both.

This is the key advantage over revalidatePath: you don't have to know which pages consumed the data.

Not all data comes from fetch — direct DB queries (Prisma, Drizzle) bypass the Data Cache. Wrap them with unstable_cache to gain tag support.

The third argument takes tags and revalidate. The same revalidateTag call busts these entries too.

The cache key parts (second arg) must be static enough to identify the function, but you often want a per-entity tag. Build the tag from the argument and pass it in the options.

Return the function so each call gets the right entity tag, enabling revalidateTag('user-7') precision.

Inconsistent tag strings silently fail — a mismatched tag busts nothing. Centralize tag construction in pure helper functions so reads and writes always agree.

This is standalone, idiomatic TypeScript: the helpers produce the exact same strings used in both next.tags and revalidateTag.

One mutation often affects several cached slices. After updating a product you may need to bust the item, the collection listing, and the search index.

Call revalidateTag for each. They all take effect together before the action's response returns to the client.

• Order doesn't matter — all are marked stale.
• Keep the set minimal to avoid over-fetching elsewhere.

Both revalidateTag and revalidatePath exist for different jobs:

• Use tags when the same data appears on many routes, or when you only want to bust data (not the rendered route cache assumptions of a layout).
• Use path when an entire route's output changed and you don't track tags there.

They compose: a Server Action can call both. Prefer tags for data, paths for whole-page structural changes.

Here is the full loop: a form posts to a Server Action, the action mutates the backend, then invalidates the precise tags. The next render of any page reading those tags gets fresh data automatically.

Note we bust both the entity tag and the collection tag because the edit changes the listing too.

A product detail fetch is tagged ['products', 'product-42'] and is also shown on the home page list (same tag). You edit only product 42's name. What is the most precise way to refresh both views without over-invalidating other products?

You learned granular, tag-based cache busting in Next.js 15:

• Attach tags via next: { tags: [...] } on fetch, and via the tags option of unstable_cache for non-fetch data.
• Use a collection tag plus a per-entity tag so mutations pick the smallest blast radius.
• Call revalidateTag(tag) inside Server Actions or Route Handlers — once per tag, atomically.
• Tags are global: one call refreshes the data on every route that read it, unlike revalidatePath.
• Centralize tag strings in helpers so reads and writes never drift apart.