cache-manager - fix: use store.deleteMany in mdel instead of per-key delete loop

[pavelmaksimov25]

Closes #1696

Please check if the PR fulfills these requirements

• Followed the Contributing guidelines and Code of Conduct
• Tests for the changes have been added (for bug fixes/features) with 100% code coverage.

What kind of change does this PR introduce?

Bug fix / performance — aligns mdel with the bulk pattern mget and mset already use.

Problem

Since #1014, mget and mset delegate to the Keyv store's getMany/setMany, so bulk-capable adapters (e.g. @keyv/redis) issue a single native command. mdel was left out: it looped the key list calling store.delete(key) per key, costing one round-trip per key.

const promises: Array<Promise<boolean>> = [];
for (const key of keys) {
	promises.push(...stores.map(async (store) => store.delete(key)));
}

For N keys across M stores that is N × M round-trips.

Change

Mirrors the mset implementation exactly:

const promises = stores.map(async (store) => store.deleteMany(keys));

M round-trips regardless of key count. Keyv's deleteMany falls back internally for adapters without native multi-delete support, so no fallback logic is needed in cache-manager.

Unchanged: nonBlocking behavior, the mdel event payload, the Promise<true> return value, and error handling.

Tests

test/mdel.test.ts:

• new — every key deleted with a single store call, asserting deleteMany receives the whole key array
• new — one deleteMany per store in a two-store cache
• new — empty key list still resolves true
• updated — the existing blocking / non-blocking tests now spy on deleteMany and assert toHaveBeenCalledOnce(), matching the shape of test/mset.test.ts

pnpm test in packages/cache-manager: 17 files, 89 tests passing, coverage 100% statements / 100% branches / 100% functions / 100% lines. Lint and build clean.

Docs

Added a line to the mdel section of packages/cache-manager/README.md noting that keys are deleted with a single bulk operation per store.

Note on scope

#1014 also mentions hasMany. cache-manager currently exposes no mhas, so adding one is a new feature rather than an alignment — left out of this PR.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9a7c30c045

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve empty-list behavior before bulk deletion

When keys is empty and the cache uses the documented @keyv/redis store, this now invokes its native deleteMany([]), which delegates to Redis DEL without any key arguments and is rejected by Redis. Previously the empty loop produced Promise.all([]) and resolved true; the new empty-list test only exercises Keyv's in-memory fallback, so it does not cover this adapter-specific regression. Return early for an empty array before dispatching to stores.

Useful? React with 👍 / 👎.