[folbricht]

minio-go v6 has been end of life for years. v7 moves every object operation onto a context and renames a few things, so the S3 store code changes with it:

• minio.NewWithOptions is now minio.New
• GetObject, PutObject, StatObject and RemoveObject take a context.Context
• ListObjectsV2 with a done channel is now ListObjects with a ListObjectsOptions, cancelled through its context
• credentials.Provider gained RetrieveWithCredContext, which StaticCredentialsProvider and RefreshableSharedCredentialsProvider now implement alongside Retrieve

The Store interface has no context to pass down, so the per-object calls use context.Background() — which is what v6 was doing internally anyway, so behaviour is unchanged. Prune already receives a context and now uses it to bound the listing as well as the loop. Worth noting separately: S3 stores still ignore the timeout store option, and now that the client takes a context there is a clean way to wire it up. I left that alone here to keep this a straight upgrade.

Dependency bumps. v7 forces several transitive upgrades through MVS — klauspost/compress 1.16.4→1.19.2, logrus 1.9.1→1.9.4, cobra 1.7.0→1.10.2, pflag 1.0.5→1.0.10, x/crypto, x/sys, x/term. All are required by minio-go/v7@v7.3.0 itself (go mod graph confirms), not incidental to go get.

Verification. The existing S3 tests run against a fake S3 server and pass, but they can't catch changes in signing or request shape, so I also ran a real MinIO server in a container:

• make and extract round-trip a 20MB blob, output compared byte-for-byte
• info reports all 304 chunks present in the store
• an index stored in S3 (s3+http://…/bucket/indexes/blob.caibx) is written and read back, and extract works straight from it
• prune drops the right chunks (377 → 73) and the remaining blob still extracts; the index object in the same bucket is untouched
• a prefixed store (…/bucket/prefixed) round-trips

go test ./..., go test -race ./..., go vet and gofmt all clean.