Skip to main content

rpc_blob_minio 2.1.1 copy "rpc_blob_minio: ^2.1.1" to clipboard
rpc_blob_minio: ^2.1.1 copied to clipboard

S3/MinIO storage adapter for rpc_blob, with presigned download URLs and streamed uploads.

2.1.1 #

  • Requires rpc_dart 7.0.0.

2.1.0 #

Changed #

  • Requires rpc_dart 6 and rpc_blob 2.3. Two core majors are spanned here: see rpc_dart's 5.0.0 and 6.0.0 entries — flow control on by default, an expired deadline as RpcDeadlineExceededException on every shape, a stream that ends without a trailer raising UNAVAILABLE, and a handler's bare Exception text no longer reaching the caller.

2.0.1 #

  • RequestTimeTooSkewed is no longer retried. Nothing between attempts corrects the clock, so retrying only turned one error into four.
  • Documented that a retry after a timeout does not cancel the attempt that timed out — the client offers no cancellation — so both can be in flight. The wrapped operations are idempotent, which is what makes that safe.

2.0.0 #

Breaking — one bucket, a prefix per collection:

  • Objects live at <collection>/<id> in the single bucket named by S3BlobStorageOptions.bucket; bucketPrefix is gone. A bucket per collection does not survive a hosted S3, where accounts are capped on bucket count and creating one is heavyweight. Existing data must be copied under its prefixes before this version reads it.
  • collectionSize returns null: S3 cannot size a prefix, and the MinIO admin API that used to answer this reports per bucket, which no longer corresponds to anything. useAdminApi is gone with it, along with the SigV4 signing it needed and the http.Client it leaked per call — the adapter now needs plain S3 permissions rather than admin ones.
  • deleteCollection walks the prefix and deletes in batches instead of dropping a bucket; listCollections reads common prefixes.

Requests per operation:

  • A write was six round trips (bucketExists, then headBlob which is itself statObject + getObjectTags + getBucketPolicy, then putObject, then getBucketPolicy again for the URL). It is one. ensureCollection replaces the per-write bucket check, a missing bucket is repaired from the error instead of polled for, immutableObjects skips the read-before-write for content-addressed stores, and publicRead states what the bucket is instead of asking it per descriptor.
  • listBlobs built each descriptor with a HEAD, making a page of N cost N+1 requests. It reads size and mtime from the listing; includeMetadata — which already existed and promised exactly this — opts back into the HEADs.
  • fetchObjectTags (default off) stops paying a GET ?tagging per head for tags this adapter never writes.

Hosted S3 readiness:

  • Throttling and transient faults are retried with exponential backoff and jitter (maxRetries, retryBaseDelay), and requestTimeout bounds a call that never answers. Listings are deliberately not retried: their paging state lives in the client, so a restart would replay rather than resume.
  • connect takes region, so a hosted bucket does not need a lookup on first use.
  • writeBlob documents that it buffers the object in memory — the checksum is verified over what is stored and a retry has to be able to resend it, which caps object size at memory / concurrent writes until multipart exists.

1.0.5 #

  • Allow to pass raw sqlite db connection to adapter

1.0.4 #

  • S3 adapter now auto-detects public buckets and returns plain download URLs when anonymous s3:GetObject is allowed; presigning is used only for private buckets, still honoring presign* host overrides.

1.0.3 #

  • Added presign-only endpoint overrides (presignEndpoint/presignPort/presignUseSSL/presignPathStyle) for S3/MinIO to sign links on a public host behind a reverse proxy; removed downloadUrlMapper to avoid generating invalid signatures.

1.0.2 #

  • S3 adapter can rewrite presigned download URLs via S3BlobStorageOptions.downloadUrlMapper (useful for serving through a proxy/CDN) and configure presign TTL via S3BlobStorageOptions.presignTtlSeconds; constructor params consolidated into S3BlobStorageOptions.

1.0.1 #

  • S3 adapter now fetches object tags (?tagging) and merges them into BlobDescriptor.metadata without overwriting metadata provided on upload.
  • Added xml dependency to parse S3 tag responses.

1.0.0 #

  • Added S3BlobStorageAdapter (S3/MinIO/Ceph-compatible) storing blobs under <prefix><collection>/<id> with metadata-based versioning and optimistic checks.
  • New S3BlobStorageAdapter.connect(...) helper for quick setup; list/Head/read/write/delete/listCollections wired to S3 operations; descriptors include a presigned download URL.
  • README now documents S3/MinIO usage; pubspec updated with the MinIO client dependency.
  • Initial version.
0
likes
140
points
726
downloads

Documentation

API reference

Publisher

verified publisherrpc-extra.nogipx.dev

Weekly Downloads

S3/MinIO storage adapter for rpc_blob, with presigned download URLs and streamed uploads.

Homepage
Repository (GitHub)
View/report issues

Topics

#rpc #storage #blob #s3

License

MIT (license)

Dependencies

crypto, http, minio, rpc_blob, rpc_dart, xml

More

Packages that depend on rpc_blob_minio