Support Matrix

This page brings the current repo support matrix into the docs site so readers can stay inside the website instead of jumping out to Markdown in the repository.

The upstream Java client remains the comparison baseline for overlapping features. The Go SDK aims to match Fluss behavior while shaping the public API in an idiomatic Go way.

Current Validation Context

The strongest real-cluster support contract today comes from the Fluss + Paimon demo environment under demo/fluss-paimon.

That harness is currently green for the implemented Go client/ surface covering:

• database, table, and partition admin lifecycle
• indexed log append and limit scan
• Arrow log fetch and projection
• all-types log round-trip
• primary-key upsert, lookup, partial update, limit scan, delete, and prefix lookup
• primary-key snapshot batch scan

These behaviors are validated against a live Fluss deployment using the upstream Java client semantics as the behavioral reference.

Legend

• Yes means implemented and usable
• Partial means in progress or intentionally limited
• blank means not implemented yet

Data Operations

Table Type	Operation	Java Client	Go Client
Log	Append	Yes	Yes
Log	Typed Append	Yes	Yes
Log	Scan	Yes	Yes
Log	Scan with Projection	Yes	Yes
Log	Typed Scan	Yes	Yes
Log	Batch Scan with Limit	Yes	Yes
Primary Key	Upsert	Yes	Yes
Primary Key	Upsert with Partial Update	Yes	Yes
Primary Key	Typed Upsert	Yes	Yes
Primary Key	Delete	Yes	Yes
Primary Key	Lookup	Yes	Yes
Primary Key	Prefix Lookup	Yes	Yes
Primary Key	Typed Lookup	Yes	Yes
Primary Key	Batch Scan with Limit	Yes	Yes
Primary Key	Batch Scan (Snapshot)	Yes	Yes

Reference:

• Fluss table overview

Notes for Primary-Key Tables

• The canonical real-cluster E2E now validates the Go client against Fluss COMPACTED KV semantics for upsert, partial update, lookup, limit scan, delete, and prefix lookup.
• These flows intentionally align to upstream Java behavior first, then adapt to the Go-native public surface.
• The current snapshot path already supports schema-id-aware remapping for older snapshot rows onto the requested target schema.
• Real-cluster compatibility is proven for the current RustFS/Paimon demo path, with built-in fetcher support for both s3:// and local file:// or plain paths.
• Broader snapshot portability across additional filesystem backends and cluster variants still deserves follow-up validation.

Notes for Log Projection

• Upstream/server behavior only supports projection on ARROW log format.
• The Go client supports Arrow append/decode plus projection-aware FetchLogWithOptions(...).
• The real-cluster support contract validates projection through a dedicated ARROW log table while the broader all-types harness remains on INDEXED tables.

Data Types

Data Type	Java Client	Go Client
BOOLEAN	Yes	Yes
TINYINT	Yes	Yes
SMALLINT	Yes	Yes
INT	Yes	Yes
BIGINT	Yes	Yes
FLOAT	Yes	Yes
DOUBLE	Yes	Yes
CHAR(n)	Yes	Yes
STRING	Yes	Yes
DECIMAL(p, s)	Yes	Yes
DATE	Yes	Yes
TIME	Yes	Yes
TIME(p)	Yes	Yes
TIMESTAMP	Yes	Yes
TIMESTAMP(p)	Yes	Yes
TIMESTAMP_LTZ	Yes	Yes
TIMESTAMP_LTZ(p)	Yes	Yes
BINARY(n)	Yes	Yes
BYTES	Yes	Yes
ARRAY<t>	Yes	Yes
MAP<kt, vt>	Yes	Yes
ROW<n0 t0, n1 t1, ...>	Yes	Yes

Reference:

• Fluss data types

Admin Operations

Notes for Admin Work

• The canonical Fluss + Paimon harness is green for safe database, table, and partition lifecycle coverage through the public Go admin API.
• Cluster-global admin mutations such as ACL changes, cluster config mutation, rebalance control, and server-tag mutation are implemented in the SDK and covered by mock integration tests, but they remain outside the narrower single-tablet demo contract.

Entity	Operation	Java Client	Go Client
Database	CreateDatabase	Yes	Yes
Database	DropDatabase	Yes	Yes
Database	DatabaseExists	Yes	Yes
Database	GetDatabaseInfo	Yes	Yes
Database	ListDatabases	Yes	Yes
Table	AlterTable	Yes	Yes
Table	CreateTable	Yes	Yes
Table	DropTable	Yes	Yes
Table	GetTableSchema	Yes	Yes
Table	GetTableInfo	Yes	Yes
Table	ListTables	Yes	Yes
Partition	CreatePartition	Yes	Yes
Partition	DropPartition	Yes	Yes
Partition	ListPartitionInfos	Yes	Yes
Snapshot	GetKvSnapshotMetadata	Yes	Yes
Snapshot	GetLatestKvSnapshots	Yes	Yes
Snapshot	GetLatestLakeSnapshot	Yes	Yes
Bucket	ListOffsets	Yes	Yes
Cluster	AlterClusterConfigs	Yes	Yes
Cluster	DescribeClusterConfigs	Yes	Yes
Cluster	CancelRebalance	Yes	Yes
Cluster	Rebalance	Yes	Yes
Cluster	ListRebalanceProgress	Yes	Yes
Server	AddServerTag	Yes	Yes
Server	RemoveServerTag	Yes	Yes
ACL	CreateAcls	Yes	Yes
ACL	DropAcls	Yes	Yes
ACL	ListAcls	Yes	Yes

Data Lake Formats

Format	Java Client	Go Client
Iceberg	Yes
Lance	Yes
Paimon	Yes	Yes

For Go, Paimon here means the SDK is validated against a Fluss deployment configured with Paimon-backed lakehouse infrastructure, covers the Java-aligned lake snapshot admin surface, and participates in the canonical demo that prints the manual-style hot/cold query behavior through Flink SQL.

Like the upstream Java client module, this does not imply a separate Paimon-native public reader API inside the Fluss client SDK itself.

Reference:

• Fluss streaming lakehouse overview