Writing a Sink

A sink implements Sink, defined in the repo’s root sink.go:

type Sink interface {
	Spec() SinkSpec
	Open(ctx context.Context, run RunSpec) error
	Apply(ctx context.Context, b Batch, opts ApplyOptions) (WriteReceipt, error)
	Commit(ctx context.Context) error
	Abort(ctx context.Context) error
	Name() string
}

Spec is the sink’s self-description — identity, config schema, and capabilities (what powers the catalog, same as a source’s Spec).
Open prepares the sink for one run.
Apply writes a Batch under an ApplyOptions.Policy (the write policy governing that batch), returning a WriteReceipt.
Commit or Abort finalizes the run.

SinkSpec.SchemaField names the pipeline-scoped config field that holds where output lands — a schema, namespace, database, or key prefix — so the server can default it from the source connection’s normalized name.

Optional capabilities

A sink can implement staged/transactional writes, key-based upserts, or typed DDL (materializing a resource’s schema ahead of its records). See Optional interfaces.

Registering it

Same pattern as a source: a register.go that registers with the engine, then a blank import wherever it should be available — see Embedding as a library.

Overview

Sources

Sinks

Building a Connector

Optional capabilities

Registering it

​Optional capabilities

​Registering it

Optional capabilities

Registering it