Struct ChangeStream 

pub struct ChangeStream {
    pub schema: SchemaRef,
    pub capability: CdcCapability,
    /* private fields */
}

An incremental change feed: renderable batches delivered as they are produced, so a UI pane paints arriving changes instead of waiting for the whole scan.

Backed by a bounded channel, so a slow consumer applies backpressure to the producer rather than growing an unbounded buffer.

Fields

schema: SchemaRef

The renderable schema every batch carries (see change_row_schema).

capability: CdcCapability

Which backend produced this feed and why (see CdcCapability).

Implementations

impl ChangeStream

pub fn channel( schema: SchemaRef, capability: CdcCapability, capacity: usize, ) -> (ChangeSink, ChangeStream)

Create a stream/sink pair with a capacity-batch buffer.

pub fn ready( schema: SchemaRef, capability: CdcCapability, batches: Vec<RecordBatch>, ) -> ChangeStream

An already-complete feed — the batches are handed over as-is. Used by backends whose engine materialises the whole result before returning (Spark Connect’s collect does); the consumer still sees them arrive batch-by-batch.

pub async fn next(&mut self) -> Option<Result<RecordBatch>>

The next renderable batch, or None once the feed is exhausted.

pub async fn collect(self) -> Result<Vec<RecordBatch>>

Drain the whole feed (tests, benches, and any caller that genuinely wants the full window before acting).