Create a CSV sink for streaming annotation chunks

The returned closure can be passed to `explore(sink = ...)` to append each per-model chunk to a CSV file without holding everything in memory.

Usage

annotation_sink_csv(path, append = FALSE)

Arguments

path

Destination CSV path.

append

If `TRUE`, new chunks are appended to an existing file.

Value

A function with signature `(chunk, model_id)`.