Maybe: table.subscribe("x", "y")
Consider:
- Should we wait to expose this in the coming multiplexing WS endpoint and keep the /stream/single endpoint simple?
- If we do expose this in the /stream/single endpoint, do we actually need separate endpoints per structure family, like /stream/array/single and /stream/table/single analogous to the REST API, so that endpoints have structure-specific query parameters—such as
?columns=…?
- Should we expose array subselection options too? Like
slice and/or decimate?