Update stream description in README #166
Description
per @dshulyak in linked thread: we want to refactor the way we use streams in the API, and need to update the README along these lines.
api is simpler with streaming
- there is no need to enforce order for pagination and correct offsets, current pagination is basically broken. and with streams it is simple
- one endpoint can be implemented for historical and live data
streaming is more efficient
- no need to keep N objects in memory during query
- also client gets first object without waiting for N objects to be read and decoded. same issue with memory but on client side
nuanced behavior can be hidden behind the stream
with streams that seamlessly integrates live and historical data it is possible to eliminate duplicates that otherwise would have to be handled by client
Originally posted by @dshulyak in #165 (comment)