This looks terrific! Being able to get results per partition is
especially cool, as this will enable AsterixDB to play much more nicely
with parallel clients (e.g., interplay with Spark and friends). The
status and metric info coming back is also really nice. We need this!
On 9/29/25 10:45 AM, Peeyush Gupta wrote:
Hi all,
Initiating discussion to enhance the async query API.
Feature: Enhancing the async query API.
Details: Currently there are several issues with the async API. This proposal
aims to improve upon the its current design and implementation.
* Handle returned has the ip address and port. we should return a URI
instead of the complete URL as the handle. The caller should prepend the host
and port appropriately.
* Timeout is not handled for the entire query and only considered only for
the duration till which the handle is returned.
* There is no partition level information.
* The status api does not return when the query was started, how long it
was queued etc.
* There is no way to explicitly discard a result set.
APE:https://cwiki.apache.org/confluence/display/ASTERIXDB/APE+28%3A+Enhancing+Async+API
Thanks,
Peeyush
