Ted Dunning wrote:
On Tue, Dec 29, 2015 at 1:24 AM, Dhruv Gohil<[email protected]>
wrote:
Just my 2cents:
I think its into greater good that J/ODBC/Native/REST connectivity
features (for SQL operations) are pushed under "Apache Calcite" Project
instead of Drill kind of projects.
Just like "Apache Phoenix<https://phoenix.apache.org/server.html>" guys
started with Query Server<
https://issues.apache.org/jira/browse/PHOENIX-971>
Drill can/should use Calcite+Avatica<
http://calcite.apache.org/docs/avatica_overview.html> combination to
support native language interfaces.
AFAIK ODBC/Native client development is in TODO and seeking contribution
effort, so if Wojtek selects to work with Calcite Community on this, All
dependent project including DRILL will get the benefit of Python
Connectivity.
Doing the development first in Drill and then porting it back to Calcite if
(when) Calcite gets the rest interface going seems like a better path to
success since it gets results sooner.
FWIW, I think I'll have some cycles to start looking into a Python
implementation in Avatica itself "soonish". I haven't really looked into
Drill's current relationship with Calcite, but would be amenable to
having discussions on what a Python API would look like (the ES project
Wojciech linked and Cassandra are two places I plan to look for
inspiration).
