In general I would say if you are trying to connect a new database engine to Calcite, an adapter is probably the way to go. You can use Calcite's existing JDBC driver and then just write planner rules to push down whatever is necessary.
-- Michael Mior mm...@apache.org Le lun. 8 mars 2021 à 06:24, Nicola Vitucci <nicola.vitu...@gmail.com> a écrit : > > Dear all, > > I am exploring Calcite and, as a study use case, I am trying to create a JDBC > interface to a non-relational database that has its own joins, aggregates, > etc. I have been playing with a TranslatableTable and some planner rules, but > given that basically all queries including joins would be pushed down to the > underlying engine, I am wondering whether an adapter (like the core JDBC) > would still be the preferred route or a different approach (custom JDBC > driver + query rewriting engine maybe?) is advised. What would you recommend? > > Thanks, > > Nicola >
