I’m doing some thinking (and prototyping) about the query preparation lifecycle. Preparing a query requires lots of pieces of information (type factory, cost factory, a list of planner rules, metadata providers, executor for reducing scalar expressions, configuration, and many more) and depending on how you use the Calcite framework (through a JDBC driver or using Frameworks or otherwise) you provide those pieces of information in different ways and different times.
Can people chime in on https://issues.apache.org/jira/browse/CALCITE-1536 <https://issues.apache.org/jira/browse/CALCITE-1536>. Issue https://issues.apache.org/jira/browse/CALCITE-1499 <https://issues.apache.org/jira/browse/CALCITE-1499> precedes 1536 (i.e. is less complicated) and https://issues.apache.org/jira/browse/CALCITE-1525 <https://issues.apache.org/jira/browse/CALCITE-1525> depends on it (i.e. is more ambitious). But answering questions such as “what is the latest point in the planning process that I can change the list of active traits?” will help. Julian
