In general these identifiers can represent either subexpressions of the expressions in the original program, or even new expressions that do not appear in the program (e.g., when stddev is expanded), so there many be no column name associated with an expression.
Mihai ________________________________ From: He, Guanlin <gbh5...@psu.edu> Sent: Monday, September 23, 2024 10:02 AM To: dev@calcite.apache.org <dev@calcite.apache.org> Subject: Question regarding calcite logical algebra's ordinal place holder Hi developers of calcite. I am wondering if there is a way to change the ordinal in RelNode's relational algebra into the field name. For example if I have logical planning like: LogicalProject(name=[$1]) LogicalJoin(condition=[=($0, $6)], joinType=[inner]) How could I change back into: and only output logcal algebra string. LogicalProject(name=[student.name]) LogicalJoin(condition=[=(student.id, section.id)], joinType=[inner]) I am currently iterating through relNode and fill exps by their rowType but this method is really mess and need alot of pre condition ( for example, case join and case agg). I am wondering if there is a easy way that I dont know could do that. Best, Guanllin He
