Andreas Seltenreich <seltenreich@gmx.de> writes:
> sqlsmith caused another internal error while testing REL_11_STABLE at
> 1b9d1b08fe. The query below on the regression DB yields "plan should
> not reference subplan's variable" for me.
Thanks for the report. Seems to be wrong order of operations in
inheritance_planner: it shouldn't be appending extra SUBQUERY_RTE copies
to a subroot that it's going to use as a parent for later iterations.
That results in too many copies of the subqueries and screwed-up RTE
numbering in the children.
regards, tom lane