Ashutosh Bapat <ashutosh.bapat@enterprisedb.com> writes:
> The member root in foreign_glob_cxt isn't used anywhere by
> postgres_fdw code. Without that member the code compiles and
> regression passes. The member was added by d0d75c40. I looked at that
> commit briefly but did not find any code using it there. So, possibly
> it's unused since it was introduced. Should we drop that member?
I think you'd just end up putting it back at some point. It's the only
means that foreign_expr_walker() has for getting at the root pointer,
and nearly all planner code needs that. Seems to me it's just chance
that foreign_expr_walker() doesn't need it right now.
> PFA the patch to remove that member. If we decide to drop that member,
> we can drop root argument to is_foreign_expr() and clean up some more
> code. I volunteer to do that, if we agree.
That would *really* be doubling down on the assumption that
is_foreign_expr() will never ever need access to the root.
regards, tom lane