Re: remove_useless_joins vs. bug #19560

Tender Wang <tndrwang(at)gmail(dot)com> writes:
> After 2ebf25e7d70a8, I hit a crash.

Oh, interesting. What we have here is:

> SELECT
> FROM (SELECT CASE
> WHEN NULL
> THEN sample_1.a
> ELSE NULL
> END AS c2
> FROM ...

What's apparently happening is that we pull up the sub-select,
and eval_const_expressions flattens the CASE to constant-NULL,
so we don't see the reference to sample_1.a and conclude that
sample_1 can be dropped from the query. But when we do that,
we try to mutate the original copy of the sub-select's tlist
which still contains sample_1.a, and so we hit the Assert
saying we should no longer see any such Vars.

Not immediately sure what to do about that ... I don't think
that dropping the Asserts would fix it, even if that were
a pleasant answer which it's not. If we allowed the Var
to become a reference to INVALID_VAR, then when we repeat
the eval_const_expressions run later, we're going to have
problems. IIRC, there are places where eval_const_expressions
looks up Var properties such as null-ness, and that'd fail.

regards, tom lane