Re: Parallel Scan Bug: invalid attnum: 0

Steve Randall <srandall@s3.com> writes:
> I have been able to reproduce the error in my development environment.
> EXPLAIN output is attached.  EXPLAIN ANALYZE fails with the error
> mentioned.

Ah, I've duplicated it.  My previous attempt to reverse-engineer your
test case was not selecting a partial-aggregation plan.

Immediate impression is that the logic for planning partial grouped
aggregation did not get the zero-sort-keys case right.  In my hands
it produces

TRAP: BadArgument("!(nkeys > 0)", File: "tuplesort.c", Line: 763)
2016-11-10 10:44:26.955 EST [14415] LOG:  server process (PID 15726) was t=
erminated by signal 6: Aborted

but in a non-assert build of course you'd get some other misbehavior.

            regards, tom lane