Re: [BUGS] SELECT "bug"?

The Hermit Hacker <scrappy@hub.org> writes:
> webcounter=> select count(1) from webhit_details_formatted where counter_id = 1;
> [ crash ]

OK, this is a boundary condition that I fixed awhile ago: nodeAgg.c
coredumped if no tuples are selected *and* no attributes are mentioned
in the targetlist.  You can work around it in 6.5.* by writing
count(*) instead of count(1).

We could backpatch this, but I'm not sure it's worth the trouble,
considering the lack of complaints and the simplicity of the workaround.

			regards, tom lane