Re: [HACKERS] No: implied sort with group by

> > And in v6.1. If b is a space (rather than a NULL), then the behaviour is correct
> > so it must be a problem in grouping NULLs.
> > 
> 
> explain select b,c,sum(a) from foo group by b,c; -- gives...
> 
> Aggregate  (cost=0.00 size=0 width=0)
>   ->   Group  (cost=0.00 size=0 width=0)
>     ->     Sort  (cost=0.00 size=0 width=0)
>       ->       Seq Scan on foo  (cost=0.00 size=0 width=28)
> 
> There sort is there before the grouping operation, so this would seem to point to
> the sort code incorrectly setting something when handling NULLs.
> 
> This doesn't seem like the same bug that Vadim found since a small data set such as
> this one _shouldn't_ be going out to a tape file.

We have a NULL sort patch for psort in 6.3.  Are you running the most
recent sources?

-- 
Bruce Momjian
maillist@candle.pha.pa.us