> > > I really need a Standard Deviation aggregate function...
> >
> > I thought about this a long time ago and had an idea but never
> > got around to trying to implement it. I was going to have some
> > functions that worked on a structure of two doubles to track
> > the sum and square instead of using only one simple type.
>
> I remember talking about this to someone, and the problem is that you
> needed the average WHILE scanning through the table, which required two
> passes, which the aggregate system is not designed to do. I may be
> wrong on this, though.
I had asked you how to calculate this and the variance early last
year. One (I think the variance) was two-pass because of the need
for the average, but I thought the StdDev would work with the struct.
Been a while and I still haven't configured #(*&^ FreeBSD ppp yet.
darrenk
