Jim Rees scripsit:

> In light of this,  I notice that it does not specify the order in which two
> middle values of an even-length vector are passed to the median procedure.
>
> The order should be the same as they appear in the sorted vector.

Indeed.  Fixed.

--
John Cowan          http://www.ccil.org/~cowan        xxxxxx@ccil.org
My confusion is rapidly waxing
For XML Schema's too taxing:
I'd use DTDs / If they had local trees --
I think I best switch to RELAX NG.