Johnny Billquist wrote:
> Having looked a bit more, I'd say yes. Also, the encode functions don't 
> code NaNs the way DEC is saying either.
> 
> See http://h71000.www7.hp.com/doc/73final/4515/4515pro_013.html for a 
> full description of the FP format of the VAX.
> 
> I wonder if they have chosen another representation of NaN on a VAX? 
> They seem to code a NaN the same way they code an INF, which don't even 
> exist as a representation on the VAX.

Correction: they code INF just as the maximum fp number you can have. 
That is strictly not INF, but atleast it's an approximation.
But why they encode a NaN as the maximum float is beyond me. Since a NaN 
can be coded on the VAX.

	Johnny