Re: # in place of a digit.

[Matthias Blume <blume@CS.Princeton.EDU>]

This is an interesting question.  So far I was thinking that 18# is
basically the same as #i180 (which in many implementations is the same
as 180.0).  However, since the property of being inexact doesn't say
*how* inexact the number really is I would think that even 0.0 could
be a ``conforming'' result.  Of course, this would be silly.

Back to the 180. vs. 185. issue:

185. seems to be the right idea, but shouldn't it be 184.5 then?
After all, assuming that `#' stands for [0..9] with equal probability
I would think that using the expected value of the digit is the way to
go.  Likewise, 1## is (100. + 199.) / 2 = 149.5, 1### is 1499.5, and
so on and so forth...

Does this sound right to anybody else? :-)

BTW, I have never used # nor have I ever seen anybody else using it.

-- 
-Matthias