To: J3                                                     J3/07-111
From: Van Snyder
Subject: Question about MAXVAL and MINVAL
Date: 2007 January 22

The description of the result of MAXVAL specifies that if the argument has
size zero the result has the value of the negative number of the largest
magnitude supported by the processor.  Similarly for MINVAL, except the
result is positive.  It seems clear that processor numbers rather than
model numbers are intended.  But... are -Inf and +Inf allowed as results?