Fwd: 1788 suggestion:
P-1788 (and Brian: I've forwarded this):
This is some food for thought, from a user.
Baker
-------- Original Message --------
Subject: 1788 suggestion:
Date: Fri, 27 May 2011 10:39:14 -0500
From: Brian Kennedy <BrianK@xxxxxxxxxxxxxxxxxxxxxxx>
To: Ralph Baker Kearfott <rbk@xxxxxxxxxxxxx>, Ulrich.Kulisch@xxxxxxx
CC: Brian Kennedy <BrianK@xxxxxxxxxxxxxxxxxxxxxxx>
(I just guessed at Ulrich's email address, so I copied you Baker since
you are chairing the standards effort.)
Hi!
I'm not a member of the 1788 standards committee... but I've been
developing test cases for some of my interval code based on what the
committee has been deciding. In particular, I used the tables in this:
http://grouper.ieee.org/groups/1788/PositionPapers/ARITHYY.pdf
I don't know if Table 7 (the multiplication table for extended
intervals) will be incorporated into anything official, but if so may I
suggest a few corrections:
= row 6 should say "a2 > 0" instead of "a2 >= 0", so it doesn't overlap
with row 5
= row 7 should say "a1 < 0" instead of "a1 <= 0", so it doesn't overlap
with row 8
= col 6 should say "b2 > 0" instead of "b2 >= 0", so it doesn't overlap
with col 5
= col 7 should say "b2 < 0" instead of "b2 <= 0", so it doesn't overlap
with col 8
Without those changes, I believe the table has a number of ambiguities
when the bounds are zero.
FWIW,
Brian