Subject: [sv-ac] boolean implication
From: John Havlicek (john.havlicek@motorola.com)
Date: Fri Mar 07 2003 - 14:42:53 PST
All:
I was just reviewing the LRM 0.80 and was reminded that
boolean implication (11.6.7) is not the same as
sequence implication (11.6.8), although both use the
syntax "=>" for the operator.
I proposed a long time ago that, for semantic reasons,
sequence implication should be moved to the property layer.
My proposal left boolean implication at the sequence building
layer, and only moved sequence implication to the property
layer. I withdrew my proposal for some reason--I can't
remember exactly why.
Surrendra's proposal, which was voted on and approved, moved
both boolean implication and sequence implication to the
property level.
This is too bad, because nestable boolean implication
as a sequence-building operator is useful. It is not
essential, because one can code
b => r
as
!b || r
Then
(1; a => (1; a1 && (b => (1; b1 && (c => 1;d)))))
can be written as
(1; !a || (1; a1 && (!b || (1; b1 && (!c || 1;d)))))
although to my eye the latter is less readable.
Personally, I am happy for boolean implication to exist as a
nestable sequence-building operator. It is not clear to me
that using the same syntax as sequence implication is a good
idea.
So, at the 11th hour, I propose that we consider returning
boolean implication to the sequence layer, with a different
syntax from sequence implication.
Best regards,
John Havlicek
This archive was generated by hypermail 2b28 : Fri Mar 07 2003 - 14:44:02 PST