'any' and 'all' compared with the rest of the Report

Marko Schuetz marko@ki.informatik.uni-frankfurt.de
Sun, 28 Jan 2001 13:46:46 +0100

From: Fergus Henderson <fjh@cs.mu.oz.au>
Subject: Re: 'any' and 'all' compared with the rest of the Report
Date: Sun, 28 Jan 2001 01:24:58 +1100

> On 26-Jan-2001, Marko Schuetz <MarkoSchuetz@web.de> wrote:
> > I don't think we have absolute freedom. Assuming we want 
> > 
> > \forall s : bottom \le s
> > 
> > including s = error, then we should also have error \not\le
> > bottom.
> You lost me here.  Why should we have error \not\le bottom?
> Why not just error \not\lt bottom?

I assumed a semantic distinction between error and bottom was intended
to accurately model the way the implementation would distinguish or defer
the erroneous computation. You are right that without this assumption
error \not\lt bottom would suffice.