Re: POSSREPs as union types

From: <compdb_at_hotmail.com>
Date: Sat, 5 Sep 2009 16:23:36 -0700 (PDT)
Message-ID: <069a2cd7-d07b-445f-b0fa-780a7a552a5a_at_e4g2000prn.googlegroups.com>



On Sep 4, 1:01 pm, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
> com..._at_hotmail.com wrote:
> > On Sep 3, 1:22 pm, com..._at_hotmail.com wrote:
> > A possrep (possible representation) maps a subset of a
> > tuple type (namely those that satisfy the possrep constraint)
> > to values. Two tuples can map to the same
> > value.
>
> Tuples are values -- no mapping required. In the situation you describe,
> two representations represent the same value.

Sorry, it's not clear to me what you mean.

In TTM a possrep maps ordered components to a value. This is why I originally used cartesian tuples. But TTM also requires names for the components. So I changed to (plain) tuples (with named elements) ("no big deal").

I was using "value" to mean represented value (of the type represented by a given possrep) and "tuple" to mean representing tuple
(of possrep components or selector invocation arguments). I characterized a possrep as a mapping from representing tuple to represented value.

philip Received on Sat Sep 05 2009 - 18:23:36 CDT

Original text of this message