Re: Intended meaning

From: Francisco J. Lopez Fraguas <fraguas_at_sip.ucm.es>
Date: Thu, 08 Nov 2007 09:32:15 +0100

> I believe that CRWL ignores residuation, which was the ONLY
> problem in my original posting.

Sorry, this was not apparent to me. In any case, the discussion evolved
to other issues also and, in particular, my posting replied to one
of Bernd not specifically dealing with residuation.

> understanding is that we do not yet have a semantics for the
> following problem: a variable is shared by the arguments of a
> choice and the variable residuate in one argument whereas narrows
> in the other.

Maybe you are right, and some aspects of residuation
need further clarification. However, you wrote:

> How can we fix this? Abolish residuation (thanks Bernd)."

Best,

Paco
----- Original Message -----
From: "Sergio Antoy" <antoy_at_redstar.cs.pdx.edu>
To: <fraguas_at_sip.ucm.es>
Cc: <curry_at_lists.RWTH-Aachen.DE>
Sent: Wednesday, November 07, 2007 6:47 PM
Subject: Re: Intended meaning


>
> Paco writes:
>> Is it related to the property that C[e ? e'] should be
>> equivalent to C[e] ? C[e'], for any context C and expressions e,
>> e' ?
>
> At the level of this discussion "equivalent" should be clarified.
> Informally (to avoid any specialized semantics) I would guess that
> any value derived from the first can be derived from the second
> and vice versa. The difficulty is to clarify what "to derive"
> means, see below.
>
>> I remark that the property is certainly true for CRWL
>> programs.
>
> I believe that CRWL ignores residuation, which was the ONLY
> problem in my original posting.
>
>> An interesting point is that the proof of it is easy
>> if one uses semantic-based reasonings.
>
> The proof is easy in other settings, too, depending on the
> notion of "equivalence".
>
>> Thus, an operational procedure can use the above equivalence
>> while preserving the semantics.
>
> I agree completely. My concern is reversing this principle, i.e.,
> we have an operational procedure, e.g., backtracking, and we
> define a semantics to comply with the operational procedure. My
> understanding is that we do not yet have a semantics for the
> following problem: a variable is shared by the arguments of a
> choice and the variable residuate in one argument whereas narrows
> in the other.
>
> Backtracking over the choice does not even see the sharing. Is
> this the semantics we desire? Other operational procedures, which
> might be more efficient and/or more complete, see the sharing.
> We semantics should they comply with?
>
> Cheers,
> Sergio
>

_______________________________________________
curry mailing list
curry_at_lists.RWTH-Aachen.DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/curry
Received on Do Nov 08 2007 - 09:34:10 CET

This archive was generated by hypermail 2.3.0 : Do Jun 20 2024 - 07:15:09 CEST