[r6rs-discuss] "Unspecified" (naming)

[Trent Buck]

On Fri, Sep 22, 2006 at 02:02:42AM +0000, Aubrey Jaffer wrote:
>> Date: Fri, 22 Sep 2006 10:29:31 -0400 (EDT)
>> From: AndrevanTonder <andre at het.brown.edu>
>> 
>> Actually, I agree with another poster that it would be nice if
>> these expressions returned zero values instead of the current 1.
>> In that case, maybe a "void" "type" would not be so far-fetched for
>> such expressions (ignoring for the moment that the language is
>> untyped).
>> 
>> The reason I would like it if these expressions returned zero
>> values is the following: I sometimes get hard-to-debug errors due
>> to inconsistent data values containing #<unspecified>, due to
>> obscure errors in logic, misplaced parentheses, etc.  I can imagine
>> that newcomers would run into this kind of error quite a bit more
>> often.
>> 
>> The problem with these errors is that they often occur long after
>> the call that inserted the #<unspecified> value in the data
>> structure.  If these expressions returned no value, the error would
>> occur immediately, as it really should, and one would not be able
>> to build such inconsistent data values in the first place.
> 
> That is a good idea.  This was not possible in R5RS because of
> 
>   "Except for continuations created by the `call-with-values'
>   procedure, all continuations take exactly one value."
> 
> But R6RS 4.5 says:
> 
>   "The continuations discarding these values accept any number of
>   values."
> 
> The language "value returned is unspecified" would be changed to
> "returns no values"; there would be no object to name.

Is there any time where one would want to distinguish between `no
values' and `the unspecified value'?
-- 
Trent Buck, Student Errant