[r6rs-discuss] Comparison procedures' number of arguments

[Elf]

On Sun, 19 Oct 2008, Andre van Tonder wrote:

> On Sun, 19 Oct 2008, John Cowan wrote:
>
>> The fact there's no dispute about the meaning of more than one argument,
>> whereas the interpretations of zero and one arguments are all over the
>> lot, suggests that the restriction is anything but arbitrary.
>
> They may be all over the lot, but I do not believe they are equally natural.
> There seems to be only one natural interpretation.
>
>> The true
>> underlying domain, I think, is not numbers but ordering itself (there
>> is no reason why these functions could not be polymorphic, as they are in
>> many other languages), and in a domain where there exist (timelessly)
>> only one item, or zero items, ordering simply doesn't apply.
>
> Are you saying you cannot sort an empty or one-item sequence?  I have
> directories with zero or one entry on my computer, and I can order them fine
> using various criteria without my OS complaining, and that's Windows!  ;-)
>

yes, but in this case, on any criterion, sorting increasing and sorting 
decreasing would give the same result, meaning < and > have the same result.
this is, of course, impossible.  the definition of monotonic increase and 
decrease relies on pairwise comparison; a sequence with one element is neither
increasing nor decreasing.

-elf