[r6rs-discuss] Comparison procedures' number of arguments

[Andre van Tonder]

On Sun, 19 Oct 2008, John Cowan wrote:

> Andre van Tonder scripsit:
>
>> I agree.  It seems clear that the base cases for /all/ of these should
>> be #t.  It would be silly if, for example, whether a sequence is
>> increasing could be changed from #t to #f by removing the last element.
>> Stated another way, all of these are really an intersection (AND)
>> of conditions, and the base (0 argument) case for AND is #t.
>
> That would, however, undermine the otherwise reliable deduction that
> when applied to the same arguments, < and => (and likewise > and <=)
> produce results of opposite truth value.  The same is true of applying <
> or <= to the reversal of the arguments of > and >= respectively.

No, it is not otherwise reliable.  For example,

   (<  1 2 1)  ==> #f
   (>= 1 2 1)  ==> #f  (not #t as you argue)

Andre