[plt-scheme] Re: to define, or to let (last try)

Paul Schlie <schlie at comcast.net> writes:

> Please, you've got to be kidding:
>
>> This is incorrect.  When I see a fragment of code written thus:
>> 
>> (foo (compute-a) (compute-b))
>> 
>> Then I know that it is equivalent to this:
>> 
>> (let ((arg2 (compute-b)))
>>    (foo (compute-a) arg2))
>> 
>> or vice versa.
>
> - no, they're equally ambiguous if compute-a/b are interdependent.
>   (so in truth, your presumed equivalence doesn't hold in R5RS)

If they are not equivalent, then this is not a legal program under
R5RS. 

>> However were it written like this:
>> 
>> (let* ((arg1 (compute-a))
>>        (arg2 (compute-b)))
>>    (foo arg1 arg2))
>> 
>> I *know* that this:
>> 
>> (let* ((arg2 (compute-b)))
>>    (foo (compute-a) arg2))
>> 
>> is a different program.
>
> - yes, just as it would be a different program than:
>
>   (foo (compute-a) (compute-b))
>
> - which would be equivalent to (if evaluation order were fixed L->R):
>
>   (let ((arg1 (compute-a))
>         (arg2 (compute-b)))
>      (foo arg1 arg2))

These last two programs are equivalent independent of evaluation order.

>   which is actually a far more useful and unambiguous equivalence to have.

How can this be construed as an argument for fixing the order of
evaluation?


----

Incidentally, why not specify the other `unspecified' things in R5RS,
such as the return value of one-arm conditionals or the result of SET!

There is *no* argument for specifying the order of argument evaluation
that does not apply equally well to specifying the return value of
SET! or one-armed IF.