Re: Help Please with numbers
On Tue, 02 Aug 2005 06:27:35 -0700, Jason Gade
<guest at RapidEuphoria.com> wrote:
>I think that the programmer would just have to be clear in what he or she
>wants.
I take it you mean the person who designs, documents, and codes sum(),
rather than the person that uses it.
> sum(1) makes as much sense as sum("123").
I did say "Holes in this example aside".
>
>sum("123", 0) should work correctly.
So, is the result {49,50,51}, as in {49+0,50+0,51+0}, or 123? The
answer is obvious once the statement "sum returns a single atomic
value" has been made, but thus far I did not see such. Recall that in
Euphoria {1,2,3}+{4,5,6} yields {5,7,9}.
>sum("123") would produce 150 because it is ambiguous.
The point I was trying to make was that the design and documentation
of the sum() function should remove any such ambiguity. If sum("123")
compiles cleanly and does not trigger a run-time error, then any
result must be valid, unequivocal, and documented.
>
>I still wonder whether it would be a useful feature. How often is implicit
>conversion from string to number needed?
It depends where the data came from, and to some extent how useful you
would consider a sum function in the first place. If we assume that
the main reason anyone would use a sum function rather than inline +'s
is because they are in a [nested] sequence of unknown length, then you
have to ask what is the most likely source? Personally I suspect text
files or web pages will be more common than say results from dir().
Regards,
Pete
|
Not Categorized, Please Help
|
|
