2009/9/28 Alaric Snell-Pym <[email protected]>:
>
> On 28 Sep 2009, at 5:31 pm, John Cowan wrote:
>> That strikes me as over-specification.  Most people in other languages
>> don't worry much about the range of the largest kind of fixnums they
>> have,

That's generally because they are either lazy or sloppy. Very few
people I have ever met bother with range analysis. And at my current
company we had to scramble when A client wanted our system to handle a
couple of orders of magnitude more transactions than our design point
and we realized that we'd run out of ids in 90 days. It's CRUCIAL
information. That's why C has the limits.h header. Scheme *needs* the
analogy for it's numeric tower otherwise it is difficult to write code
with predictable accuracy and performance.

>> and it should not be a requirement that Scheme implementations signal
>> an error on fixnum overflow -- returning an inexact result should
>> still be fine.

That gives me the screaming heebie-jeebies. It really does.

> 2) (as an SRFI) implementations may provide special sets of numerical
> operations with prefixes - fx+ fx- etc - that have special semantics:

Please no. It is far simpler to just provide the O(1) limiting size.
If inexacts are good enough for overflow they're good enough for the
entire computation.

david
-- 
GPG Public key at http://cyber-rush.org/drr/gpg-public-key.txt

_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss

Reply via email to