On Nov 6, 2006, at 1:06 PM, Bryan Sant wrote:
On 11/6/06, Levi Pearson <[EMAIL PROTECTED]> wrote:
When programming in C, C++, or Java (among others), standard numeric
No, in java you use the BigInteger or BigDecimal class to deal with
big numbers. These are arbitrary-precision numbers. They will grow
until you run out of memory (or address space -- 32-bit systems blow).
Did you not read what I wrote? You must be pretty defensive about
Java to feel the need to defend it against accusations I did not make.
I specifically said 'standard numeric variables' have this problem
and noted later on that there are arbitrary precision libraries you
can use if you need to. The fact is, the standard numeric types in
Java are just like those in C, C++, and even OCaml. They overflow
and wrap around. If you don't anticipate this and explicitly use
arbitrary-precision numbers (or some check for overflow, at least)
where necessary, your program WILL fail for sufficiently large inputs.
Having this kind of data type is a good thing for efficiency, but
treating members of machine-length integers like mathematical
integers is a recipe for disaster when overflows are possible.
Common Lisp errs on the side of producing correct results in the case
of machine integer overflows at the price of efficiency, though you
can force it to always use a machine integer if you wish. The other
languages I listed err on the side of default efficiency and possible
overflow danger. These are, as I said, semantic properties of the
language and I made no value judgement regarding which was better.
--Levi
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
