Ulf Moeller <[EMAIL PROTECTED]> in epsilon.openssl.dev:
>> It is more convenient to have the range; also the previous
>> prototype was misleading because max was larger than
>> the actual maximum.)
> OK, so it would make sense to rename the variable in the prototype.
> But it is a frequent operation to generate a random number in [0..n-1]
> and [1..n-1] and the function should facilitate both.
BN_rand_range(min, max) (previous semantics) cannot be efficiently
implemented for the general case without using an additional temporary
variable range := max - min. Only if 'min' is small, it makes sense
to evaluate BN_rand_range(0, max) repeatedly until the result
is >= min.
We could change BN_rand_range so that it always computes 'range' from
'min' and 'max'. But this would mean unnecessary computational work
since often 'min' *is* small.
Alternatively, we could write an additional function for the special
case that 'min' is small or even for the special case min = 1. Then
only this special function would include the 'r < min' loop. But
since such loops can be easily implemented in the code that uses
BN_rand_range (just two lines, 'do' and 'while (...)'), I don't think
we need one more library function.
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
