Sergey Gromov wrote:
> 
> Wouldn't it be better if BigInt implemented share-on-copy, copy-on-write
> semantics?  This really boils down to a single reference counter within
> BigInt's allocated data and working struct constructors/destructors.

Sure.

(Given the lack of activity on this thread, I'm suspecting we just
solved the problem and everything is OK.)

Reply via email to