Re: [boost] Input archive loads seemtocorrupt sharedcharacterstd::strings

From: Peter Dimov (pdimov_at_[hidden])
Date: 2007-12-05 12:28:06

Robert Ramey:

> I've been reading the posts on this subject with interest.  I don't have
> strong position myself.  I do believe that some extra effort and/or
> risk is justified to get best performance.  I'm leaving for you guys
> to do battle over the various alternatives. I strongly believe
> in const correctness and believe that things are "const" for a reason
> so I'm not enamored with a const_cast of any kind.  I would
> much prefer a "guarenteed correct" solution if its close to best
> efficiency.
>
> Hence I like the current suggestion.  But does't it presume
> that s.begin() points to a real array which would be an implemenation
> feature?.  That is, do I know that a standard conforming string
> is actually an array of adjacent characters?

The standard doesn't guarantee that a std::string is a contiguous array, but 
all existing implementations are. You can use &s[0] and take the risk, or 
you can prefer dispatching on whether s.begin() returns a pointer.