Re: mutex_traits<>

From: bill_kempf (williamkempf_at_[hidden])
Date: 2002-01-31 11:30:37

--- In boost_at_y..., Jeremy Siek <jsiek_at_c...> wrote:
> On Thu, 31 Jan 2002, bill_kempf wrote:
> willia>
> willia> Making lock() operations easily accessible is likely to 
lead to
> willia> newbie mistakes.  Not exposing them at all makes it
> willia> difficult/impossible to extend the library, such as with a 
lock_ptr<>
> willia> concept.
> 
> Whatever mechanism we decide on, the name should indicate the fact 
that
> newbies shouldn't use it. For example, mutex_access_private might 
be a
> good name.
> 
> How about make this a wrapper class with member functions that 
forward to
> the real mutex, and this wrapper class is declared as a friend by 
all the
> real mutex classes.

I've already done that much.  The lock.hpp file defines a lock_ops 
class that does precisely this, though it's in detail land instead of 
exposed in the public interface.

The real issue here is actually how to interact with 
boost::condition, however.  I can't quite see how to document this 
well enough in the public interface to allow for mutex extensions 
that work with the condition variable.

Bill Kempf