Re: [boost] [thread] Question about MT API C++0X N2094

From: Peter Dimov (pdimov_at_[hidden])
Date: 2006-11-01 15:08:54

Roland Schwarz wrote:
> Peter Dimov wrote:
>> Indeed, sorry for not paying attention.  But the only point of
>> locking a mutex is to be able to do that from two different threads
>> at the same time. If you know that only one thread is accessing the
>> scoped_lock at a time, you wouldn't need a scoped_lock in the first
>> place.
>
> Could you please provide an example where it is not possible to
> solve the problem with scoped_lock only? I simply cannot find
> one. Currently this would be the only "killer argument" for my
> proposed aggregate/POD type mutex. Therfore I am very interested
> in this.

The problem mainly arises when you have to implement an interface that has 
separate lock and unlock functions.

class X
{
private:

    mutex mx_;

public:

    void lock();
    void unlock();
};

If you could change the interface of X and force its clients to use 
X::scoped_lock for locking, there wouldn't be a problem. But sometimes you 
can't; for example, you may be implementing a C API or a COM interface, and 
there is no way to impose scoped locks on their clients.

If the entire program can be persuaded to use scoped locks, direct 
lock/unlock calls are probably never necessary (given movable locks, 
otherwise you can't return a lock from a function).