Boost.Threads - 'lock' name (was Re: Boost.Thread

From: williamkempf_at_[hidden]
Date: 2001-07-19 15:47:05

--- In boost_at_y..., "Alexander Terekhov" <terekhov_at_d...> wrote:
> 
> > > which is simply a convenient way to deal with
> > > (acq/rel) LOCKS (mutexes, semaphores,
> > > read/write locks, spinlocks,...).
> >
> > A mutex is not a lock, though.
> 
> are you kidding ?

No, I'm not.

> well, fyi..
> 
> http://ace.cs.wustl.edu/cvsweb/ace-cvs.cgi/ACE_wrappers/ace/Synch.h?
rev=4.117&content-type=text/x-cvsweb-markup
> 
> /**
>  * @class ACE_Thread_Mutex_Guard
>  *
>  * @brief This data structure is meant to be used within a method or
>  * function...  It performs automatic aquisition and release of
>  * an <ACE_Thread_Mutex>.
>  *
>  * This class is obsolete and should be replaced by
>  * ACE_Guard<ACE_Thread_Mutex>.
>  */
> class ACE_Export ACE_Thread_Mutex_Guard
> {
> .
> .
> .
> protected:
>   /// Reference to the mutex.
>   ACE_Thread_Mutex &lock_;
>                    ^^^^^^ !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
> 
> 
> also, fyi..
> 
> http://ace.cs.wustl.edu/cvsweb/ace-
cvs.cgi/ACE_wrappers/ace/Synch_T.h?rev=4.90&content-type=text/x-
cvsweb-markup
> 
> /**
>  * @class ACE_Guard
>  *
>  * @brief This data structure is meant to be used within a method or
>  * function...  It performs automatic aquisition and release of
>  * a parameterized synchronization object <ACE_LOCK>.
>                                                ^^^^ !!!!!!!!!!!!!!
>  *
>  * The <ACE_LOCK> class given as an actual parameter must provide at
>  * the very least the <acquire>, <tryacquire>, <release>, and
>  * <remove> methods.
>  */
> template <class ACE_LOCK>
> class ACE_Guard
> {
> public:
> 
>   // = Initialization and termination methods.
>   ACE_Guard (ACE_LOCK &l);

That only shows that the ACE library has chosen to use this name in 
this manner (inconsistently, BTW).  It doesn't mean that the choice 
was correct.  You acquire a lock on a mutex.  This means a mutex is 
not a lock.

Bill Kempf