Re: first sight

From: bill_kempf (williamkempf_at_[hidden])
Date: 2002-01-16 13:48:55

--- In boost_at_y..., "terekhov" <terekhov_at_d...> wrote:
> --- In boost_at_y..., "bill_kempf" <williamkempf_at_h...> wrote:
> > --- In boost_at_y..., "terekhov" <terekhov_at_d...> wrote:
> > > --- In boost_at_y..., "David Abrahams" <david.abrahams_at_r...> wrote:
> > > > Do you feel that someone is failing to account for the 
> > considerable
> > > > experience of the pthreads committee? That would certainly be 
a 
> > > mistake, but
> > > > I didn't notice it happening.
> > > 
> > > For example:
> > > 
> > > http://groups.google.com/groups?as_umsgid=3C33232D.EE55A880%
> 40web.de
> > > 
> > > My problem is that under PTHREADS, a thread 
> > > reference/id value object (pthread_t) provided 
> > > by pthread_self() IS SHARABLE and I use this. 
> > > Boost.Threads, AFAIK, does not provide such 
> > > functionality, unfortunately.
> > 
> > The link talks about a totally different subject from whether or 
> not 
> > this is shareable.  A boost::thread object created with the 
default 
> > constructor (equivalent to a pthread_self()) is shareable in any 
> > event.  
> 
> I mean *thread-*shareable:
> 
> "Constructors
> 
>     thread();
> 
>  Effects: Constructs a thread object representing the current 
thread 
> of
> execution.
> 
>  Postcondition: *this is non-joinable.
> 
>  Danger: *this is valid only within the current thread"
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Ahh.  Sorry, I forgot I specified this in the documentation.  This 
may be an overspecification on my part and easily fixed.  I probably 
specified this to make the Win32 implementation as fast as possible.

> > It's just not joinable.  Do you actually call pthread_join() 
> > on references returned from pthread_self()?  Why?
> 
> To let someone else join (compare/cancel/etc) the "current" 
> thread. For example, initial/main thread.

I've yet to see an example where this actually constitutes a problem.
 
> > I still may reconsider this decision.  
> 
> Please.
> 
> regards,
> alexander.
> 
> ps. the other point I was trying to make in my posting
> was that Boost.Thread thread objects are completely 
> different beasts from PTHREADS thread objects, just 
> do a search on "object" in e.g. Compaq PTHREADS-impl 
> docu...

I don't have access to this document.  Could you elaborate?

Bill Kempf