Re: Threads: Barrier Synchronization

From: dmoore99atwork (dmoore_at_[hidden])
Date: 2002-03-07 15:58:27

Hi,

I wanted to run something by you - after posting this library, I got 
a couple of private emails from a "semaphore zealot" and a "spinlock 
zealot" who seemed to view my posting as an opening salvo in some 
kind of thread revolt.  It was really quite humorous.

Anyways, I've been reading/lurking on the threads library for quite a 
while, and it seems that your focus is on getting the core right, and 
in getting the core of the threads library submitted for 
standardization.  Cancellation issues, etc. seem to dominate there.

I've started using boost::mutex and boost::condition in real code, 
and I'm interested in fleshing out the higher-level constructs like 
barrier, rw_mutex, etc. 

Are you comfortable with this occurring in parallel with your 
standards-oriented work in the core?  The two activities seem fairly 
othrogonal to me, and any construct like a thread-safe message queue 
which isn't fundamental enough to be in the standard could at least 
be added to a set of nice examples of using boost::threads...

But, if you think this would be a distraction to working through the 
tough cancellation point type issues, let me know and I'll keep this 
stuff to myself until you launch the Boost::Threads list...

Thanks,
Dave