Subj : Re: RCU+SMR
To   : comp.programming.threads
From : Joe Seigh
Date : Tue Jun 28 2005 04:42 pm

Joe Seigh wrote:
> Joe Seigh wrote:
> 
>> Yeow!  I've been using hazard pointers like RCU.  That plainly does
>> not work.  Different rules.  Time for a testcase rewrite. :)
>>
> 
> Interesting.  Reference counting wouldn't be an option here as it
> would be for normal hazard pointers.  Plus arbitrarily requiring
> reference counting turns reference counting into a problem rather
> than a solution.  I think I have something that will work though.
> 
w00t! (works, I think)  No testcase rewrite needed.

If C++ goes with deferred reference counting, they're going to run into
the same issues RCU+SMR would have with reference counting so it wouldn't
even be very feasible.

-- 
Joe Seigh

When you get lemons, you make lemonade.
When you get hardware, you make software.

.