[concurrency-interest] Lock-free mania

Brian Goetz brian at quiotix.com
Tue Apr 17 20:44:57 EDT 2007


I think that's right.  Just as eight years ago the same crew built
elaborate thread-based mechanisms to solve problems that were entirely
compute-bound and deployed on single-processor systems -- which would
have been more effectively solved with a sequential approach.

But its not productive to try and "outlaw" foo-free algorithms, any more
than it was productive to outlaw threads -- the best we can do is
educate people when NOT to use them.  Of course, teaching people that
they're not as smart as they think is an uphill battle.  Time tends to
be a better teacher of these things than those who actually know better.

Larry Riedel wrote:
> I think if there was a question of will the increasing availability and
> trendiness of lock-free, wait-free, obstruction-free, blahblahblah-free
> tools lead to hordes of "architects" and "principal engineers" choosing
> those tools for no technically justifiable reason to create inferior and
> broken solutions to nonexistent problems, I think it would be safe to
> agree the answer is yes.  But I think the overall net impact of the
> availability and trendiness of those tools is still going to be positive.
> 
> 
> Larry
> 
> _______________________________________________
> Concurrency-interest mailing list
> Concurrency-interest at altair.cs.oswego.edu
> http://altair.cs.oswego.edu/mailman/listinfo/concurrency-interest



More information about the Concurrency-interest mailing list