On Wed, Oct 22, 2003 at 07:26:07PM -0700, [EMAIL PROTECTED] wrote: > > There is no need to synchronize access to make a reference container > > thread-safe. > If you want to update it, I don't see how you can guarantee updates will > not collide, but I'm willing to be educated, if you can provide a link to a > web resource. "High Performance Dynamic Lock-Free Hash Tables and List-Based Sets" by Maged M. Micheal. It's the first thing you'll find with a Google of the title..
> Actually, it's quite difficult -- it's a chicken-and-egg problem. ++ on > integers is not safe at all -- there are many architectures where this > operation can and will be interrupted. Atomic integer increments are not > available on all of the platforms Xerces-C supports, so some sort of > synchronization primitive is needed. Yes, that is a problem. You'd need an atomic_increment function or somesuch. Can't help you there for the moment - haven't actually really needed one yet. rlc --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
