[Openmcl-devel] Non Blocking Nearly Wait Free Hash Table questions...
Jon S. Anthony
j-anthony at comcast.net
Mon Jan 11 14:13:57 PST 2010
On Mon, 2010-01-11 at 16:48 -0500, Gail Zacharias wrote:
> At some point, aren't you just reimplementing locks? I'm getting out
> of my depth here, Gary will have to say whether doing this sort of
> thing by hand is actually any better than using system locks (or
> semaphores or whatever).
Yes, I was thinking that very thing. And maybe locks are implemented
using this very sort of operation, in which case this is just a crappy
reinvention of the wheel...
> Anyway, if you want to experiment, the functions that do this sort of
> thing are ccl::%store-node-conditional and
> ccl::%store-immediate-conditional. Use at your own risk.
I'll check it out, but I am pretty sure this is a lame approach.
More information about the Openmcl-devel