Re: Proposal Two: Sample Meta Registrar

["Rick H. Wesson" <wessorh@ar.com>]

On Jan 20,  8:53am, Aveek Datta wrote:
> Subject: Re: Proposal Two: Sample Meta Registrar
> Excerpts from mail: 20-Jan-97 Re: Proposal Two: Sample Me.. by David
> Collier-Brown@Cana
> >   I did an early proof of concept of a socket protocol, and a
> > very bad web interface to that, then suspended work pending
> > a clear indication of what to do in the exceptional cases.
> >   It's not terribly interesting, except for the ease with
> > which one can do simple things like this: the man page for
> > sdmpd follows
>
> What are the exceptional cases you speak about?  The only cases I can think
> of are the borderline cases of two requests coming in at exactly the same
> time. Rather than deal with the problem in an "extremely scientific manner"
> I propose that the main server chooses a resolution (like one minute) and
> then scans forward in the queue for up to one minute away. If there are any
> requests for the same domain name (or any other conflict) the server
> randomly chooses one and rejects the other. I think that with the various
> services out there we can synchronize clocks to that resolution.
>

If this 'system' is implemented using threads and mutex locks on the objects
then whomever grabs the mutex when creating the object gets it. Its real simple
because only one thread can have the mutex at a time. Once an object is locked
there is little room for deciding resolution.

-Rick

-- 
Rick H. Wesson