I agree. On Thu, 2005-10-20 at 15:01 -0700, Sean Hefty wrote: > >But that require changes to CM APIs vs a module on top of it > >to parse and populate private data field. > > I'm wasn't advocating this change. What I think needs to be defined here is a > *service* that provides TCP/IP connection semantics, similar to the definition > of SDP. Applications can make use of this service or not, but the goal is > that > all services that use TCP/IP addressing to establish a connection would do so. > OpenIB would provide an implementation of this service. > > The service is defined by one or more service IDs, plus a private data format. > > Moving beyond defining this service to changing the CM REQ, or separating the > definition of the service into a private data protocol and application defined > service IDs seem like a step in the wrong direction. > > - Sean > > _______________________________________________ > openib-general mailing list > [email protected] > http://openib.org/mailman/listinfo/openib-general > > To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
