[openib-general] OpenIB and OpenRDMA: Convergence on common RDMAAPIs and ULPs for Linux

[Roland Dreier]

    Sean> My personal view is that any higher level abstractions could
    Sean> combine common functionality or simplify the interface,
    Sean> possibly with a loss of lower level control.  For example,
    Sean> an interface could be created that allowed the user to
    Sean> operate with more abstract addressing, hiding SA query and
    Sean> CM interactions.  Such an interface may not be needed yet,
    Sean> but if there were multiple RDMA transports, it could be
    Sean> valuable.  However, I do not think this necessarily
    Sean> translates into the need to abstract concepts like PDs, QPs,
    Sean> and CQs.

Yes, that is probably a better way of saying what I was trying to
drive at.  There's a significant portion of kDAPL that doesn't seem
necessary or useful to me.  For example, why do we need a new API for
posting work requests?  Why do we need the ability to combine
completion events and CM events into a single queue?

 - R.