On 3/1/06, Caitlin Bestler <[EMAIL PROTECTED]> wrote:
> [EMAIL PROTECTED] wrote:
> > Aren't attach_mcast and detach_mcast also ib-specific?
> >
>
> Only until transport neutral unreliable datagrams are defined,
> or to be precise when transport neutral QP-centric unreliable
> datagrams are defined (SOCK_DGRAM is quite arguably an existing
> transport neutral unreliable datagram API).
>
> But that leads to an interesting point. When something *might*
> become transport neutral, do we leave it in a transport neutral
> location, or put it in a transport specific location on the
> theory that the eventual transport neutral definition might
> be somehow slightly different?

My preference is to put thing in transport specific locations until
there's a transport neutral version in the neutral location.  This is
the same reason we don't add features that aren't needed by any ULP.

- Fab
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

Reply via email to