> Thus, the question under discussion is: what should we do under
> sustained overload (or attempts to sustain overload) of the if_snd
> queue?  Specifically, when an app using UDP (or other unreliable
> datagram protocols) uses non-blocking I/O to persistently overflow th=
e
> if_snd queue?
>
> The most correct anwser is: we should drop packets.

First off, as noted by others, I think the entire argument started
with discussion about what we should do in case of *blocking* send()
operations.  In that context, I'm not sure I agree with the above
answer.

There also seems to be a sentiment from some of the participants in
this discussion that says approximately the following: "because the
network might drop a packet further down the line, we don't have to
care at all about any opportunity we might have for intra-host flow
control".  I think I disagree with that.

I'm, like Allen, leaving anything about the application or multicast
out of this equation.

Regards,

- H=E5vard