Re: terminological nit

[Steve Deering <deering@parc.xerox.com>]

Even nittier nits:

In "supporting unidirectional paths in the Internet", the word "paths"
should probably be changed to "links" or "subnetworks".  If we support
unidirectional links properly, unidirectional paths (i.e., multiple,
contiguous unidrectional hops) should just automatically work.

In the text of section 2.1 of that document, and in Figures 1 and 2, there
are references to a "Server", which seems irrelevant to me.  The server/
client relationship is assymetric, whereas we need to support symmetric
relationships between machines at either "end" of a simplex link, e.g.,
as well as "home" clients accessing servers "inside" the Internet,
clients in the Internet might access a server in someone's home, and
there are non-client/server-based applications, like telephony and
conferencing.  Is the notion of "server" somehow relevant to the problem,
as INRIA sees it?

Steve