Re: Allowing IP addresses as path-identiites

[Frank Ellermann <nobody@xxxxxxxxxxxxxxxxx>]

Forrest J. Cavalier III wrote:

> Does that mean that someone saying NO for path-identity is
> also saying that they reject uses of IP for diagnostics in
> path-list

No, that part would be still TBD, you added "keyword dot IP"
to the list, Charles proposed "IP dot keyword", and I vaguely
recall a dot-atom-text construct from Harald (but that didn't
work because it contained path delimiters).

> You may argue that an implementor knows whether they are
> inserting a server name or a diagnostic, and that there is
> no issue.  But the grammar only allows path-identity, folded
> white space, keywords, and delimiters.  (Well, it allows a
> tail entry which is a path-identity.)

We'd update the Path-ABNF allowing some kind of diagnostics
if the poll results in "no IPs as path-identity".  With this
idea all problems of IPs elsewhere (Xref or Injection-Info)
are eliminated or rather reduced to "diagnostics in the Path".

IMHO it's cleaner than Charles' approach (a SHOULD NOT for
IPs as path-identity).

> Harald can count that as a "YES", "NO", or a "NO RESPONSE"
> however he likes.

What you said sounds like NO for me, and what Charles said
like YES.
                          Bye, Frank