Re: Mandatory From field, anonymity, and hacks

Charles Lindsey wrote:
> On Thu, 29 Jul 2004 22:51:35 -0400, Bruce Lilly <blilly(_at_)erols(_dot_)com> 
> wrote:
> > I do have the text, and there is no such "recommend[ation]".
>
> It says
>
>       ".invalid" is intended for use in online construction of domain
>       names that are sure to be invalid and which it is obvious at a
>       glance are invalid.
That is not a recommendation for anything -- it is a statement of intent;
intent for use "online" and not via hard-coding -- and is certainly not
a recommendation for use in address fields.

> > You have missed the point.  Evidently you haven't looked at RFC
> > 3696.
>
> I have now looked at RFC 3696, and what Usefor proposes seems entirely 
> consistent with same. Essentially, it says that if you intend user
> agents  to check for and act upon particular TLDs, then you need to be
> pretty sure  that  your filtering is fully consistent with the relevant
> published  specifications. A registration with IANA that a given TLD
> will _never_  resolve to anything seems as clear in that regard as one
> can get.
You have utterly missed the point of RFC 3696. The point is that
hard-coding domain names is foolhardy; whether that is a hard-coded
assumption that TLDs are 3 characters, or that specific strings are or
are not TLDs, etc.  And you have far too much faith in IANA; RFC 2046
reserves the name "ASCII" which is not supposed to be used as a charset
name or alias -- yet IANA has registered it as a charset alias.  Now,
IANA doesn't control TLDs, ICANN does; whether you have more or less
faith in ICANN vs. IANA is another matter...