--On Sunday, June 29, 2008 7:46 PM -0700 Bill Manning
<bmanning@xxxxxxx> wrote:
I'm suggesting it would be helpful if there were an RFC
directing IANA to establish a registry that contains both
labels and rules (e.g, no all-numeric strings, no strings
that start with 0x and contain hexadecimal values, the
string 'xn--', the 2606 strings, etc.) that specify what
cannot be placed into the root zone. As part of future
IANA actions, any time a protocol defines a new TLD (e.g.,
.local) an entry should be placed into that registry.
Would there be the downside to this?
several come to mind...
heres one. wrt numeric strings, you have examples of the
ambiguity in implementations on -how- to process non-base 10
numbers. but it is clear that hex encoding is -not- tossed
out. how 'bout octal? or base 36? are numeric string
representations now, after 30 years going to be outlawed? if
so, on what basis?
creating a useful RFC that creates a registry and maintains
it in a timely fashion -in this century- seems a bedtime
fable.
The other two things that seem to be getting lost in this
discussion is that one can write all of the RFCs one like, but
rules like this are ultimately useless unless ICANN agrees to
them, presumably via the gNSO, one at a time, and via a PDP
process. The odds of the very-commercially-oriented gNSO
agreeing to the IETF's being able to pull a name out of the
potential namespace by simple IETF consensus (or less) are, IMO,
around zilch. Worse, as we move toward IDN TLDs, the odds are
high that there will be no practical difference between an IDN
gTLD and an IDN ccTLD other than the ability of the operator to
shield itself from any potential ICANN enforcement action --even
of agreements that were signed to get the domain-- by claiming
national sovereignty and the rule that, in general, private
bodies can't sue governments without the permission of those
governments.
The complementary problem is that, in the present DNS
environment in which typing errors are monetized, an effect of a
reservation of "example.com" is registration of, e.g.,
"examlpe.com" and "exmaple.com" as placeholders, perhaps in case
they draw enough traffic that someone would like to buy them
(the web page for the first of these contains an explicit offer
to sell it).
To at least some extent, our reserving a name and (if the use in
plain-text examples, rather than, e.g., MIB examples actually
drives non-trivial traffic toward those names) publicizing it
makes spelling variations of that name more valuable. I don't
know if the gNSO would like that or not, but it seems to argue
that we should be conservation about what names we reserve and
thereby promote.
That has an interesting corollary: to the extent to which
domains, especially web sites, consider unexpected access to be
a potential profit source, rather than an annoyance, using
someone else's domain name in an example may be "welcome free
advertising", rather than a "rude intrusion". One can argue
against use of those names on either basis, but let's stop
pretending that we have complete knowledge of what is going on
here and of the consequences of our actions.
Perhaps we should ask ICANN to reserve all single-letter TLDs
(in any script) for IETF use. That would make the examples very
clear, would avoid driving traffic to alternate sites (since we
would "own" all of them) and would constitue a relatively closed
list :-)
john
_______________________________________________
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf