On 01/05/2015 09:15 PM, Andrew Sullivan wrote: > On Mon, Jan 05, 2015 at 08:16:26PM +0100, Christian Grothoff wrote: >> Usability. Especially on small screens (mobiles, etc.), every character >> matters. > > Who even types domain names any more? On small screens, you don't > type domain names. You use apps. The domain names are embedded in > places. When I use the onion browser on my mobile, I follow links.
And I call people by typing (yes, console) /call nickname.gnu and I prefer to keep doing just that to avoid repetitive strain injury. > In fact, I can see a stronger argument for, "More octets in the name > takes away from the space of the 255 octets we have to work with," > except of course since these names _aren't_ DNS, they don't have that > limit. Except of course maybe they do, because people seem to want > these alternative names to work just fine in every domain name slot. > Fundamentally, this is where the problem lies: every one of these > systems wants to do "DNS-ng" without fixing some of the big > limitations. Right, GNS could be much nicer without IDNA insanity, 63-character label limitations and 255 character limits. But if we do stick to them, then telnet, ssh, and Firefox can use GNS without changes to the application. So this is the catch 22: _some_ compatibility will have to be maintained for some time, because we won't see direct application support until we have many users, and we won't get many users unless there are applications that can use the system. So GNS offers a DNS-compatible API (and even a dns2gns proxy) where it doesn't hurt too much (i.e. the limitations are not that painful for the user). > I have a great deal of sympathy for that desire, because > I agree that "reformat the Internet" isn't really an option. But the > fit is rather awkward. ".alt" is IMO worse. >> Also, we're not "alt" (German for "old"), we're new! DNS is "alt". > > If the primary objection to _that_ draft is the string, the problem is > easily resolved. I'll add markup to my sarcasm next time. >> I personally also refuse to accept that ICANN somehow "owns" the entire >> global name space. > > ICANN does not own it; indeed, the very existence of top level names > in the special-names registry is evidence to that effect. But the > IETF has in fact delegated the responsibility of managing the root > zone to IANA, and the IANA operator is ICANN. Having made that > delegation, it seems rather arbitrary of us to come along and yank > back chunks of it for political reasons. Hence my concern. Not political reasons, these are technical reasons. Usability is a technical concern. Using privacy-preserving, end-to-end secure name resolution is a technical matter. We can't do those with DNS, so we need a (name)space to enable/explore those matters. With GNS specifically, we tell users that the labels match exactly the entities they rely on for resolution (no out of bailiwick, no glue or other funny business). If you append some semi-random DNS name, you destroy this key aspect of usable security where the user's intuition about what is going on matches what is happening.
signature.asc
Description: OpenPGP digital signature
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
