Re: [DNSOP] New Version Notification for draft-adpkja-dnsop-special-names-problem-00.txt

["John R Levine" <[e-mail]>]

> There's talk about protocol switches.  I think that's a misnomer.  There
> are resolution switches.  I see a lot of utility in it being the top-level
> name in a Domain Name.  (I'm not ready to say that's the best way to go.)

Until .onion, the protocol switch for all of the special names was at the 
point where you map a name into an IP address, since localhost and .local 
give you a real IP that you use the same way as any other IP.  You can 
open a TCP socket for web or submit, you can send UDP packets for DNS, you 
can send ICMP packets for ping.  For .test, .example. and .invalid. the 
mapping always fails.

Now .onion comes along and the switch is at a different layer, at whatever 
level SOCKS is.  You can open TCP-like virtual circuits, you might be able 
to do DNS if your SOCKS driver simulates UDP, you can't do ping ping, 
since SOCKS doesn't simulate ICMP.

.onion is a special case for a variety of reasons, but it's not clear to 
me whether people think that slicing at the SOCKS level rather than the 
address resolution level is an exception, or we will be defining a new 
application API for every new special name.

R's,
John