On 3/21/2018 8:03 AM, Paul Vixie wrote:
dave, i wasn't going to reply at all, since your snark is a turn-off.
Snark? You think my note was "crotchety, snappish; sarcastic,
impertinent, or irreverent"? It wasn't any of those things, though
perhaps it's interesting you thought it was. But only perhaps.
Your note made no obvious sense to me, since it began with what is
really a platitude (that I really do agree with) and ended with a
generality that had no obvious linkage to the current work.
however, john decided to make this thing real, so now i'm stuck with it.
srv has a registry. that's working. that need not change.
adding another registry for other rr types who want to have well known
underscored names will harm nobody and i'm unopposed.
Much of the discussion of the current topic -- previously and now -- has
tended to stray from the pragmatics, whereas that is the only thinking
driving my concerns and suggestions. In particular, some people seem to
have a mystical -- or equally impractical -- view of how name collisions
will be avoided between independent registries making assignments out of
the same name space.
I prefer to take as simple an approach as possible: have a single
registry control all allocations out of a name space.
So if you will please explain:
1. How the SRV underscore registration process and the new
registration process will avoid collisions
2. How people revising existing specs that use _underscore naming
will know that their document needs to use the new registry
it would be helpful.
d/
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net
_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www.ietf.org/mailman/listinfo/dnsop