Fixing display variations (was: AW: Eszett and IDNAv2 vs IDNA2008)

[Andrew Sullivan]

On Fri, Mar 13, 2009 at 05:22:33PM +0000, Shawn Steele (???) wrote:

> The broken and clutzy part is that when the browser displays it the
> ß is lost.  That isn't a requirement to make distinct DNS entries
> for these names.  It is a requirement for fixing display variations.

I really want to put an end to that line of thinking, which has
cropped up more than once in this disucussion.  We said we were out to
internationalize DNS, not fix problems in the way domain names are
displayed in different contexts.

DNS is case-preserving but case-insensitive, which means that most of
the time the answer you receive will be exactly what you put in the
question section (due to the use of compression pointers; it's not
relevant whether that's strictly speaking what the scriptures say the
behaviour ought to be).

Other protocols, we have learned, require that everything be downcased
(presumably so that comparisons become easy, but again, the reasoning
isn't important).

What these "extra display variation prevention" schemes seem to be
adding is a new rule that the preferences of the domain holder need to
be reflected in the way the domain is displayed in clients.  This is a
completely new requirement that is not analagous to any existing
behaviour.  Quite apart from it being a new topic that hasn't been
addressed and that will hold up the WG's progress even more, it also
open a host of new issues about presentation issues that I don't think
we're in a position to answer. 

I really don't think this is a fruitful line of inquiry.

Best,

A
-- 
Andrew Sullivan
ajs at shinkuro.com
Shinkuro, Inc.