Charter changes and a possible new direction

Andrew Sullivan ajs at shinkuro.com
Wed Jan 21 00:14:25 CET 2009


Hi,

Sorry for the delay in replying.

On Wed, Jan 14, 2009 at 09:46:56PM +0100, Patrik Fältström wrote:

> Note that Paul, as far as I can see, actually suggest two different  
> things that should not be mixed with each other:
> 
> 1. Should we use a rule based approach that can be applied to any  
> version of Unicode?
> 
> 2. Should we still have the exception table, backwards compatibility  
> table, and specifically, the contextual rules?

I think that distinction is just right.  You're also right that, if we
say "no" to (2), then the rule-based approach is a no-brainer.

So the question is, what's the answer to (2)?  My reading of Paul's
draft is that the benefits of (2) are not worth the cost.

> So I claim that what Paul *really* suggest is to remove contextual  
> rules, minimize the exceptions table. Not so much that we should use a  
> table approach.

My reading seems to be what you think, too.  Now, the question is what
other WG participants think.

I don't have to implement IDNA any more, and thinking of it from that
point of view, I completely understand the desires of those whose
needs are met by these additional rules that they be included under
the protocol as well.  But the consequence of doing that is the
complexity that some of us are worrying about.

I know, I know, just use the well-known libraries, &c.  But we are
talking, after all, about data that finds its ultimate expression at
_every_ level of the DNS.  Do we really think that everyone will use
the right libraries and get all the details quite right at every
level?  What if they don't?  I don't know; has anyone done that analysis?

If I think the way I would were I to have to implement IDNA2008 at a
top level domain, I am instantly aware that there is non-trivial work
entailed.  In particular, we have a considerably greater number of
aliases that almost certainly need to be maintained.  Significant
policy needs to be written.  Probably, extensions to EPP are needed.
I argued on this list, a few weeks ago, that the protocol need only
provide the _facilities_ to do those things.  But the more I consider
the objections, the more reasonable they seem to me.  I genuinely
don't know what to say about this.

I would like, very much, to hear whether there are others who are of
two minds on this question, or (for that matter) buy the view that the
advantages of the exceptions, contextual rules, &c. are not worth the
cost. 

A

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


More information about the Idna-update mailing list