AW: Mapping?

Kenneth Whistler kenw at sybase.com
Thu Dec 3 02:28:56 CET 2009


> The opponents of PVALID for these contentious characters are claiming  
> not that they're useless.  

Exactly. In fact I suspect that we would all go further and
stipulate that they are:

  A. Fully useful.
  B. Required for orthographic spelling distinctions.
  C. In wide use.
  D. Important.
  
And frankly those (and other reasons) are why they are separately
encoded in Unicode (and in every other significant character
encoding that covers German or Greek character usage).

So there really is no point in going on and on in these threads
about how important these characters are, and how spelling
distinctions need to be made with them, and so on and so on.
We get that already.

> They're claiming that the change from  
> IDNA2003 is harmful. 

Exactly.

It isn't that you can't use a sharp-s or a final sigma in
IDNA2003. You just can't make a domain name *distinction*
from "ss" or medial sigma, respectively, for them.

If we change that in IDNA2008, and if every implementer doesn't
end up using a mapping external to the protocol that results
in the same behavior for these characters as in IDNA2003,
my understanding is we are going to end up with identical
links mapping to different punycode, depending. Depending
on what? Well, we're not really sure, and for darn sure the
users won't be able to know. And if they map to different
punycode, then in principle they could end up resolving
to different domains, depending. And I consider that harmful.

As I see it then, it is a matter of quantifying the
potential harm against the claimed potential benefits.

And it may well be that that calculation stacks up
differently for the sharp-s than it does for the final
sigma (or for that matter, the other forgotten
characters here, the ZWJ and ZWNJ).

> It's a serious charge, and we need to focus on it  
> (I've certainly failed to be consistent on that) because it's the core  
> dispute.

I agree that that is the core dispute, for these
characters, at least.

--Ken

> 
> A
> -- 
> Andrew Sullivan
> <ajs at shinkuro.com>



More information about the Idna-update mailing list