UI vs. Protocol: No clear boundary (was: Re: Consensus Call on Latin Sharp S and Greek Final Sigma)

Patrik Fältström patrik at frobbit.se
Tue Dec 1 09:38:02 CET 2009

On 1 dec 2009, at 09.08, Martin J. Dürst wrote:

> Overall, there is no boundary line between what is "user interface" and 
> what is "protocol element".

I agree with this.

> In particular, Web page authors type (which 
> means user input, i.e. "user interface") IRIs (and therefore IDNs) 
> directly into a Web page, at which point (inside HTML) they become 
> protocol elements. Nobody is there to pull these out and filter them and 
> map them if necessary.


> The ideal answer to this might be "well, once the browsers don't map 
> anymore, the authors will learn their lesson".

From my point of view the mapping has been "too close" to the protocols (if you understand what I mean). Comparisons of DNs, certs, kerberos realms etc have been in a gray zone where I think the world would have been better with either A- or U-label. And not things that can map to other things.

Same for href's in HTML.

But, what about "IRI's or an IDN-email address in plain text"?

The only thing we already know from IDNA2003 is that "the fewer characters that are mapped that are used" the better.

Just like we from day one in 2003 saw that people had to include the Punycoded version of IDNA2003-IDN in hrefs.

We will not be able to force people to do things in one and only one way. Which some people might use as an argument that things will not work.

I do not agree, as I trust developers to do "the right thing" -- regardless of what it is. Specifically if we really do define A-label and U-label, as we do today in IDNA2008.

> Unfortunately, the 
> current tendency of (at least some) of the browser makers seems to be to 
> want to include all the IDNA2003 mappings, including those that 
> essentially remove some IDNA2008 PVALID codepoints (the ones we are 
> discussing in this thread), and then some (essentially all IDNA2003 
> style mappings extended to IDNA2008).

Indeed it looks like it. More about this in a response to a later mail from Shawn.


More information about the Idna-update mailing list