Re: [DNSOP] Terminology: IDN

[Paul Hoffman <[e-mail]>]

On May 4, 2015, at 8:13 AM, Edward Lewis <[e-mail]> wrote:
> I'm going to ask this as a question, not a comment, directed to those who
> know something about IDN.
> 
> Isn't IDNA2008 a "convention" on top of the representation?  

IDNA2008 is a set of standards-track documents. It is the only standardized way to convert between DNS names and representation of non-ASCII characters in the DNS.

> And, this
> isn't about display, it's about, well, encapsulation, maybe?

It is about those, and about on-the-wire representation as well.

>  And isn't
> IDN tied to Unicode?

For someone being pedantic, your use of "tied to" seems awfully vague. :-) IDNA2008 refers normatively to the Unicode Standard, yes.

> I ask for a few reasons - "non-ASCII" might include raw binary or bit
> encodings other than Unicode.

It does. And there is no standardized way of representing those in domain names in applications yet. If someone creates one and gets it standardized, we can update the definition. I'm willing to bet this will never happen.

> The punycode algorithm maps Unicode (code points) to DNS-label compatible
> strings.  (No-case, etc.)

Yes.

> IDNA 2008 is a treatment of labels generated by the punycode algorithm to
> achieve a number of other goals.

It is also a standardized definition of how to generate those labels.

> (I guarantee you I am missing something in the above, like Normalization
> Forms and such.)

Yes. But note that I'm not suggesting that you read the RFCs, just answering your questions.

> When I use "IDN" I think of a representation of a Unicode string in a DNS
> label according to a bunch of conventions.  

Good! That's a reasonable guess, although a subset of what it really is.

> Perhaps IDN is taking on a
> wider meaning.  I don't know.

The purpose of the terminology draft is to point people towards the general definition of a term, and give pointers to the exact definitions where possible. For "IDN", I believe this is the best we can do. Other suggestions are welcome.

--Paul Hoffman