layer model

Erik van der Poel erikv at google.com
Thu Jan 11 14:20:23 CET 2007


Here's a slightly different take on the layer model:

1. ISO 10646 and Unicode. ISO 10646 talks about 0..7FFFFFFF but
110000..7FFFFFFF are permanently reserved. Unicode talks about
0..10FFFF.

2. Unicode versions. As Unicode grows, more and more characters are added.

3. Unicode properties. As Unicode evolves, more and more properties
are defined for particular processes.

4. IETF specs. The old IDNA2003 was based on Stringprep2003, but the
new IDNA200x may become a stand-alone spec (not based on a
Stringprep200x). IDNA200x and/or Stringprep200x may use a new Unicode
property.

5. Registries. Each TLD registry and each operator of lower-level
domains has a policy for allowed characters and strings in IDNs.

6. Label creation and lookup. Registrars create labels in registries
and user agents lookup FQDNs containing a number of labels.

7. URIs, IRIs and other protocol elements. User agents discover IDNs
inside URIs and/or IRIs and/or other protocol elements such as email
headers.

8. Human users. Human registrants buy labels through registrars. Human
users access FQDNs through user agents.

9. Politics. ICANN and others govern various aspects of IDNs.

In this model, layers 1 and 2 are clearly out of scope of
idna-update at alvestrand.no. However, it is not clear to me that layer 3
is out of scope. Some of the participants here may spin off an effort
to define a new Unicode property specifically for IDNA. Layers 5 to 9
are out of scope, but we may occasionally talk about them or even
reference them in specs.

Erik


More information about the Idna-update mailing list