On graphs and equivalences

[Eric Brunner-Williams]

The assertion that in Script X a casing property is required to form 
persistent resource identifiers appears to be an assumption present in 
  Vaggelis' note on the desirability of an aliasing mechanism between 
two or more directed acyclic subgraphs existing in one or more root 
server entries.

This assertion does not appear to have been proved, and it appears to 
be a special case of a requirement to define equivalence classes for 
characters, or their sequences, and necessarily any subgraphs anchored 
from the nodes where elements of the equivalence classes exist.

An approach, "intermediate tables", was proposed in-protocol during 
the course of the IDN WG. The proposal was not adopted and was not 
incorporated in the IDNA2003 specifications.

Another specification, published as RFC 3743, Joint Engineering Team 
(JET) Guidelines for Internationalized Domain Names (IDN) Registration 
and Administration for Chinese, Japanese, and Korean, proposed a 
solution not dependent upon the correct functioning of the IDNA2003 
protocol.

I lack the subject matter expertise to form an opinion on the 
necessity claim for tonos-free repertoire ("upper case") for the 
meaningful formation of resource identifiers, and therefore the 
necessity to form character equivalence classes.

If the necessity claim arises from "words", then it suffers the same 
defect that the claims made for other "word properties", in that they 
are out of scope for a working group engaged in the specification of 
persistent resource identifiers.

Eric