[ChineseGP] Seeking Advice from the IP on Opinion on Appropriateness for Considering Initial and Subsequent Versions of the CJK LGRs for certain CJK Han Characters

[Edmon Chung]

Dear Members of the IP,

After extensive deliberations, we are first of all happy to report that many
of the identified differences between the CJK communities (in particular
between Korean and Chinese definition of IDN Variant sets) for a majority of
the Han character and IDN Variant sets have been resolved.  We are positive
that we can eventually resolve the issues for all characters.  Nevertheless,
at present, it appears that there may be around 50-60 IDN Variant sets
(involving around 100-150 Han characters) that will remain unresolved
(between CGP and KGP), until much more investigation can be considered.

In view of the many IDN gTLDs and IDN ccTLDs already in the root without IDN
Variants, and operational experience and understanding from the Chinese VIP
Case Study Report that a significant percentage of queries do potentially go
to the IDN Variant domains, which means that the Chinese Domain Name
experience for users around the world is currently disenfranchised further
every day the CJK LGRs cannot be implemented.  Therefore, we believe there
is urgency to move forward with the CJK LGRs even if a small number of IDN
Variant sets remain unresolved.

As such, we are looking at a few options to provisionally handle these
unresolved IDN Variant sets.  One of the options involve rescinding the
characters of concern.  While we consider the different options, we would
like to seek the the IP's confirmation and advice on the following 2
interrelated but separate items:


1. Is it reasonable to expect the possibility of updating the CJK LGRs
within the next few years after a first version is implemented?

Given the complex linguistic history and differing level of operational
experience in Han character IDN registrations and usage, a small number of
characters (and corresponding IDN Variant sets) will require much more
extensive discussion internally and jointly before full consensus may be
reached.  There is now an interest between the CJK LGRs to set aside this
small group of characters in order for a first version of the CJK LGRs to be
implemented, enabling IDN Variants for the corresponding IDN gTLDs and
ccTLDs to be used.

The CJK GPs nevertheless is concerned whether it is possible to submit a
version of the CJK LGRs for consideration and thereupon immediately start
work on the remaining characters, with an aim to update to a next version of
the CJK LGRs within a few years (e.g. in 2 to 3 years' time).  We are
seeking the IP's confirmation and opinion on whether it is a reasonable
expectation that such an approach would be appropriate given the context of
the situation, understanding the overarching principles of stability and
security.


2. If the around unresolved characters are to be first disallowed for
application, how should it be implemented in the first version of the CJK
LGRs? 

We have identified around a number of IDN Variant sets in the Han character
repertoire shared among CJK communities for which a divergence in the
definition of IDN Variant relationship cannot be immediately resolved
(especially between CGP and KGP).  Therefore, the CJK GPs would like to
provisionally disallow application of IDN TLD strings involving those
characters for the first version of the CJK LGRs.

The CJK GPs have identified 2 potential approaches to effect such
disallowance:

Method A: to not include the Affected Characters in the repertoire of the
CJK LGRs

Method B: to include the Affected Characters in the repertoire, but assign
them with type="invalid"

Advantage of Method A may be that the resulting CJK LGRs would be "cleaner",
i.e. that no character is "invalid" for all 3 communities but yet included
in the LGR.  However, Method B has the advantage of being more complete,
i.e. explicitly indicating to the public and to technical implementers that
the CJK Han repertoire should really include those characters, and prompt
interested implementers to find out more about the background for why they
are assigned as "invalid".

We believe both methods should yield the same result technically, i.e. that
the Affected Characters are not allowed for inclusion in a TLD string.
However, we ask the opinion of the IP, which Method is more appropriate? And
if there is a preference from the IP, what the rationale is?  Further, we
ask if the IP has other suggestions?


Finally, we understand and are committed to document the rationale and
specifics of the provisional withholding of the unresolved IDN Variant sets
(and corresponding Affected Characters) should the above action be taken.
Again, we emphasize that this is one of the few options we are considering.

We look forward to your advice and feedback on the matter.

Sincerely,
Edmon Chung

cc: CJK Generation Panels