Comments on Arabic VIP team report

[John C Klensin <klensin+icann@xxxxxxx>]

                                   ICANN VIP Project: Arabic
                                                John C Klensin

---------------------------- 

Note: This review was prepared at the request of the ICANN
Variant Information Project Team and partially supported by
ICANN.  It reflects the author's personal views and may not
reflect the views of ICANN staff, the members of the VIP teams,
or other personnel associated with ICANN.  The author had the
opportunity to do partial reviews of working drafts of this
document and prepared comments for the team.  Some of those
comments are reflected in the report as posted and hence do not
appear here.  Sections of it draw heavily on other work by the
author that bears on the issues discussed.

---------------------------- 

This is a well-written and thorough report that shows the
influence of careful thinking about the script and its use.  It
is careful to distinguish between string considerations for DNS
labels and those that might apply to running text.  The
document also provides a valuable tutorial on the use of
characters from this script.  The report is also noteworthy in
that it exposes topics on which the team was unable to reach
agreement (and even some on which agreement was achieved) and
spells out the tradeoffs and relatied issues rather than simply
noticing lack of consensus.

Like the Devanagari report, but unlike most of the others, the
Arabic report looks beyond strings and naming to find issues
that might affect variants and the practical use of the script.

The use of Arabic Script IDNs is complicated by at least two
issues: 

(1) For some strings that could reasonably be proposed as
labels, the size of the variant label set (the number of
variant labels) might become quite large.  Unless these are
used only to block registratios (which the report appears to
not consider suitable) very long lists may be a problem (see
Section 8 of the report).

(2) Arabic script labels will, in practice, end up embedded in
highly structured contexts (e.g., IRIs) that have been defined
and structured using left-to-right assumptions.  This issue is
discussed in my "Overview" review.