SC22/WG20 N877


Comments on PDAM 1 on ISO/IEC 14651:2001, accompanying a NEGATIVE vote.
(Preliminary, need to be confirmed by Swedish NB)



If the MAJOR comments below are accommodated in an acceptable way, our vote will be changed to a yes.



SE 1.    (MAJOR) Changes to clause 3 should be listed in the amendment:

            * refer to ISO/IEC 10646-1:2000

            * refer to ISO/IEC 10646-2:2001

            Have a 'note' saying that those two together have the same repertoire and code point

            assignments as Unicode 3.1.  References to older editions of 10646 and its amendments

            should be removed.


SE 2.    Annex B: All of the tailoring examples should refer to the new table.


SE 3.    (MAJOR) Annex A: The table should be extended to cover all of Unicode 3.1.  If, during

            the processing of this amendment to 14651, amendment 1 to 10646-1:2000 is

            approved, consideration should be given to cover also those characters.


SE 4.    (MAJOR) The ordering of runic letters should follow the FUÞARK based ordering.

            This is the ordering in which runic letters are traditionally presented.  ABC-ordering

            of the runic letters should be left to tailorings for specific purposes.  Even though

            ABC-ordering of runic letters may have been used during the late period of runic letter

            usage, no evidence of this has been presented by runic experts, and would even so

            be an influence of the then dominating latin script, not the order in which runes where

            originally thought of.  Further, when using runic letters as counters (for calendaric

            calculations, at least), the FUÞARK order was used, much as ABC order of the

            latin script is used for item list numbering.


SE 5.    All Han based characters (CJK URO, CJK ext A, CJK ext B), should be ordered

            together, in the order mentioned (currently ext A comes before the URO).


SE 6.    (MAJOR) All <MIN> subweightings at level 3 that correspond to (pseudo-) combining

            characters should be changed to <BLK> (possibly name-changed to <ACC>), in order

            to make the weighting system more transparent to a human reader. (NOTE THAT

            <MIN> should NOT be read as 'third level MINimum weight', but as 'third level weight

            for MINuscule (or caseless) letter/digit'; <BLK> (=<ACC>) is the minimum third level



SE 7.    US NB suggests to introduce a <MAX> weight; but since <MIN> is not minimum, the

            name <MAX> would be misleading, making the reader think that <MIN> is mimumum.

            Instead of <MAX> the name <LIG> is suggested.


SE 8.    (MAJOR) Insert as a new annex E "Order-preserving subkey reduction"; moving the

            bibliography to annex F.  The text is attached at the end of these comments, and can

            be obtained in electronic form from the Swedish NB. The kind of subkey reduction

            described in (new) annex E is inherent in the "position" directive, and this annex

            motivates why the "position" directive is sound.  Further, annex E describe "high-level"

            optimisations, not just implementation level optimisations.


SE 9.    (MAJOR) The clauses in 14651 should be rearranged as follows (i.e. insert this as an

            instruction to the editor, in the amendment text; as a basis for the next edition of 14651):


1. Scope

2. Conformance

3. Normative references

4. Definitions and symbols [from current 4 and 5]

5. Common Template Table: formation and interpretation

               [current 6.3, except 6.3.5 and 6.3.6]

6. Declaration of a delta [current 6.4]

7. String comparison [current 6.1 and 6.2]

8. Conditions for collation equivalence [from current 6.3.5 and 6.3.6]

Annex A - Common Template Table [and its name, current 6.5]

[and then the other annexes: B to F]


            The subclause divisions are intended to be kept, even though the subclauses are not

            listed above.


            This is a more logical exposition order of the contents of 14651, without making any

            normative change.


SE 10.  The note for I 3 should use <BASE> instead of <BLANK>.


SE 11.  (MAJOR) It has been suggested that the ordering after this amendment should not

            be changed, INCLUDING that added characters are collated (via the CTT) at the end,

            leaving better ordering of them to tailorings. Sweden does NOT support this LATTER

            part of stability, since the CTT should give a good default ordering for characters

            that are not of immediate concern to tailor.



Annex E...


September 27, 2001