SC22/WG20 N791
L2/00-369

From: Kenneth Whistler [kenw@sybase.com]
Sent: Friday, October 06, 2000 8:43 PM
Subject: WG2 in Vouliagmeni (Athens)

Unicadetti,

As I am occasionally wont to do, I have decided to provide a
detailed report on the goings-on in Athens.

The first thing of note is that the meetings were not actually in
Athens, but outside of town in the rich seaside suburb of Vouliagmeni.
A couple miles to the southeast of us was the private harbor where
the super-rich park their Onassis-class yachts. But where we were was
nice enough -- right next to a beautiful Aegean beach, where on occasion
Asmus and Erkki could be spotted wading in their shorts. Or, if you
were up before dawn at 6:30, Arnold Winkler could be seen hiking
briskly down the seaside causeway, leading a laggard retinue of morning
dog-spotters. And there were 3 lovely Greek restaurants with covered 
pavilions right down next to the shore, for the lunch and evening dining
pleasure of those who didn't care to deal with taxi-drivers to head
downtown to eat the same things that could be had by crossing the street
from the hotel.

And if that is not enough to wonder why you missed it -- ELOT flew the
entire group to Santorini on the weekend and put us up overnight in
little inns in Kamari, the black beach resort town. So ELOT certainly gets
full marks for hospitality for its visitors to Greece! Since the
next WG2 meeting is hosted by the U.S. in Mountain View, this left the
U.S. delegation wondering whether a bus tour to the Museum of Tech
Innovation in downtown San Jose would be considered as not quite making
the grade in the way of planned outings for next spring's WG2 meeting. ;-)

Now down to the business at hand. The resolutions from the WG2 meeting
are available as document WG2 N2254 (= SC2 N3489). You can go get that
document for details, so I won't repeat everything here. However, there
are a number of issues that do have implications for the UTC business,
so I will focus on interpreting those actions.

====================================================================

1. 10646-2

The handling of the disposition of comments for 10646-2 was the main
agenda item for WG2. The summary of what took place is that the FCD
was successfully progressed for an FDIS ballot, with only minor changes
resulting from the ballot comments. The FDIS ballot should be issued
at the end of December -- the main reasons for it taking that long are
getting the revised, high quality font for the Plane 2 Han characters,
plus giving the IRG time to resolve North Korea's request to
include DPRK source information in the publication. In terms of synchronizing
with Unicode 3.1, the target completion dates are now for approved FDIS 2001-05
and (published) IS 2001-12.

Here are some of the notable changes for 10646-2 that resulted from
the ballot comments and the committee discussion:

a. Title change. Instead of the horrendous mouthful:

   Information technology -- Universal Multiple-Octet Coded Character
   Set (UCS) -- Part 2: Secondary Multilingual Plane for scripts and
   symbols, Supplementary Plane for CJK Ideographs, Special Purpose Plane

   all of the planes are being amended to be referred to as "supplementary
   planes". So now we will have simply:

   -- Part 1: Basic Multilingual Plane

   -- Part 2: Supplementary Planes

   And the text inside 10646-2 is being corrected appropriately. Chalk up
   one point for editorial sanity over committee standardese.

b. Old Italic

   All of the U.S. comments were accepted. In addition, the Irish and
   German comments resulted in the pulling of the 5 most problematical
   numerical characters. The UTC will need to reaffirm the result, since
   names changed, some characters were removed, and a few positions changed.

c. Byzantine Musical Symbols

   The character repertoire was unchanged, but the names were all
   changed to start with "BYZANTINE MUSICAL SYMBOL", omitting "GREEK" from
   the names. All of the parenthetical stripe comments were removed, too,
   and the discussion of these symbols was simplified in the annex, pending
   more definitive information about how these should be implemented.
   The UTC will need to reaffirm the name changes.

d. Western Musical Symbols

   WG2 accepted the reorganization of this block, based on the report
   of a U.S./Ireland ad hoc before WG2, including the input of the authors
   of the original proposal. The names were shorted to "MUSICAL SYMBOL ...",
   omitting "WESTERN" from the names. The UTC will need to reaffirm the
   entire block, since names and encodings changed, and there were a few
   drops and adds. The net effect, by the way, is a much more logical
   ordering than in the original proposal.

e. Mathematical Alphanumeric Symbols

   The U.S. comments to replace the open-face italic with
   boldface fraktur were adopted. "OPEN-FACE" was changed to "DOUBLE-STRUCK"
   in other names for consistency. Otherwise the entire set survived intact.
   The UTC will need to reaffirm the name changes, but there were no
   encoding changes otherwise.

f. CJK Compatibility Ideographs Supplement (Plane 2)

   Michel corrected the gaps for this set of CNS compatibility characters,
   but during the meeting an additional problem turned up, where a character
   whose unification was in question popped up in this compatibility set,
   instead of being dumped into the bucket for Vertical Extension C. That
   character will be removed from the FDIS. The UTC needs to reaffirm,
   because of encoding changes.

g. Vertical Extension B (Plane 2)

   China delivered to Michel what purports to be the final data files for
   Vertical Extension B, and pledged that from this point forward all they
   are working on is improvement of the font. There will be no additions,
   deletions, reencodings, or unification changes for anything in Vertical
   Extension B before the FDIS ballot. So I think we finally have a stable
   encoding. The UTC needs to reaffirm this set, which results from the
   application of the IRG editor's report on the last document.

h. Language Tag Characters (Plane 14)

   No changes.

====================================================================

2. Amendment 1 for 10646-1

The other major agenda item for the WG2 meeting was to roll all the
math symbols and miscellaneous other items from the WG2 "bucket", plus
various required text changes, into an omnibus amendment for Part 1.
This was accomplished, but not without a fair amount of dickering on
the details.

The good news for the math folks is that the entire set of math
symbols went in almost unscathed, and was the vehicle whereby we
carried all the rest of the additions.

However, there were a small number of changes which need to be dealt
with, as well as the additions that the UTC will need to take up.
And one major omission.

For those who have access to the WG2 document site, who want to
read along in detail, the document going in was WG2 N2263 "Working Draft
of Tables & Character Names for Proposed Amendment 1 to 10646-1:2000".
The document prepared during the meeting, incorporating all the additions
and changes from the meeting, as input to the editor, was WG2 N2281,
with the same name. The WG2 resolution M39.25 initiates the PDAM-1,
"Mathematical symbols and other characters", with the working draft
due at the end of October, and with the PDAM ballotting to be complete
by 2001-04. So we're on our way, folks, for Unicode 3.2!

a. U+0363 COMBINING GRAPHEME JOINER

WG2 declined to accept this character, which the UTC had approved.
The main objector was Ireland, on the grounds that the documentation
provided was woefully inadequate to explain the character and how
it might react with other characters like the ZWJ, etc. The U.S.
and the Unicode liaison argued for its inclusion, as directed, but
didn't have any good documentation of use to point to, so didn't
carry the day. The UTC needs to provide more convincing documentation
of the need for this character, presumably to accompany ballot comments
on the PDAM. But in any case the code point will have to change, since
WG2 put something else in that position.

b. Medieval superscript letters

In response to the German request for superscript (combining above)
letters, WG2 took the conservative approach of encoding just the 13
attested letters, rather than a full alphabet. These got added at
U+0363..U+036F, and the UTC will need to consider and approve these.
These will pose some interesting property assignment problems, since
they are *alphabetic* non-spacing combining marks.

c. Misc. Greek symbols

As might be expected in Greece, the Greek symbols included in the math
symbols proposal came up for intensive (and passionate) discussion.
To preserve the repertoire for the STIX set, we acquiesced in a number
of name changes and the moving of a couple characters.

The Q-SHAPED KOPPA characters were renamed to ARCHAIC KOPPA, after a
long discussion of alternatives. The Greek committee strongly objected
to "Q-SHAPED KOPPA", even though that seemed like an innocuous,
descriptive term to the rest of us.

The qualifier "WITH STRAIGHT BAR" was dropped from the name of
U+03F4 GREEK CAPITAL THETA SYMBOL.

The move involved the straight epsilon symbols:

U+213B GREEK SYMBOL STRAIGHT EPSILON 
--> U+03F5 GREEK LUNATE EPSILON SYMBOL
U+213C GREEK SYMBOL REVERSED STRAIGHT EPSILON
--> U+03F6 GREEK REVERSED LUNATE EPSILON SYMBOL

The insistence on "LUNATE" was a Greek preference. These are handwritten
variants of the epsilon, and are seen as exact parallels to the lunate
sigma. The mathematician's preference for "straight epsilon" will be
handled by a Unicode name alias.

Additionally, there were some further name modifications for all of
the other letterlike symbols from the math symbols proposal, though
all of them stayed at their proposed encoding positions.

The UTC will need to consider and approve all these modifications, since
the UTC had already approved the math symbols proposal in toto.

d. Komi Cyrillic

16 Komi Cyrillic characters were added at U+0500..U+050F. This required
establishing a Cyrillic Supplemental block at U+0500..U+053F -- which
was conveniently available there, courtesy of Joe Becker's guesstimate of
about how much bizarre Cyrillic was likely to trickle in eventually.
The UTC will need to consider and approve these.

e. (ZERO WIDTH) WORD JOINER

U+2060 WORD JOINER was approved by WG2, but without the "ZERO WIDTH" in its
name. This because no one on the committee, including the Unicoders, could
get the phrase "ZERO WIDTH WORD JOINER" out of their mouths consistently,
without mixing it up with "ZERO WIDTH JOINER". Rather than bequeath that
problem on the world, we went with the simpler and less confusing
"WORD JOINER". The UTC will need to approve this name change.

f. U+20B0 GERMAN PENNY SYMBOL --> GERMAN PENNY SIGN

A simple name change that the UTC needs to approve.

g. U+21F4 DOWNWARDS WHITE ARROW WITH CORNER LEFTWARDS
   --> U+23CE RETURN SYMBOL

This is the return symbol from JIS X 0213. The Japanese committee objected
to the name, and various people pointed out that this really is a keyboard
symbol and may either be represented as a filled or a hollow glyph, so the
description as "WHITE ARROW" was inappropriate. To resolve the impasse, we
renamed it and moved it into the misc. technical symbols block with the
other keyboard symbols. The UTC will need to consider and approve this change.

Incidentally, *all* of the rest of the STIX-derived math symbols survived
with no name or encoding changes -- which should be a boon to the people
working on MathML. So the front-loaded pain of name changes from the
Beijing meeting paid off in stability in Athens.

h. U+30A0 KATAKANA DOUBLE HYPHEN --> KATAKANA-HIRAGANA DOUBLE HYPHEN

The Japanese committee objected to adding "KATAKANA" to their suggested
name "DOUBLE HYPHEN", but when presented with the argument about the
confusion with the meaning of "double hyphen" in Western typography,
acquiesced in labelling it "KATAKANA-HIRAGANA", so as not to indicate
a constraint on its usage to Katakana only. The UTC will need to approve
this name change.

i. JIS X 0213 Compatibility ideographs

This batch of 57 (or is it 61?, or is it 60?) compatibility ideographs
is a botched mess. The draft I brought to the WG2 meeting in WG2 N2263
contained 57 characters, based on WG2 N2197 (from the Japan NB), "Update:
CJK Compatibility Ideograph request", minus the 4 duplicate radicals
that the U.S. and the UTC declined to endorse.

Japan countered in Athens with WG2 N2273, misleadingly labeled as
PDAM text, "AMENDMENT xx: JIS compatibility ideographs", which was
provided as "a correction to N2142", and which Japan insisted was the
correct input to the editor for the PDAM text. That document contains
60 [sic] characters, not the 61 previously advertised, including the
4 radicals in dispute. So it apparently is missing the one extra character
that was included at the end of the list in WG2 N2197. (A compatibility
form of U+9B2D, if anyone is still counting.)

At any rate, WG2 N2273 was designated to be used as input to the editor
for this block of characters. So the UTC action now will have to be
to draft comments on the PDAM text firmly requesting that the 4 duplicate
radicals be taken out again (and presumably a conditional request to add
back the one missing character from WG2 N2197, if it is really needed).

j. Circled Numbers

Given that the UTC couldn't come up with an usable alternative, and
given the Japan NB's insistence on getting an encoding answer, the
extra circled numbers from JIS X 0213 were just dumped into the PDAM.
This, by the way, also made the DPRK happy, since it covered part of
the mapping requirements for their standard as well.

The additions were:

In a gap in the Enclosed Alphanumerics block:

U+24EB..U+24F4 negative circled 11 -- 20
U+24F5..U+24FE double circled 1 -- 10

In two gaps in the Enclosed CJK Letters and Months block:

U+3251..U+325F circled 21 -- 35
U+32B1..U+32BF circled 36 -- 50

At least it was possible to accomodate these in the otherwise useless
holes in these enclosed symbols blocks, where like meets like. The
UTC will need to consider and approve these encodings.

Japan promised not to do this again. ;-)

k. Philippine Scripts

WG2 formally endorsed adding the 4 Philippine scripts into the PDAM.
These are Tagalog, Hanunóo, Buhid, and Tagbanwa, and they were just
plunked in with the names and encodings already approved by the UTC.
So there is no UTC action needed on these, except possibly a huzzah that
some more minority scripts have finally been progressed in 10646.

l. Recycling Symbols

WG2 approved the recycling symbols proposal, encoding the UNIVERSAL
RECYCLING SYMBOL at U+2672, and the 7 specific plastic type symbols
at U+2673..U+2679. The UTC will need to consider and approve these
encodings.

m. UCS Sequence Identifier

In addition to the character encodings for Amendment 1, there are a
number of important textual changes for 10646 that the UTC is interested 
in. The first of these is for the UCS Sequence Identifiers. WG2
endorsed the proposal that Uma wrote up, based on UTC input. The
only change is that the USI's will be known as UCS Sequence Identifiers,
rather than Unique Sequence Identifiers. The ambiguity of the term
"unique sequence" was the reason for this change. Otherwise, the
syntax and intent is exactly as the UTC proposed.

n. U+ Notation

WG2 completely endorsed the proposed change for the U+ short identifier
notation, as written up in WG2 N2234. So the PDAM will contain the
text which allows 5- and 6-digit U+ notation, as suggested by the UTC.

o. Permanent Reservation (in process codes)

WG2 also accepted permanent reservation of U+FDD0..U+FDEF for internal
processing purposes. So this also matches the UTC request.

p. WG2 acceptance of UTC proposed characters

WG2 accepted the addition of the terminal graphic characters and of
the THAANA LETTER NAA, which were already approved by the UTC. So that
puts WG2 in synch with the UTC on those characters.

====================================================================

3. DPRK Issues

The DPRK delegation (North Korea) was again present and active. Only
3 people showed up (instead of the 7 who came to the Beijing meeting),
but they had a full set of revised documents for consideration.

Once again DPRK asked for name changes and reordering of all the Korean
characters in 10646, and once again WG2 refused to do this. But this
time, the delegates from Germany and from Sweden, who have been active
in the WG20 work on ISO 14651 (International String Ordering) were able
to spend a fair amount of time explaining the alternatives for ordering
that make it unnecessary to insist on particular orders for encoding
in 10646 itself. Particularly helpful was Kent Karlsson bringing a
draft 14651-style collation tailoring showing what would be required
to get an ordering that would match the DPRK requirements. Several
people also spent time talking to the DPRK delegates about commercial 
and open source implementations of collation.

The DPRK also brought in a properly restructured request for additional
character encodings to match the content of their national standard.
We were able to work with them offline to further pare down and
refine that request, and we can expect them to come back with something
for the next WG2 meeting that will be close to ready for prime time
consideration, finally.

The major accomplishment, however, was one not actually on the formal
agenda of documents, but clearly on the DPRK's actual agenda. WG2
authorized a continuing ad hoc group on Korean issues, inviting the DPRK,
ROK, and any other interested parties to participate. Germany and
Sweden will be participating, and it would make sense for someone
from the U.S. more knowledgable about the Korean script than I am
to participate as well. What this really is is a backdoor channel to
work on IT technical issues that will result from the prospective
reunification of Korea. It is quite clear to everyone that once the
DMZ falls and the inevitable political and economic reunification
occurs, that the South Korean economy and technology will engulf the
North -- probably even more decisively than what has happened in
the West/East Germany reunification. But the DPRK technologists are
looking for answers for interoperability during the transition. Basically
they will just need conversion and collation solutions and tools, and
need channels open to find out about such things while the politicians
and diplomats are groping about for the bigger political solution.

So WG2 in this case was able to softpedal the confrontational
approach that we saw at the Beijing meeting. Instead of "These
proposals are ridiculous -- and we'll just say no", the approach
from Athens turned more towards "We understand you have a number of
technical problems in character encoding -- let's find a way to
continue the dialogue and figure out mutually satisfactory solutions
to the problems."

====================================================================

4. Editorial Corrigenda & Publication Issues for 10646-1

WG2 resolution M39.5 bundles together a bunch of editorial corrigenda
for glyphs, requests the Unicode Consortium to prepare the updated
tables, and instructs the editor to forward to corrigenda to ITTF
to publish as a Minor Revision to the standard. When that happens,
it will put the 10646-1 charts back in synch with the revised charts
that the UTC has posted for the Unicode Standard, reflecting all these
glyph charts.

Discussion of this raised the sore point for WG2 about the delay
that ITTF has caused in the publication of 10646-1:2000. As of
the Athens meeting, no one yet had a published copy of 10646-1:2000,
even though ITTF had the contents delivered to them in February. There
were some hot-tempered comments about this state of affairs. It is
certainly making ISO look bad, in comparison with the publication
track record of the Unicode Consortium. The whole issue was escalated
up to the SC2 plenary for discussion, where it resulted in some
careful editing of the text of the relevant resolution -- with
the SC2 chair trying for a polite, nonconfrontational resolution,
but with others pushing for a very strongly worded resolution
*demanding* that ITTF get off its butt, so to speak, and get the
thing published.

--Ken
	8