The future of "Transitional": Japanese concern

[Rick Jelliffe]

(My background for having an opinion for this is that I used to work in 
the Japanese publishing industry in SGML, that I have produced a 
Chinese/English typesetting system for a Singapore client for the PRC 
trade laws system, that I was employed in Taiwan for several years to 
research in the area of CJK page structures that were not supported well 
by Western-sourced software, that I was a member of the CJK DOCP experts 
group on document processing, a former member of the W3C I18n SIG.  The 
Japanese and Koreans can speak for their own requirements, but I can 
speak for what I have seen and used in the region,  in support of them.)

BACKGROUND COMMENT

As a general background comment, I think we non-CJK-ers are not in a 
good position to judge how necessary some feature claimed to be required 
for CJK processing is. I think we need to *strongly* defer to the 
particular National Bodies in this area: if the CJK NBs say they need 
something, I recommend  NBs adopt a policy of voting "concur" on 
whatever technical outcomes arises, once strategic issues such as the 
relationship between S & T are resolved.

A group of European experts can reasonably be expected to be able to 
discuss whether so-called French Spacing should be supported as well as 
so-called English Spacing.  They will have looked at enough authentic 
documents to have an idea of how it looks, the best way to support it, 
which kinds of publishing uses it, whether anyone cares, whether it is 
already supported by other means, and so on.

But I think there are many CJK issues where,  though the technical 
issues be arcane but graspable, the non-CJK NBs are not in a strong 
position to make any opposing statements about how important or 
desirable a feature is.

Despite this, it is frequently the habit of us loud Westerners on 
standards bodies to take the attitude "Because you cannot prove you need 
it, we are not in a position to approve it."  In effect it means that 
only things that are also found in Western typesetting are allowed, or 
the grossest or most jarring features only (LTR, ideographs, etc). I 
think several of us have seen this over the years: Japan's plea at the 
BRM "please do not block us" is something that should not have needed to 
be said, in a perfect world.

Now of course, SC34 is highly internationalized, and i18n has everyone's 
goodwill and commitment, and the CJK NBs are in the driver's seat for 
the Secretariat and so on. So please I am not trying to cry wolf nor 
light fires. I don't think a white guy can play the race card by proxy! 


TECHNICAL COMMENT ON GRIDS

I would like to confirm that the W3C note on Japanese Text Layout at 
http://www.w3.org/TR/jlreq/  is very good and worth support, not just 
for Japanese typesetting, but for many kinds of Chinese typesetting too. 
Indeed, many of the typesetting rules for other counties in the region 
are regional adaptations of Japanese analysis. 

The W3C note might be a little clearer to non-Japanese if it substituted 
a term like "grid layout" rather than kihon-hanmen. 

But I think it is reasonable to say that the grid is a fundamental 
graphical principle for modern and historical CJK print (more obvious in 
some periods than others.) Japanese and Chinese learn to write in grids 
at school. This is not to say that the characters are always neatly in 
their boxes or in half-width boxes, or that kerning is not used, or that 
type is always solid, or that there are not other objects that can 
interrupt the grid. 

Art alert!
For an extreme example from calligraphy, see this:  
http://www.nigensha.co.jp/kokyu/en/c09.html
For a less overt example: http://www.nigensha.co.jp/kokyu/en/p06.html
For an example of flexibility within the grid, see this: 
http://www.nigensha.co.jp/kokyu/en/c21.html
For an example of the grid used in traditional culture, see fan 6: 
http://www.nigensha.co.jp/kokyu/en/c20.html
Or the vertical text at the bottom: 
http://media.photobucket.com/image/japanese%20magazine%20layout/trfjason/tamiya_ta051.jpg

Perhaps one way to explain one of the issues is to use the analogy of 
tables:  we can specify a table width in a few ways: one way is to 
specify a *measurement* like standard column width (and the table is as 
wide as that number times the number of columns);   and another way is 
to specify a *count* that there are x number of columns all the same 
size (and the column widths are the table width/x).   Most systems allow 
a combination of all these methods, because people apply different 
graphical ideas to different tables. However, for paragraphs and lines, 
Western typesetting systems support only a *measurement* approach while 
the CJK grid is much more like  a *count* based approach:  I want to fit 
twenty lines vertically each with 50 characters (in the design font and 
size)."

I agree with Murata-sensei's comment "I guess that it is possible to 
create OOXML extensions so that documents can contain specifications in 
terms of KIHONHAMEN *as well as* margins."  

But while the grid parameters can be converted for any one frozen page 
into combinations of margins, line spacing, tracking, kerning, etc,  
and  opened and displayed in a non-grid-aware OOXML system, I would 
expect that editing the text in the non-grid-aware system would 
eventually mess up this hardcoded formatting to some extent. Think of 
how horrible it is to enter a table width in percents, then move it over 
to an external system that has converted everything to inches! The 
kihon-hanmen approach seems a very simple way to specify important 
parameters of page, paragraph and line design with just a few 
parameters, for pages of CJK text.

I read that Office 10 is supporting more of the Open Font features: it 
would be useful to know what the impact would be on things like 
baselines, kerning and linespacing.

Features for CJK that have existing fallbacks using MCE have the big 
advantage that they can be added to OOXML (perhaps in a new part:  
"Features which should be used with fallbacks")  without any impact on 
implementers'  product cycles or other complicating deadlines.

I would contrast the grid layout with CJK tables in this regard. Full 
support of the kinds of complex divisions that one sees in CJK tables 
(stemming from the typically compact nature of headings with one or two 
letters)  would be quite difficult.   And  it would be really useful to 
mirror the changes being proposed for ODF-NG by the Chinese companies in 
this regard  (they go a little further than OOXML does with its diagonal 
cells, but not as far as I found would be necessary to handle the kinds 
of cell divisions seen in CJK printing.)  The Japanese Layout note does 
not attempt to deal with issues such as tables, but that does not mean 
complex cell divisions are not lurking in the shadows.

Cheers
Rick Jelliffe

P.S. I think the requirement to figure out policy and relationships 
between S & T is essential. Otherwise the WG will make changes ad hoc, 
which we then will be stuck with.