From: 	MIKE_KSAR@hp-paloalto-om4.om.hp.com[SMTP:MIKE_KSAR@hp-paloalto-om4.om.hp.com]
Sent: 	18 September, 1997 10:50
To: 	CARROLL_DON/HP-Boise_om8@boi167.boi.hp.com; kenw@sybase.com
Cc: 	Hart, Edwin F.; x3l2@listproc.hcf.jhu.edu
Subject: 	Re: draft SHARE comments on CG model

     
Ed,

I tend to agree with Ken's feedback especially in his statement that says the 
issue is not that a character code, in this case Arabic, can map to 4 possibly 
different glyphs (which happen to be coded in 10646 in the Presentation Forms 
part of the RU Zone).  It is important not to confuse the reader, especially 
ones who could claim that a character code, again for Arabic, can be represented
by a ligature, which is "contains" several character codes rendered as one 
glyph.

Your proposed changes to ensure that the document is uniformly formatted are 
also important.  I would suggest that you should get some input from Keith 
Brannon of ISO Central Secretariat (ITTF) on the camera-ready copy that you 
intend to provide before you submit your final document for publication.  

Mike

______________________________ Reply Separator _________________________________
 
Subject: Re: draft SHARE comments on CG model
Author:  Non-HP-kenw (kenw@sybase.com) at HP-PaloAlto,shargw3
Date:    97/09/16 2:58 PM


     
Ed,
     
I've looked over your comments, and have the following comments on the 
comments (metacomments?)...
     
Re comment 3:
     
I think the description of M-to-N sounds correct to me as it stands. The 
intent is to cover the following cases:
     
   1 character maps to 0 glyphs (e.g. spaces of various sorts, etc.)
   1 character maps to 1 glyph  (e.g. most Latin letters, Han ideographs) 
   1 character maps to 2 or more glyphs (e.g. compatibility Roman numerals
                        decomposed and rendered with individual glyphs)
   2 or more characters map to 0 glyphs (e.g. sequence of format codes)
   2 or more characters map to 1 glyph  (e.g. combining sequence rendered
                        with preformed glyph; Hangul jamo sequence rendered 
                        with Hangul syllable font, etc.)
   2 or more characters map to 2 or more glyphs (e.g. typical case for
                        Indic script rendered with high-quality font with 
                        conjunct glyphs, where rendering logic may map on 
                        a syllabic basis)
     
The issue here is *not* the number of positional variants of a glyph which 
may be associated with a single character. I don't think that issue
should be raised here, as it may tend to confuse the basic relationship 
described above.
     
Re comment 8:
     
In your suggested Table 1, I would change "Coded Font" to "Character-coded Font"
. 
All fonts are "coded" in the sense that they use code indices to access the 
glyphs they contain. The big difference for the "Character-coded Font" is 
that the index code to the glyphs is identical to the character code, rather 
than being logically separated from it as a glyph identifier.
     
--Ken