L2/04-327

Source: John Hudson
Date: 2004-08-03 18:17:24 -0700
Subject: Distinction of Vav Haluma and Holam Male

Here goes, as succinctly as possible:

I consider the visual distinction of vav haluma and holam male to be an  
important one, which is clearly desirable in a great many documents,  
especially those of a scholarly and educational nature. Some contibutors to  
discussion on the Unicode Hebrew list have made the point that the  
distinction is optional, a feature of 'more exact' typography. I have no  
trouble accepting this, so long as the implication is understood: that the  
choice is that of the document creator, i.e. it is an authorial or  
editorial choice. For this reason, it is important that the distinction be  
reliably made when desired by the document creator. It is not acceptable  
that the choice be left to the rendering system. The distinction is  
optional insofar as the document creator desires to make the distinction,  
not insofar as a rendering system may ignore the document creator's intent.  
I believe that some contributors, including Jony Rosenne, favour an  
encoding that is ignorable, and hence unreliable. Since the visual  
distinction signifies a semantic distinction, I find an ignorable solution  
to be unacceptable.

Further, I have concerns about the reliability of an encoding that uses  
the control character ZWNJ (document L2/04-307) in existing rendering  
systems, which may legitimately choose whether to paint control characters  
at any stage during rendering. At best, it seems that the font mechanisms  
available to render the distinction between <vav, holam> and <vav, ZWNJ,  
holam> at the glyph level are severely limited. It seems to me that, if  
control characters are not painted, and the glyph-level distinction has to  
be handled through application of specific layout features, the available  
mechanisms for rendering the distinction are very limited. A restriction  
such as always treating one sequence as a ligature and the other as not  
seems to me an inappropriate limitation on rendering technology to be made  
in a character encoding standard. Since I do not claim exhaustive  
understanding of all aspects of rendering, my hope in raising these  
concerns is that the proposals for distinguishing vav haluma and holam male  
will be closely reviewed by UTC members who are experienced in the design  
and coding of rendering engines, particularly those who have varying  
philosophies on the painting of control characters. I have spoken with  
Peter Constable at Microsoft about this, and also hope that Eric Muller at  
Adobe will involve himself.

If the proposal to use ZWNJ is judged to be unreliable as a means to  
distinguish vav haluma and holam male when desired by a document creator,  
then I believe the UTC should probably approve the counter-proposal to  
encode a new holam haser for vav (document L2/04-310). I have asked Jony  
Rosenne to describe the interchange problems that he believes will result  
from this proposal, since I have not managed to imagine what they might be.  
I see no reason why existing documents should be affected at all, and I  
believe that a sensible fallback mechanism in Hebrew rendering engines --  
to handle acceptable display of new documents with existing fonts that do  
not support the new character -- would be easy to implement because of the  
acceptability of not making the distinction in 'less exact typography'. If  
the new character is not supported in a given font, the rendering engine  
could use the existing holam character for display purposes.


Regards, John


--

Tiro Typeworks
www.tiro.com Vancouver, BC
tiro@tiro.com