L2/08-027

Date: 16 Jan 2008
From: Mark Davis
Subject: Default Ignorables

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

In the standard, the default ignorable code points are defined as those that should be invisible in rendering if unsupported. However, as a legacy of the development of the default ignorables (where for a time they also were meant to be ignored in general processing if unsupported), there are three types of default ignorable characters that should use a regular missing glyph, instead of what the definition states. (This came out in a long discussion on unicore.) These are:
 

• Controls (minus whitespace, which are not in DI)
   
• Surrogate code points
   
• Noncharacter code points
   

The presence of any of these code points indicates some problem with the text, and our recommendation should be to display missing glyph if unsupported, rather than be invisible. (As a matter of fact, the latter two are always and forever "unsupported".)

So, my proposal is that we remove these from the default ignorable code point property. That would reduce the possibility of erroneous implementations, make the property actually hew to its own definition, and make the situation much clearer to users of the standard. While we generally want to be conservative about property changes, for this particular property, I don't see any problems with making such a change, and do see much benefit.

--
Mark