Rather, it defines that they do not. But since this is not true on any
reasonable intuitive definition of "interact typographically" (as we
have seen with Hebrew vowel points), this statement makes sense only as
a counterintuitive definition of "interact typographically".
Exactly. The words in the standard don't fit the facts, though one can read through the words and understand what is meant.
From 4.3:
<< For example, it is crucial that the combining class of the cedilla be lower than the combining class of the dot below, although their exact values of 202 and 220 are not important for implementation. >>
This is not explained, but obviously the reason why it is "crucial" that the combining class of cedilla be lower is that the cedilla indeed might interact with characters in other combining classes and so must belong to a combining class which is applied first so that other diacritics that might be applied beneath are positioned below the cedilla which cannot have any other diacritic between it and the base character.
The implication is that a font designer
cannot assume that glyph positioning adjustments are required only
between adjacent characters in canonical order, despite what the
"interact typographically" rule might suggest.
Therefore the standard should so indicate. Currently it really says the opposite, though that obviously isn't right and not what is intended.
Jim Allan
