> Asger, is the proposal of Wed, 18 Mar 1998 still valid?
> I still think ElementType class you proposed is the best for addressing
> this problem. By letting ElementType getNextElementType() to return
> "Word" if the cursor is on <CJK> or other east Asian character, and
> if the next element is not a certain punctuation, we will be able to
> have a working line breaking algorithm. UnicodeData will be of
> use here.
The principle is still valid. The implementaton will be different: Instead of
having one getNextElementType that will handle all kinds of elements, I think
we should introduce a separate iterator for each type.
So, we will have a "breakPoint" iterator that will give us the points in the
data structure where we can be a line break. This one will be able to support
the line breaks that are needed for your situation. And the flow logic will use
this one, so I think we have solved this problem for Asian languages.
Greets,
Asger
