On Wednesday, August 6, 2003, at 3:53 PM, Peter Kirk wrote:
This answer presupposes that there is a well-defined concept of which base character a combining mark belongs to. That is not always true. The particukar combining mark which precipitated the debate may be situated above the gap between the (logically and phonetically) preceding and following characters, or may move on to the preceding or the following characters depending on the precise context and on the typographer's preference.
If its behavior is substantially different from that of existing combining marks, then there's no reason not to suggest it be added with its own properties. Just don't add it as a combining mark. (This is basically what happened, e.g., with the ideographic description characters.)
Good idea in principle. The trouble is, it already is a combining mark and presumably has to remain so, at least it has to retain its combining class! Is there any way we can adjust the properties without encoding a new character?
Ken suggested looking at Tengwar as a tutorial case to refine Unicode principles. Well, Hebrew can also be a tutorial case to refine Unicode principles. Since Tengwar, and some aspects of Indic scripts, don't fit nicely into the conflicting principles, we shouldn't try to squeeze Hebrew into this straitjacket as a matter of principle.
Anyway, John J, what code are we talking about that has to work from the positions of the combining marks back to the underlying representation? Are you talking about OCR?
No, the issue is more how to start from a base form and work forward to encompass the whole series of characters which need to be treated "as one" in certain processes, which can include cursor movement, hit testing, display, line breaking, collation, normalization.
Thanks for the clarification. I certainly see the issues with cursor movement and hit testing. Normalisation is unfortunately so fixed that it is not an issue.
-- Peter Kirk [EMAIL PROTECTED] (personal) [EMAIL PROTECTED] (work) http://www.qaya.org/
