Re: Industry Thai Cell-Clustering Rules

[Theppitak Karoonboonayanan <thep links nectec or th>]

On Mon, Nov 06, 2000 at 10:33:40AM -0800, Chookij Vanatham wrote:
> 
> So, let's make the change of Thai Pango engine to have Wtt2.0 cell-clustering
> and also, adding Mac/Microsoft shaping (tis620-1, tis620-2).
> 
> I'm changing/updating Thai pango engine now.
> Attached are the Wtt2.0 cell-clustering rules.

OK. That's also a good guide for me to Pango code. Thank you.

I would like to discuss some points before going on..

As you once stated about the cursor movements and editing behaviors within
Thai text :-
- It is natural to move cursor by one cell instead of by one character.
- Text insertion is only allowed at the cell boundary.
- Text deletion with Del key should delete the whole cell under the cursor,
  not only the baseline character. This can prevent confusion when the
  upper/lower vowels from the right cell jump to combine with the left cell.
- However, text deletion with backspace can be done on character basis,
  because there could be no confusion in that case.

Therefore, the meaning of "cell" here is quite sensitive for the special
case of SARA AM, which occupies two cells.

Moreover, the consistence between the cell counting and the physical
appearance is quite important in some matrix-based display, like those in
web browser's <TEXTAREA> and in terminal emulators.

So, I think the cell retrieval routine you have provided could be used in
other areas as well, and it could be shared for consistency. But how could
we make its meaning more firm?

I think the twisting of the cell meaning comes from the need to render
SARA AM in two cells. So, how about letting two consecutive cells combine
in shaping stage?

If you agree, I will design the shaping mechanism based on this concept.

Regards,
-Theppitak.