Elazar Leibovich wrote:
> Maybe it's a problem with the code flow. Actually, the whole idea of
"patching" the cursor position in RtL paragraph seems wrong to me. One
should IMHO, to calculate the position it should've been if everything
was English, and then to reverse it (subtract the window width with
it, etc.).
I'm not sure I agree with that, because we want to be able to deal with
*mixed* RTL and LTR (Bidi...). I'm not sure that just mirroring
everything would work in this case.
The fact is, LyX has traditionally had very good support of bidi, and
it's been breaking only because there hasn't been anyone to yell when a
change which affected bidi behavior occurred. We need to patch these
things up, and add additional incremental improvements along the way.
For now, I think that your proposed selection is the correct approach,
we just need to work out the details. But let's focus on that until
proved otherwise...
Dov
