Awhile back, we complained about the cursor movement in XXE when the
cursor is inside Arabic script (or, I presume, any right-to-left
script). The problem was that the visual position of the cursor did not
correspond to the insertion point. (It appears to us that the insertion
point behaves correctly when it passes from left-to-right script into a
stretch of right-to-left script, namely it jumps to the right-hand end
of the right-to-left stretch. The visual representation of that
insertion point, i.e. the visible cursor, is what's broken: as you press
the right arrow key, it moves left-to-right through the right-to-left
script.) Your response was that this is a Java issue, and therefore not
something you could easily fix.
(BTW, I don't want to give the wrong impression by using the word
"complain": XXE support is wonderful, and I realize that you have to set
priorities, and that there are probably few people out there who want to
use Arabic script in XXE.)
We've discovered another issue with Arabic script: selecting a word (a
bit tricky, because of the above cursor problem, but still possible) and
marking it for emphasis (i.e. wrapping it with the <emphasis
role="bold"> tag) results in switching the order of the words before and
after on the screen.
For example, if the following were Arabic text:
word3 word2 word1
then tagging word2 would result in the following on-screen display:
word1 _word2_ word3
(where I have represented the on-screen bold emphasis by leading and
trailing underscores; I don't think most Arabic fonts support italicization)
I presume the problem is that what was once a single stretch of
right-to-left text, and therefore laid out on-screen that way, has now
been broken into three spans, each of which is laid out right to left,
but the three spans are concatenated in a right-to-left fashion.
Is there any chance of fixing this?
--
Mike Maxwell
What good is a universe without somebody around to look at it?
--Robert Dicke, Princeton physicist
