[MACTCL] Re: 8.4.6...

[Benjamin R. <Ben...@ep...>]

Hi Daniel,


"Daniel A. Steffen" <st...@ic...> writes:
> great! so this is fixed in the update?

Yes, but see the Problem that James noticed.  I'll see that I can fix
this in the evening. 

> I traced it down yesterady to ATSUBreakLine hanging in certain
> circumstances, but couldn't figure out what exactly was causing it
> to happen...

ATSUBreakLine runs into an endless loop, if the maximum space that it
is given is too small.  It's not clear yet, how much exactly "too
small" means.

> I was just referring (very imprecisely) to the fact that the
> comparison between the widget demo main window with ATSU vs without
> looks quite different, esp the line spacing, c.f. comparison
> 	http://rutherglen.ics.mq.edu.au/~steffen/tcltk/atsu/
> 	atsu_widget_demo.pdf
> (ATSU on the left). This may well be ok, I don't know how important
> it is to preserve existing text spacing

Line spacing is calulated differently between QD and ATSU.  QD shrinks
characters, until they fit into the line.  While ATSU by default just
draws outside the theoretical limits.  Because this causes problems
with common accented characters, I inserted code to expand the line
spacing. 

> Another issue I found is with your test file from the patch with
> arabic, hebrew & syriac: when clicking on the roman text that starts
> off one of the lines with non-roman text, part of the non-roman text
> disappears, c.f.
> 	http://rutherglen.ics.mq.edu.au/~steffen/tcltk/atsu/atsu_test_ok.pdf
>	http://rutherglen.ics.mq.edu.au/~steffen/tcltk/atsu/atsu_test_bad.pdf
> funny things also happen when trying to select anything on these
> lines, text tends to disappear and reappear etc... this may have
> something to do with bi-di?

Yes.  tkTextDisp.c gets confused with bidi text.  Similar things
happen on Windows.   I have ideas about this, but I haven't yet
written anything.  There is quite a lot that has to happen to make
bidi interaction work. 


Thanks for the valuable feedback. 


benny