Re: Displaying Plane 1 characters (was Re: TV teletext)

From: Rick McGowan (rmcgowan@apple.com)
Date: Wed Oct 14 1998 - 20:15:35 EDT


This is off the Plane 1 topic, but on the terminal emulation topic,
actually... Frnak asked, regarding the Mac OS X Server lack of bidi...

> No doubt due to bidi issues?

Precisely. Not that it wasn't *possible* it just wasn't a marketing or
customer requirement at the time.

> don't have much experience with Arabic, but as to Hebrew, I can say that there
> has been plenty of online computing in Hebrew, going back many years and still
> going strong today, that is based on ISO 2022 and VT100/220/320 terminals
...
> As to Arabic, I'm sure it could work the same way, but from my limited
> understanding of the situation, there are rendering requirements beyond those
> of Hebrew that make display of Arabic (and probably Farsi, etc) in a
> rectangular grid of screen cells unacceptable to most readers, even though
> PC code pages have been available for use in DOS for many years.

Actually... terminal emulation doesn't require fixed-width cells at all. It
just requires cellular addressability. I proved this thesis to my own
satisfaction (in a previous incarnation prior to even the dawn of Unicode).
I once wrote a terminal emulator for English & Arabic using ASMO-449 as the
codeset (and shipped it to China, but that's another story...). This
emulator handled the Arabic shaping rules, and had variable width "columns"
and correct cursor positioning in a variable-width physical space.
Conceptually, the display was a grid of cells; the cells' width and physical
coordinates (in pixels) varied depending on the characters they displayed.
We had both "vi" and "emacs" running, doing mixed Arabic & English processing
together, without ANY modification or bidi knowledge on the part of the text
editors. The emulator handled the conceptual to real space mappings &
cursor positioning, including bidi -- and it all just worked.

Unicode processing of Arabic, however, is a rather different game because of
the non-spacing marks, ligatures, etc, etc. I think it's less amenable to
the fixed-cell model because the on-screen "cell" could contain potentially
many "characters".

        Rick



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:42 EDT