tlug: was email software --> UNICODE input

["Stephen J. Turnbull" <turnbull@example.com>]

>>>>> "Craig" == Olinsky, Craig <olinskyc@example.com> writes:

    Craig> 	What would constitute a "Unicode Input Method" -- as
    Craig> opposed to a collection of separate input methods by
    Craig> language/script?

I'm not really sure.  :-)  Quail is a good start, although I don't
think the interface is unified enough.

    Craig> 	Obviously, you'd want the ability to do raw-hex input,
    Craig> although this wouldn't be as much value to endusers.  You
    Craig> could also put up a char-chart (by language/script/or code
    Craig> point) and have the user click, but that wouldn't be an
    Craig> efficient way to type either.  Handwriting recognition
    Craig> would be an interesting (but difficult) possibility.

The first two are not really input methods, in the sense that they're
far too inefficient.  The third is not as difficult as you might
think.  Try the URL http://www.mit.edu/~cadet (I think it is) and look 
for the package `strokes.el'.  But it's not going to beat Canna,
either.

    Craig> 	I supposed the only possibility of "unifying" input
    Craig> would have to be by the physical appearance of the glpyh --

That's one approach, using the 4-corners aproach (this is already
coded into Quail, I believe).

    Craig> since anything pronunciation-based would be
    Craig> language-specific, and require a certain language base for

No.  The first thing would be a method that recognized from context
that you aren't speaking Japanese anymore.  Or an easier way to switch 
from one to another than the current LEIM (which isn't all that bad,
but...).  That's why I called `Quail' the closest thing to true
multilingualization.

And you could `pun'.  Type Japanese and tell the wapuro that what
you're really looking for is Chinese output.  Weird?  Sure---but we're
talking truly multilingual usage here.

    Craig> every language used.  Plus there's a number of
    Craig> non-pronounceable characters.  Of course, there are many
    Craig> glyphs with alternate forms, and others with similiar
    Craig> appearances, so there couldn't be anything close to a
    Craig> one-to-one correspondance.

Finally, there's also the tcode approach (arbitrary keystroke
sequences for each character).  It's not as bizarre as it sounds; it
evidently really works, although the learning curve is pretty shallow.
But it's quite efficient (fewr keystrokes than Wnn4 or Canna 3).
---------------------------------------------------------------
Next TLUG Meeting: 11 April Sat, Tokyo Station Yaesu gate 12:30
Featuring Tague Griffith of Netscape i18n talking on source code
---------------------------------------------------------------
a word from the sponsor:
TWICS - Japan's First Public-Access Internet System
www.twics.com  info@example.com  Tel:03-3351-5977  Fax:03-3353-6096