Re: yet more term.el fixes #2

[Miles Bader <miles@lsi.nec.co.jp>]

Dan Nicolaescu <dann@ics.uci.edu> writes:
> Looking closer, the characters > 158 decimal have char-width 4 in my
> setup. 
>
> Use this to print all the widths: 

Hmmm, using that code I get simular (odd) results:  the result of
`char-width' seems to be completely unrelated to the actual displayed
width of the characters -- some values are displayed as a single column
character, but have a `char-width' value of 4, and some are displayed as
a 4-column escape-sequence, but have a `char-width' value of 1!!

Here's a version that uses `decode-coding-string' to get correct
results; probably term.el should be using something like that:

   (defun print-width ()
     (interactive)
     (let ((i 0))
       (while (< i 256)
         (let ((decoded (decode-coding-string (string i) 'iso-8859-1)))
           (insert (format "%d %d %s\n" i (string-width decoded) decoded)))
         (setq i (1+ i)))))

Using this, all the char-width values match the displayed representation.

When simply inserting raw character values in to the buffer, it seems
that some sort of default decoding takes place at insert/display time,
but an equivalent default decoding is not done by `char-width'; whether
that represents a bug, and if so, where, I don't know, but I think it's
probably always safer to be explicit as in the above code.

-Miles
-- 
The secret to creativity is knowing how to hide your sources.
  --Albert Einstein