Re: Insert Euro symbol

["Stephen J. Turnbull" <stephen@xemacs.org>]

>>>>> "Miles" == Miles Bader <miles@gnu.org> writes:

    Miles> "Eli Zaretskii" <eliz@is.elta.co.il> writes:
    >> > It's very convenient for people who don't use a latin-x input
    >> > method normally but want to only insert a character or two.
    >> 
    >> Is that because you have some other input method active?

    Miles> Yes.  I normally use the japanese language environment,

etc.

FWIW I agree with Miles 100%.  LEIM is broken by design in the same
way that XIM is broken, that is, it assumes that input methods are
modal.

IMO, input methods should not be modal, they should be context-
sensitive.[1]  Ie, if your LANG=ja_JP.eucJP and you visit a file which
is known to be in Polish (Latin-2), you should be able to start typing
and get a Latin-2 input method.  More, if you're a Buddhist scholar
with a trilingual (Sanskrit, Chinese, Japanese) document and you
overstrike a Japanese character, you should be in a Japanese input
method no matter what your most recent previous insertion used.  If
for some reason you really want to overstrike with Sanskrit, then you
explicitly invoke an input method, through a mechanism like C-x 8.

This would be really hard to implement, of course (without
telepathy.el which I hear is still totally non-functional :-).  Since
most people and most documents are monolingual, a modal input
framework is a close approximation to the context-sensitive framework
(in number-of-keystroke terms, not in annoyance terms, YMMV).  But
C-x 8 is a really important facility for making this bearable to
multilingual users.

Note that most fullfledged Japanese input methods provide both modal
aspects (is the basic input romanized phonetics or Japanese
phonetics?) but also C-x 8-like "single shifts" to menus of symbols,
radical/stroke count input, direct input of JIS codes, and even more
exotic methods.  These are only used for "rare" characters that may
not be in the IM's dictionary or that the user can't pronounce.

By analogy, I would suggest that C-x 8 could be considered a
single-shift to a designated secondary input method.  The default
should be "traditional C-x 8", but it should be reconfigurable to a
"single shift" to any LEIM IM.  (This probably requires extensions to
LEIM.)  C-x 9 could give a menu of available input methods (ordered by
user preference), then you enter one char using that IM.  C-u versions
of C-x 8 and C-x 9 could allow changing the defaults (C-u C-x 9
therefore probably should invoke a Customize method).

This is quite complex, but the typical monolingual "The sysadmin set
LANG in /etc/profile and I've never noticed" user still won't notice.
Nor will Latin-2 users who need to enter the Euro sign (once U+20AC
gets added to C-x 8 :-).  Buddhist scholars and other polyglots will
get a lot of benefit from it, I think, although they really want a
'input-method text property.


Footnotes: 
[1]  I don't know of any context-sensitive implementations, except in
edict.el, and that's broken at the moment.  AFAIK Emacs is the only
app with any hope of getting it right, anyway.

-- 
Institute of Policy and Planning Sciences     http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
 My nostalgia for Icon makes me forget about any of the bad things.  I don't
have much nostalgia for Perl, so its faults I remember.  Scott Gilbert c.l.py