Re: mule Chinese input inconveniences

[Dan Jacobson]

mailed & posted
>>>>> "K" == Kenichi Handa <handa@etl.go.jp> writes:

K> "Eli Zaretskii" <eliz@is.elta.co.il> writes:
>> I cannot even understand what is he talking about, let alone try to
>> see if this is fixed in Emacs 21 ;-)

K> I think I understand what is his complaint.

>> ------- Start of forwarded message -------
>> From: Dan Jacobson <jidanni@kimo.FiXcomTHiS.tw>
>> Subject: Re: mule Chinese input inconveniences
K> [...]
>> Small complaint about when I'm entering Chinese after using
>> toggle-input-method: OK, I've already got the character I like where I
>> want to put it.  Now say I want to move the cursor two characters
>> rightwards.  Well, can't use C-f, that is for selecting different
>> characters.  Can't use right arrow either, same problem.  Can't use
>> ESC, as that isn't independent at this point... Can use the finger
>> twisting C-\, which is bad when I still want to keep on entering more
>> chars... I just want to change the cursor position.  Can use SPC DEL
>> or RET DEL.  But that is sloppyish.  Perhaps "hands off C-f" "no fair
>> using up both C-f and right arrow and not leaving one left for
>> stopping character selection with cursor movement".

K> Such a feature is not supported, and it's quite doubtful
K> that the feature is surely useful.  Why on earth do you
K> start typing Chinese characters not at the place where you
K> want to insert them?

no,
Say I have a line
XXXX  XXXX  XXXX
and I want to put one chinese char in each of the two gaps.
I go to the first blank, select the correct char., then want to move
to the second blank with a few C-f ... 
I type

b a n g 4 C-f 

because i see the default choice is correct already
I dont need to select say the 3rd character with

b a n g 4 3 C-f 

If I use just the 4 version, and not the 4 3 version, I will still be
in character select mode, and the C-f will just rotate choices instead
of moving the cursor... which is very unnatural after so many years of
habit.

If C-f were liberated back to its former tradition role, there still
would be lots of ways to select characters at this point: arrow keys,
pick a number [3...] etc.   Other users wouldn't mind C-f's new role,
but would want the arrow keys left alone...

Anyways, the feeling is "OK, I've already got what I want on the
screen, now can I please move around?"  It seems that the easiest way
under the present scheme is to hit SPC Backspace, to break out of
character select mode... a wee bit easier to type than C-\.

>> Another problem is that ones eyes don't want to look up and dig around
>> in the text to see what character has been selected each time.  One
>> wants to look at the same spot in the minibuffer... however looking
>> there will not necessarily tell the typer what character will actually
>> be inserted... especially later in a session when the remembered
>> sequences kick in... I'd say put the selected char in the minibuffer too.

K> In such a case, perhaps what you want is such a mode that no
K> charactes is inserted automatically even if there's single
K> candidate for the key sequence you typed (e.g. the case of
K> "a4" and "ang3" for chiense-by-b5).  They are inserted only
K> when you explicitly select one by typing number or C-SPC, or
K> you type the key sequence for the next character.  Then, you
K> can always see what character will be inserted even if you
K> are looking at only the echo area.

K> But, the better way is to show the candidate list just above
K> or below the cursor line.  Then, your eye doesn't have to
K> go up and down.  I'm now working on such a method.  But, it want
K> be included in the comming Emacs 21.1.

I think that what xcin does now.  What I'm saying is when emacs starts
up, the default choice [1.] in the minibuffer is guaranteed to
be the character inserted at the cursor.  My eyes can check there with
out having to hunt around on the screen for where the cursor is upon
each character I type.  [I need to see the keycaps when I type because
I'm a bad typer, so it's one eye trip to the keyboard to screen for
each character.]

However, after a few minutes, emacs has 'learned' which characters I
like to type more, and then the #1. character of the choices in the
minibuffer is no longer necessarily the default character to be
inserted where the cursor is.  On must now leave that certain fixed
spot with the eyes for the varying spots where the cursor might be to
check...

Therefore it would be nice for the area in the minibuffer to also show
the default choice in addition to the 1. 2. 3...choices.

well, ok maybe like xcin a big blue floating blob wit the choices
would be easy also to find with the eyes...


>> Currently for longer [than 20 character] entry jobs, I do
>> $ emacs -nw file
>> and use xcin which knows multi character entry... yes, I know in the
>> future emacs can use xcin...   by the way emacs' v. v, v! etc. seems
>> smarter than xcin's alt-shift-o, alt-shift-, and I still haven't
>> figured out how to type big5 "!" in xcin.

K> As far as I know, xcin also works as XIM.  If so, the
K> comming Emacs 21 will be able to use xcin directly on X.

K> ---
K> Ken'ichi HANDA
K> handa@etl.go.jp

K> PS.  By the way, could you please stop using this kind of
K> incorrect mail address, <jidanni@kimo.FiXcomTHiS.tw>?   It's
K> so annoying to delete "Fix" and "THiS" everytime.

The kimo machine is closing down so i must soon come up with a new
stupid spam strategy.  How about the "deadspam.com" method?

Gee, you GNU Professionals don't seem to be bothered by spam, I have
to download it on my 15Kb rural phone line, then use procmail...

Will I not really get spammed if I have a true reply-to?
-- 
http://www.geocities.com/jidanni Tel+886-4-25854780 e-mail:restore .com.