[SCIM] The new features I'm implementing.

[Kenichi Handa]

In article <278a3d0411060720b637948 at mail.gmail.com>, Zhe Su <james.su at gmail.com> writes:

>   Currently I'm implementing the following features in the next major
> SCIM release:
[...]
> 5. Keyboard Layout remapping, eg. remap QWERTZ, Dovrak layouts to QWERTY. (TODO)

There are two kinds of input methods:
(1) based on a character generated by typing a key
(2) based on a location of a typed key

For (1), such a remapping should not be done.  For instance,
m17n-latn-post generates a-grave by typing the sequence
"a`", and that should be done regardless of the location of
keys "a" and "`".  Most input methods based on
roman-transliteration are in this category.

For (2), such a remapping is mandatory.  For instance,
m17n-sk-kdb (simulating Slovak keyboard layout) generates
o-circumflex by typing ";" on normal querty keyboard, and
that should be done even if one type a key at the same
location on a keyboard that doesn't generate ";" by that
key.  Most input methods simulating a localized keyboard are
in this category.  All m17n-based input methods in this
category currently assumes VT100 keyboard layout.

To distinguish them, each input method has to provide
information about the necessity of keyboard remapping.

I was going to provide the facility of keyboard remapping in
the m17n-lib itself, but if SCIM is going to support it,
I'll make a new API something like below:

MSymbol minput_keyboard_type (MInputMethod *im);

It returns Mnil when IM doesn't require keyboard remapping.
Otherwise, it returns a symbol representing a type of
keyboard IM assumes.

What do you think?

---
Ken'ichi HANDA
handa at m17n.org