[SCIM] SCIM and Wubi

Ming Hua minghua at rice.edu
Sat Jul 3 21:39:27 PDT 2004


James, I think you misunderstood Alexander.  From my understanding, he
has no problem setting his keyboard as Dvorak keyboard layout, and he
has no problem using SCIM with his dvorak keyboard either.  The point
is, it's very inconvenient.

Wubi put similar parts of a character as a group (if I remember
correctly, there are five groups, each group has five keys.  And the
five groups are horizontal strokes, vertical stokes, etc.), and the keys
for each group are put together on the keyboard (qwert, yuiop, asdfg,
etc.), so it would be easier to remember the positions.  However, with a
dvorak keyboard layout, the keys in the same group are spread out in the
keyboard without any regularity.

Since the parts represented by a Wubi key actually has nothing to do
with the actually english character on it, it makes more sense to keep
the keys in the same group together, regardless the actually english
keyboard layout.  This way, Wubi users remember the keys by position on
keyboard instead of the corresponding English character, and that's more
intuitive.

The problem is that in dvorak layout the characters and the punctuation
symbols somehow changes positions, you can see the actual layout at
    http://www.mwbrooks.com/dvorak/layout.html
for example.  To keep the same Wubi layout, we need to group the keys as
something like: ',.py fgcrl aoeui etc.  However in SCIM punctuation
symbols are treated differently than characters.

I see a straight forward way (it may still be hard, I haven't looked at
the code yet) to solve this would be to allow punctuation symbols in
scim-table IM engine, and then we can just generate a dvorak-wubi table
by some simple 1-to-1 substitution, and all are set.  But from a
long-term point of view, I agree with Alexander that allowing different
xkb layout for different input method may be the right way to go.


On Sun, Jul 04, 2004 at 09:05:09AM +0800, Zhe Su wrote:
> Hi,
>  I think the keyboard layout issue should be handled by XFree86 or
> linux kernel. Did you set your keyboard type correctly in XFree86
> configuration or linux configuration?
>  I know nothing about other linux distributions, but in SUSE Linux,
> there is a config file /etc/sysconfig/keyboard, in which the keyboard
> layout can be specified.
> 
>  Indeed, I know little about the keymap issue, does anybody know this
> issue? Please give us more detailed information.
> 
> Regards
> James Su
> 
> On Sat, 3 Jul 2004 01:40:01 -0700, Alexander Poquet <atp at csbd.org> wrote:
> > Hi folks.
> > 
> > I recently moved from XCIN over to SCIM because it really has become a
> > superior system.  Good work.
> > 
> > However, I have a problem with the Wubi input method.  Usually I like
> > to type with Wubi because it is fast.  When typing English, I use the
> > dvorak keyboard layout for the same reason.
> > 
> > The problem is that Wubi is a structural input method.  That is to say,
> > the letter which is on the key is not important; rather, its position
> > on the keyboard is important.  Those of you that type Wubi know that
> > it divides the keyboard into 5 zones based on stroke, and further
> > subdivides those zones based on the number of strokes.
> > 
> > The problem is that because I use dvorak and not qwerty, the key in
> > the 'q' position is not q, for example, it is an apostrophe (').
> > Essentially all the keys are in completely different places.
> > 
> > This makes Wubi very hard to type.
> > It also makes it less efficient, because before all similar strokes
> > were grouped together, but now they are scattered all over the
> > keyboard.
> > 
> > This was a problem in XCIN too.  What I did in XCIN was simply write
> > a perl script to remap all the keys to dvorak, so a sequence like 'cex'
> > would become 'j.q' (because the key in the c position is a j in dvorak,
> > etc).  This worked fine, although it was an ugly hack.
> > 
> > In SCIM, I cannot seem to do this.  Looking over the code for
> > scim-make-table, it appears that punctuation characters are not allowed
> > in the table.  This is a problem for dvorak keyboards.
> > 
> > Now, I am not much of an X programmer, but the most elegant way to cope
> > with this sort of problem is to somehow allow tables to bypass the
> > higher levels of XKB and deal with keycodes directly.  I do not know
> > how difficult this would be.
> > 
> > Admittedly, dvorak is used by only a minority of people, but many non-
> > qwerty layouts are in widespread use (AZERTY in France, for example).
> > 
> > This annoys me enough that I would be willing to lend coding time to
> > fix it.
> > 
> > As far as I am concerned, any structural layout (ie, non-phonetic)
> > should probably operate using key position rather than key value.  If
> > I understand properly, X already provides us with the layout of the
> > keyboard (physically) through XKB.
> > 
> > As a disclaimer, I am a C programmer.  C++ has never been my thing. :)
> > 
> > Alexander

Ming
2004.07.03

--------------------------------------------------------------------------
Physics is much too hard for physicists.                 --- David Hilbert
--------------------------------------------------------------------------



More information about the scim mailing list