Re: X Keyboard for Kurdish with Arabic script

[Oibane <pflm52td at w6 dot dion dot ne dot jp>]

Hi, Erdal Ronahi.

As Mohammed Sameer said, X doesn't allow strokes of codepoints assigned 
to one key touch, so you need workarounds.
One is obvious, to have the ZWNJ(Zero-width-non-joiner) on the 
keyboard, but this is clumsy for the Kurdish language, right?
(I'm Japanese, and I don't have the slightest knowledge of Kurdish.)

Another which came in my mind is an "input method". Input method
is an interface between the keyboard input and the X applications,
and translates strokes of key punches to letters, often by 
providing intermediate states. A simplest example is something
like a ' -> á.
Japanese writing system is complex, and we need input methods.
(The ordinary Japanese can write around 2000 characters or more.)
Also in Korea, China and Vietnam input methods seem to be used, as far
as I know.

Popular input methods, provided under the framework of scim or uim,
are rather big. But a dedicated input method for "ae" will be really
compact, and may help. It would just replace one "ae" key to 0647+200c.
It would be almost transparent, and users may forget its existence.

And yes, urging X people is the right way.


The other problem, which codepoint should be used, should also be
clarified, but please wait for the words of experts. There're quite a
few in this mailing list. I'm not the one, so I just remain
to review the facts. 

First the answer: it is (another!) flaw in Unicode, and the definite
answer cannot be given, at least for the time being. But because 
there already is the MS way, it may be to be followed.

Backgrounds:
U+06D5 Arabic letter AE has been there from the very beginning of the
Unicode version 1. From the ver 3.2, it is defined to be "teh marbuta"
class in its joining-group propery, and corrected to be 
rightward-joining. (It is strange that only few fonts support it, 
if it's such a veteran.)

What makes the story complicated is the U+200C Zero-width-non-joiner
(ZWNJ). It first appeared in the ver 4.0. It's rather new.
(To note, U+200D Zero-width-joiner ZWJ is from 3.2.0. They were 
not classmates.) The consistency of the policy lacks between 6d5
and 647+200c.

It is a problem, absolutelly, to have twin-in-visual. If they 
are not distingushable in their appearance, then confusions happen.

At least, the policy needs to be established. Just sitting back 
does not solve the problem.
a. If U+06D5 is to stay, then equivalence with 647+200c has to be defined.
It better be annotated to be "Kurdish" to reduce possible confusion,
although the annotation does not seem to be mandatory.
b. It can be deprecated.


Best regards,
"Oibane"
pflm52td at w6 dot dion dot ne dot jp
------------------------------
Arabic on Linux - written by a non-Arab for non-Arabs
http://www.k2.dion.ne.jp/~oibane/aonl/en/index.htm