Re: Is X lam-alef key OK?; technical detail

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

Let me summarize:

1. It is more than common among Arabs to use X keyboard Arabic layout.
The 'b' and other lam-alef keys (shifted t,g,b) are still a bit
problematic. At least some, perhaps many don't use this key.

I wanted to know if I could recommend the layout to the Japanese 
at my site.

2. (I guess) If pango is used correctly, even if the key input falls into
the range of Arabic presentation form B, it is displayed correctly.
It does not mean that the letter is saved in a file as lam U+0644 and 
alef U+0627.

It remains a guess, because I don't know gtk+2 and other related
libraries.

3. In katoob it is ensured that the decomposiotion mentioned above
is always done. I don't know exactly when it is. 
(It is done in the function katoob_document_handle_lam_alef, used as
 a callback in the file katoobdocument.c)


Still ambiguous points:
> pflm52td at w6 dot dion dot ne dot jp (me, Oibane) writes:
>> Lam-alef, bound to key "b" belongs to the Unicode presentation block B.
>> Imagine, it is like a piece of metal type of "isolated lam-alef", 
>> picked up as an element of printing press.
>> Components of this category are to appear on your PC screen, or
>> to be printed on paper.
>
Thamer Mahmoud wrote:
> I've seen this before on some programs. It became less of a trouble
> after switching to a UTF-8 locale (i.e. being able to handle
> inputing/saving characters in the presentational forms range).
>
> Also while using applications that are unaware of this problem,
> it is very much reproduceable when using non UTF-8 encodings. For
> instance, when saving files that includes a keyboard inserted
> presentational lam-alef in Windows-1256 files.

Nadim Shaikli says:
> (Snip) So if you
> are asking if the 'b' should in fact send 2 characters and be
> stored on disk/media in that manner then the answer is yes.

  I'm not getting the point clearly yet.
Does Thamer mean that the reason softwares don't complain is that ligated
lam-alif U+FEFB is uh, say, possible, if utf-8 is chosen, as opposed
to cp-1256 ? Then it remains used.
  Nadim states the way it should be. OK. To be strict to Unicode standard, 
saving a presentation form is not preferable (no, not allowed). 
But how about the reality, especially in saving files?

  One apparent con is web-search failure. I had a small test:
"Invalid" as-salaam hit only 79 pages, but the normal one hit 694,000
on google. So almost everyone seems to avoid the "bad" lam-alef.
(As for input, I remember when I ran firefox with locale "C", mouse 
drag & drop of french letters was impossible.)

By the way, to have Arabic and Japanese at the same time Unicode
is a must, so we don't have any other choice. :)


Others. Still open:
1.
> > X Window System does not allow a sequence of symbols to be bound to
> > a key. Only one symbol is possible. The architecture is now found
> > unsatisfactory, but it can not be changed easily. It would be a big, big
> > change.
>
> Is there a bug or a discussion about this somewhere?

Anyway, fixing X is the correct answer, unless it is decided to abandon
current layout.
I had a browse on new feature page of Xorg x11 R6.9/7.0 and R7.1.
There's no description about it, so it won't come in the nearest future.
It does not seem to be on the todo agenda either. Finally, I've searched
the word "bind" in Xorg ML to find no related topic.

I present you a possible helper here.
X keyboard architecture is criticized in libxklavier home:
http://www.freedesktop.org/Software/LibXklavier
The explicit address of the criticism paper is:
http://xlibs.freedesktop.org/xkbdesc/antixkb.pdf
The author Sergey Udaltsov has an acquaintance with Ivan Pascal,
one of the author of XKB extension manuals.
So contacting Sergey, who looks influential, may help,
adding our point to his.
His email address is given in that short paper.

More directly, X org mailing list is here:
http://wiki.x.org/wiki/XorgMailingLists

Don't ask me to do this job. It would be far less persuasive if 
requested by an Arabic-speaking (much closer to non-speaking) Japanese.

2.
Lam-alef selection problem of firefox, referred by Thamer Mahmoud:
https://bugzilla.mozilla.org/show_bug.cgi?id=271057
still survives on my firefox-1.0.6. I use Gentoo Linux pre-build version,
which seems to enable pango at configuration time.
(Well, Thamer says in bugzilla that it won't happen if pango is used...)

0.
> I completely ignore the 'b' key, but I'm not sure how
> significant that would be for Arabic touch typing. I also find other
> solutions like scim to be truly an overkill.

In Japan, Korea, China, and seemingly in Vietnam, input method is so
ordinary, but I know what you mean. Most key inputs are direct enough 
in Arabic.
You know, when I first saw the Arabic layout, I could not understand why
there exist lam-alif key. Lam and alef are already there!


Thank you all for your words.
Good day.

"Oibane"
pflm52td at wsitta.dion.ne.jp
# sitta is 6