RE: tlug: Kinput2

[TBaetzle@example.com]

Disclaimer: I don't know what I'm talking about ;-)

> Jonathan Byrne wrote:
> Is Kinput2 a global FEP, or does it only work on things that are run
> from
> the same Kterm window in which Kinput2 was started?
> 
It's global to the local display, and it works even if you don't 
start it in a KTerm. Just make sure you start it _before_ you
start any KTerms that you wish to use it - AFAIK KTerm
does not query dynamically wether a Kinput has become
available if it didn't find it during startup.

> Next, a "how does it work" question: on MacOS, Windows 95/NT, and
> presumably
> OS/2 (haven't tried OS/2 J), switching to Japanese input mode will
> attempt
> to put Japanese text into any application, whether it's double-byte
> enabled
> or not.
> 
I don't know about Macs, but on Win and OS/2 (and of
course AmigaOS) all applications are event-driven. The
keyboard and mouse drivers create events when you hit
a key or move the mouse. These events are then passed
down a hierarchy of so-called event handlers which add,
modify or remove events. The mapping from keyboard
scan code to a character value accoring to the currently
enabled character set would be done by a handler.

What the FEPs do on these systems is add another 
event handler at a high priority that does the conversion.
The applications, which get the "input events" much
later in the chain of event handlers basically can't
tell what has been going on. The application really
has to know that when it gets a code sequence like 
0x82, 0xA9, 0x82, 0xC8 that it should display "kana" 
in Hiragana and not some weird combinations of
other characters.   

>   If the app with the focus isn't written to accept inline Japanese
> input, you get the little input box that pops up on your screen and
> you can
> type Japanese and choose kanji.  When you hit enter it goes into your
> document at cursor position, and sometimes you are lucky and it will
> work if
> you chose a Japanese font, under times it just won't and you get
> garbage.
> But the OS will always try and force it in there.
> 
And weird things do happen - you see n characters
in the display, but the cursor is really positioned at
n + m characters :-(

> What is the difference between that approach and the Kinput2 + Canna
> or Wnn
> approach, where it won't even try to force Japanese into an app that
> isn't
> written for it, such as Netscape Mail?  Can a force-it-in everywhere
> approach be done on Linux/UNIX, or does the way things work under UNIX
> make
> this not possible?
> 
The "Kinput" approach is that the application queries
Kinput for the input when it's available. You could 
probably create a "transparent" FEP in a Unix/X 
environment since this also uses an event mechanism.
However, I don' think that many applications would
be able to cope with it...

Thomas
---------------------------------------------------------------
Next TLUG Meeting: 11 April Sat, Tokyo Station Yaesu gate 12:30
Featuring Tague Griffith of Netscape i18n talking on source code
---------------------------------------------------------------
a word from the sponsor:
TWICS - Japan's First Public-Access Internet System
www.twics.com  info@example.com  Tel:03-3351-5977  Fax:03-3353-6096