Re: [tlug] [News] input methods for linux approaching standardization

["Stephen J. Turnbull" <stephen@example.com>]

Curt Sampson writes:

 > On 2008-01-28 11:03 +0900 (Mon), Gernot Hassenpflug wrote:
 > 
 > > http://spacehunt.info/2008/01/25/input-method-api-now-nearly-standardised-on-linux

Don't hold your breath, unless your purpose is to turn blue.

 > Is that standardized for "Linux" or for X11?

If they have any clue, it should be independent of X11.

If you read the blog post, you will see that Hideki Hiura is a three-
time offender.  If he gets extradited to California or Texas, he'll go
away for life.  Look at the process: after three years, with two major
preceding specifications with full implementations to reference, they
couldn't agree on a spec ... and so they wrote code?  Visions of Lucy
snatching away the football as Charlie Brown charges forward to kick
it fill my mind ... Aaaaaaaaaaaaaaaargh!  Here's what we find in the
reference implementation[1]:

    Revision 50: /trunk/imbus/docs

        * ..

Yes, Virginia, that *is* the *whole* *docs* *directory*.  Sheesh, we
never do learn, do we.[2] :(

Standardization of input method APIs is a 20-year standing joke, and
recent trends show ever less real interest in sharing the sandbox.
Both XIM and IIIMF were plausible standards (I've worked with them
both), both required a lot of discipline from the programmer, and for
that reason, open source implementations sucked badly because M17N
hackers have never had much discipline.  So many open source hackers
chose to ignore the standards for the last 10 years (basically, since
IIIMF was published).  I see no reason why things will be different
this time around -- input methods themselves are nontrivial, and
generalized input managers are downright hard -- unless the hackers
have acquired better habits ...  in which case it doesn't much matter
which standard they follow.

And this crap about "wait for publication, wait for the ISO", who are
we trying to kid?  If the standard is any good, the draft site should
being getting slashdotted and adaptions of the current flock of IMs
should be appearing already (cf ANSI C, C89, C99, HTML 2.0--3.0--4.0--
XHTML, etc, etc).

Footnotes: 
[1]  http://imbus.googlecode.com/svn/trunk/

[2]  For those who aren't in on the joke, if you don't know what
you're doing, writing code to do it anyway makes your ignorance far
more precise.  This is more than compensated as what you do know
becomes exuberantly inaccurate.