[emacs-bidi] Re: [emacs-bidi]status? news?

[Eli Zaretskii]

On 18 Jan 2001, Matan Ninio wrote:

> I've looked for traces of this code in the pretest emacs21, but could
> not find any.  Do you know if this been integrated into the mainstream
> code?

That code is not in the current Emacs sources.  As Handa-san said, we are 
trying to find a way of performing the logical to visual transformation 
without prefetching text into some buffer and converting it all at once, 
which is how a typical bidi implementation works (and not surprisingly 
so, since the Unicode algorithm is defined in a way that makes it very 
hard to make it work per character).

> There are some cluefull people around that may be able to deal with the
> BIDI part of the work, but there seem to be non that can do this and
> know emacs internals well (at least non that have free time :)

This issue is on the top of my TODO, and I'm thinking about it and 
testing ideas all the time.  However, I will probably be unable to
start coding anything before Emacs 21.1 is released.  In part, this
is because I need Gerd's help and feedback, and Gerd is swamped by
the pretest.

People who are unhappy about the progress of adding bidi support to Emacs 
(I know _I_ am unhappy ;-), and are motivated enough to invest some of 
their time, are hereby encouraged to study the Emacs display code in 
order to make themselves acquainted with the relevant internals.  Then 
they could offer real help, both in discussing implementation details and 
in writing parts of the code.  It is not rocket science to understand
how does the display engine work, at least in principle.  It will
certainly take time, but given that nobody is actively coding anything
in this area, time is not an issue right now.

> A partially working version can be a good bases for some "classic
> evolution" code development, adding one feature at a time, one bugfix at
> a time.  This may not be the ideal code writing mechanism, but it can
> work nicely nevertheless.  Waiting for The Perfect Solution may take
> much longer.

We don't wait for a pefect solution.  We are looking for a solution that 
is a reasonable starting point, which can be developed and improved in
the future.  A solution whose design is wrong is a dead end and working 
on it is a waste of resources, even if it partially works.

Faced with the implementation you mention, Gerd Moellmann felt that it 
does not fit well into the display engine operation.  It is IMHO not 
wise, to say the least, to go against Gerd's opinion on this, since no 
one knows more than he does about the current display code.