Now: Paragraph Direction Detection and Harmonization -- Was: Re: Bidirectional editing in Emacs -- main design decisions

[Stephen J. Turnbull]

Mohsen BANAN writes:

 >   Eli> So the Emacs display is correct, is that
 >   Eli> what you are saying?  
 > 
 > I am saying that emacs display is correct

Oh, why is there no period here?

 > but that there are interoperability problems.

Please, let's not go there.  We know one good path (not perfect, but
good) to interoperability: Implement UTR#9.  The other paths are the
road to interoperability hell, most likely.

So it's too bad for the users of the other application, then, unless
they're programmers capable of hacking the code themselves.  As a
20-year veteran of the Japanese interoperability disaster, which was
two decades in progress when I joined it and which continues to this
very day, please, have patience, for five years if it takes that long.
(< 5 40) => t!

The thing is, every concession to broken implementations delays true
harmony (by which I mean "reliable interoperability", not "pedantic
standards conformance").  Not to mention strengthening proprietary
software, which often seeks to exploit the "tippiness" induced by
interoperability problems.

 > The nature of bidi text is such that it demands
 > harmony.

True, but overly specific.  s/bidi//  ;-)  Everybody thinks their
problem is special.  Of course it is in one way, but in another, it's
not.

 >   Eli> I don't care about Firefox
 > 
 > I have a different view. I see Emacs and Firefox as joint sisters.

Of course users care.  But user wishes don't generate code.  Eli cares
about users; he's proved that amply over the years.  So if he says he
"doesn't care", that's not just making obscene gestures at the users,
that's his expert judgment that it's not worth doing.  He may not be
able to "prove" it, but his expertise is proven to me by 2 decades of
acquaintance.

If you have the ability to code, Eli will help.  I can testify to
that, too.  But "don't care" means Eli isn't going to do it himself,
it's too much work compared to the gains possible by working on other
areas.

And it's not like the Firefox developers don't care, either.  It would
be much better to concentrate on lobbying them.  They *want* to be
correct, and will be made happier once Firefox conforms to the
standard.

 > I think the solution is to generate html and
 > specify paragraph direction explicitly in html.

That's OK IMO, although somewhat undesirable.  (It will slow the
adoption of correct and full implementations of the bidi algorithm.)

That's not an Emacs problem, though.  That's an MUA problem.  I guess
you're talking to the right person on that, since RMail is still the
official Emacs MUA, and Eli uses and (sometimes, at least) contributes
to RMail.  But if you want other people, who use a wide variety of
MUAs, to do it that way, you'll need to talk to larsi and/or Ted
Zlatanov (Gnus/message-mode) and Uday Reddy (VM), at least.

 > Beyond the basic bidi capability in emacs, there
 > are several layers above it that we now need to
 > cultivate.

+1 on that, though.  I hope Eli will focus on that, personally.  (But
that's 'cause I'm a standards geek, so you can multiply that by about
0.1 importance factor -- I have no personal need for bidi, or
experience with it. Someday ... :-)