Re: [emacs-bidi] mirroring of glyphs

[Eli Zaretskii]

On Wed, 21 Nov 2001, Ehud Karni wrote:

> > > > I think the only way to work reasonably well in such cases is to look
> > > > at preceding and following characters, and implement some of Ehud's
> > > > wisdom to fix UAX#9 lossage.
> 
> I'm blushing -:) 

Why should you?  UAX#9 should learn from what you did, instead of
learning from Microsoft.

> > > The following characters might not exist (yet), and generally inserting
> > > bidi characters beyond the users back is a risky business.
> > 
> > I agree that we should minimize the insertion of formatting codes,
> > but there are a few specific cases where UAX#9 does nonsensical
> > things, which users won't forgive us.
> 
> I think the real solution is to work in some way like the good old
> Einstein Writer (always a good reference). When there is some phrase
> that the UAX#9 is just doing mishmash of it (like mathematical
> expression [see the difference between 1-5 1/5 1*5] or physical
> constants like -5.0 C enclosed by Hebrew text) then Emacs should have
> "visual mode" that will let user enter each character from left to
> right (or right to left in RTL paragraph) disregarding it UNICODE
> type (this like LRO ... PDF).

IMHO, this is a separate feature (I do agree that we need to support
it).  From my experience, visual insertion is a pain, so it should be
the last resort.

> The real tricky (and wise) part will
> be to convert that string to the best (not defined yet) logical order
> so the appearance will be the same but search (done in logical order)
> will find the relevant parts (like 23.45 for numbers).

Perhaps we could do this in some cases by inserting LRM or RLM.

> We need to be comply with UNICODE, but we also need to make life
> easier for the users.

Agreed; and therein lies the problem, since Unicode doesn't always do
that.