Re: Emacs as word processor / Text Properties

[Eli Zaretskii <eliz@gnu.org>]

> From: Jambunathan K <kjambunathan@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Fri, 29 Nov 2013 19:12:32 +0530
> 
> You are saying that explicit markers could be used in place of
> per-paragraph tagging to achieve the desired effect.

No, I'm saying more than that: inserting those control characters _is_
the Emacs way of tagging the paragraph as having a certain base
direction.  Since Emacs is mostly about plain-text files, there's no
other way to achieve that effect, while being sure that any other
random UBA-compliant application will obey that.

> May be it is not a question of "enough" but a question of
> "manageable". i.e., minimizing or eliminating undesired effects while
> moving - copy + pasting - the text around.

I see no adverse effects from copy/paste due to these controls, except
those effects that are already here to stay: in general, pasting
bidirectional text into a middle of another text can already have
profound non-local effects on how the paragraph into which you paste
is displayed.  RLM and LRM are just 2 strong directional characters,
so their effect is the same as that of some Arabic or Hebrew
character.

> The FAQ below talks about avoiding paired control codes (to avoid
> goofing up of structure) and says RLM and LRM are "less problematic".

Don't take what w3.org says about (X)HTML and XML as having any
relevance whatsoever on what Emacs does or needs to do.  Emacs deals
with plain text that is generally devoid of any persistent tags or
other meta-data, so it cannot possibly DTRT by tags and markup, it
must use only characters and directional controls.  By contrast,
w3.org is about markup languages, where the situation is entirely
different, and actually using LRM, RLM, and other Unicode controls is
"considered harmful".

> The scenario I am placing is this:
> 
>      Org-mode table + Multi-paragraph table cells + Mixed scripts.
> 
> >From the FAQ above,     
> 
> ,----
> | ... Using Unicode controls results in a lot more work to achieve the
> | same result. Also it is difficult to know how to achieve effects like
> | reversing table columns and right-aligning text with just Unicode
> | control codes.
> `----

"A lot more work" might be relevant for a human who prepares a Web
page by hand, but it is no excuse for an Emacs mode.  A layman might
have hard time figuring out the effect of this or that technique wrt
bidirectional text, but we in Emacs have enough knowhow and experience
to deal with that correctly, and the infrastructure already exists to
make it possible.  All we need is for modes to use that.