Proposing new arrow characters with Bidi_Mirrored=Yes

[Eli Zaretskii]

> Date: Thu, 10 Apr 2025 19:27:43 -0700
> From: Asmus Freytag via Unicode <unicode at corp.unicode.org>
> 
> On 4/10/2025 5:14 PM, Mark E. Shoulson via Unicode wrote:
> 
>  does glyph-mirroring make sense for things that don't already have a mirrored glyph at another
>  codepoint? 
> 
> It does. Contour integral needs to be "mirrored" but the direction of the arrow (circular) needs to be
> preserved. There are no mirrored code points, as the (partially) mirrored shape *only* exists in RTL display.
> 
> However, math font providers need to know about this and have the correct glyph available. 

That takes us back to the shaping engine and the font part, which I
thought someone said was a separate issue?

For showing plain text on a text-only terminal, which doesn't know
about the underlying font and its rendering, any features that require
to interact with the font are a non-starter.  By contrast, the simple
rules of bidi mirroring stated in UAX#9 work in all cases, including
when showing bidi text on dumb text-only terminals.

In addition, a related discussion on the HarfBuzz forum
(https://github.com/harfbuzz/harfbuzz/discussions/5278) reveals that
the relevant OpenType features are probably supported only by a small
set of fonts, if at all.  So leaving this to fonts and shaping engines
will at best solve only a small part of the use cases.