On 10/04/2025 19:29, Asmus Freytag via Unicode wrote:
> On 4/10/2025 4:46 AM, ((Nitai Sasson)) via Unicode wrote:
> > I honestly cannot think of any issue or potential pitfall with this 
> > solution.
> That in itself is a problem.

I'm open minded. Show me I'm wrong - I've been convinced by good arguments 
multiple times in this chain already.

On 10/04/2025 14:46, Marius Spix via Unicode wrote:
> I'm afraid, that this would open Pandora's box for more homograph attacks. ∃ 
> (U+2203) is already a confusable for E in a BiDi context. But this would add 
> more characters like ꟼ (U+A7FC), ⅃ (U+2143) or ↋ (U+218B) to the list.

Letters are strongly directional, so under normal conditions would not be 
affected by <BDM>. In other words, L<BDM> would always be rendered as L, never 
⅃.

Except... I actually have had some thoughts about potential exceptions to this, 
but I'm not sure if right now is the best time to get into the details of what 
I had thought of. I first wanted to get feedback for the intended use case of 
the proposed <BDM>.

Of course, *today* within LTR text you can insert: <RLI>)<PDI> and end up with 
a homograph of (. Here it is: "⁧)⁩". This would not create a brand new issue, 
this is merely an inevitable consequence of bidi text.

I also pretty much agree with Joao on this:
On 10/04/2025 19:45, Joao S. O. Bueno via Unicode <[email protected]> 
wrote:
> Other security measures have to be taken for this kind of attack -
> and having more ways to spell out a forward arrow is certainly not
> changing the current landscape where an homograph, or close homograph
> for that matter, to every ascii letter can probably be found nearly 20
> times in the unicode charset.

Reply via email to