<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<div class="moz-cite-prefix">On 7/3/2022 2:20 AM, Andreas Prilop via
Unicode wrote:<br>
</div>
<blockquote type="cite"
cite="mid:7B6DAF42-3ABF-4C45-BD11-29EBD6B512BB@fn.de">
<pre class="moz-quote-pre" wrap="">On 3 July 2022, Eli Zaretskii wrote:
</pre>
<blockquote type="cite">
<pre class="moz-quote-pre" wrap="">I thin UAX#9 clarifies it perfectly: numbers are displayed in LTR order.
</pre>
</blockquote>
<pre class="moz-quote-pre" wrap="">
This is undisputed.
I ask about the differences
“555-2368” vs. “2368-555”
“1=3−2” vs. “1=2−3”
“1999-12-31” vs. “31-12-1999”
The Bidirectional Algorithm is responsible for these differences. But why?</pre>
</blockquote>
<p><font face="Candara">The real answer is that this matches
differences in displaying lists of numbers (!) not order of
digits, in Hebrew vs. Arabic.</font></p>
<p><font face="Candara">The Bidi algorithm uses the classes AL and
AN (and rules that resolve them) to implement these inherent
differences in the way the various scripts handle such cases
(multiple groups of digits separated by punct).</font></p>
<p><font face="Candara">As I mentioned, I raised a public review
issue to make sure that UAX#9 either *specifically and
explicitly* cites or, alternatively, incorporates language that
explains scripts have different preferences in resolving groups
of numbers (not: digits) and points in a high level to where in
the spec these preferences are addressed.</font></p>
<p><font face="Candara">I agree, it's not enough to reverse engineer
the algorithm and conclude that it behaves as specd. It should
be a simple matter to understand why it was designed the way it
was.<br>
</font></p>
<p><font face="Candara">A./</font><br>
</p>
</body>
</html>