<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<div class="moz-cite-prefix">On 10/22/2021 4:07 PM, Richard
Wordingham via Unicode wrote:<br>
</div>
<blockquote type="cite" cite="mid:20211023000704.34f03ca0@JRWUBU2">
<pre class="moz-quote-pre" wrap="">Indeed, I have trouble with the auxiliary 'should' in TUS. If I read
TUS as a specification, vast swathes evaporate. </pre>
</blockquote>
<p><font face="Candara">The IETF sense of "SHOULD" is to make it
something you can opt out of, if you know better or if
circumstances require. But it's considered that you have a very
good reason to disregard the statement. (There's an RFC that
describes the terminology).<br>
</font></p>
<p><font face="Candara">Now, conformance testing can't test the
validity of your reasoning, but I'd start with the presumption
that if you "evaporate" all these statements, your reasons don't
meet any strength criteria at all, which would put you at odds
with that definition of SHOULD.</font></p>
<p><font face="Candara">Now, TUS is not an IETF standard, but it
exists in the context of standards where the drafters have taken
pains to describe what they mean by "should". So, I'd argue, the
general reading of "should" that you "should" apply to TUS is
not the full range of meaning in ordinary language, but the more
restricted reading common to standards (and preferably standards
that cover similar or closely related fields).</font></p>
<p><font face="Candara">Therefore, I argue, the proper reading of
"should" in TUS is to treat it initially as mandatory, but to
consider its effects and if it fails in an edge case or seems
otherwise inapplicable after specific reflection, you may opt
out without affecting conformance.</font></p>
<p><font face="Candara">There may be an interoperability cost if you
opt out; but it may be outweighed in certain scenarios.</font></p>
<p><font face="Candara">In contrast, the term "MAY" for IETF, or
"may" in Unicode describes something that is truly optional, and
where, perhaps there exists even probability for preferring one
over the other. (If opting in had been considered something that
should only be done rarely and where truly necessary, the
statement would have used "SHOULD NOT").</font></p>
<p><font face="Candara">Hope this helps,</font></p>
<p><font face="Candara">A./</font><br>
</p>
</body>
</html>