New issue from Hewill: "Missing Preconditions for take/drop adaptor"

Author: Dani-Hub

xml/issue4214.xml

@@ -0,0 +1,87 @@
+<?xml version='1.0' encoding='utf-8' standalone='no'?>
+<!DOCTYPE issue SYSTEM "lwg-issue.dtd">
+
+<issue num="4214" status="New">
+<title>Missing <i>Preconditions</i> for <code>take</code>/<code>drop</code> adaptor</title>
+<section>
+<sref ref="[range.take.overview]"/><sref ref="[range.drop.overview]"/>
+</section>
+<submitter>Hewill Kang</submitter>
+<date>15 Feb 2025</date>
+<priority>99</priority>
+
+<discussion>
+<p>
+The <code>take</code>/<code>drop</code> adaptor does not explicitly require <i>N</i> to be non-negative (although
+the view class does), which makes it possible for some specialized cases to be well-defined when <i>N</i> is negative, 
+since no <i>Preconditions</i> are violated:</p>
+<blockquote><pre>
+auto e = std::views::empty&lt;int&gt; 
+       | std::views::take(-1);   // []
+auto i = std::views::iota(1, 5)
+       | std::views::drop(-1);   // [0, 1, 2, 3, 4]
+auto r = std::views::repeat('a', 2)
+       | std::views::drop(-1);   // ['a', 'a', 'a']
+</pre></blockquote>
+<p>
+This is not the intention, we should ban these cases.
+</p>
+</discussion>
+
+<resolution>
+<p>
+This wording is relative to <paper num="N5001"/>.
+</p>
+
+<ol>
+<li><p>Modify <sref ref="[range.take.overview]"/> as indicated:</p>
+
+<blockquote>
+<p>
+-2- The name <code>views::take</code> denotes a range adaptor object (<sref ref="[range.adaptor.object]"/>).
+Let <code>E</code> and <code>F</code> be expressions, let <code>T</code> be
+<code>remove_cvref_t&lt;decltype((E))&gt;</code>, and let <code>D</code> be 
+<code>range_difference_t&lt;decltype((E))&gt;</code>.
+If <code>decltype((F))</code> does not model <code>convertible_to&lt;D&gt;</code>,
+<code>views::take(E, F)</code> is ill-formed. Otherwise, the expression <code>views::take(E, F)</code> 
+is expression-equivalent to:
+</p>
+<ol style="list-style-type: none">
+<li><p>(2.?) <ins>&mdash; <i>Preconditions</i>: <code>static_cast&lt;D&gt;(F) &gt;= 0</code> is <code>true</code>.</ins>
+</p></li>
+<li><p>(2.1) &mdash; if <code>T</code> is a specialization of <code>empty_view</code> (<sref ref="[range.empty.view]"/>),
+then <code>((void)F, <i>decay-copy</i>(E))</code>, except that the evaluations of <code>E</code>
+and <code>F</code> are indeterminately sequenced.
+</p></li>
+</ol>
+</blockquote>
+</li>
+
+<li><p>Modify <sref ref="[range.drop.overview]"/> as indicated:</p>
+
+<blockquote>
+<p>
+-2- The name <code>views::drop</code> denotes a range adaptor object (<sref ref="[range.adaptor.object]"/>).
+Let <code>E</code> and <code>F</code> be expressions, let <code>T</code> be
+<code>remove_cvref_t&lt;decltype((E))&gt;</code>, and let <code>D</code> be 
+<code>range_difference_t&lt;decltype((E))&gt;</code>.
+If <code>decltype((F))</code> does not model <code>convertible_to&lt;D&gt;</code>,
+<code>views::drop(E, F)</code> is ill-formed. Otherwise, the expression <code>views::drop(E, F)</code> 
+is expression-equivalent to:
+</p>
+<ol style="list-style-type: none">
+<li><p>(2.?) <ins>&mdash; <i>Preconditions</i>: <code>static_cast&lt;D&gt;(F) &gt;= 0</code> is <code>true</code>.</ins>
+</p></li>
+<li><p>(2.1) &mdash; if <code>T</code> is a specialization of <code>empty_view</code> (<sref ref="[range.empty.view]"/>),
+then <code>((void)F, <i>decay-copy</i>(E))</code>, except that the evaluations of <code>E</code>
+and <code>F</code> are indeterminately sequenced.
+</p></li>
+</ol>
+</blockquote>
+</li>
+
+</ol>
+
+</resolution>
+
+</issue>