New issue from Jonathan: unique_copy passes arguments to its predicate backwards

Author: jwakely

xml/issue4269.xml

@@ -0,0 +1,94 @@
+<?xml version='1.0' encoding='utf-8' standalone='no'?>
+<!DOCTYPE issue SYSTEM "lwg-issue.dtd">
+
+<issue num="4269" status="New">
+<title>`unique_copy` passes arguments to its predicate backwards</title>
+<section><sref ref="[alg.unique]"/></section>
+<submitter>Jonathan Wakely</submitter>
+<date>29 May 2025</date>
+<priority>99</priority>
+
+<discussion>
+<p>
+For the `unique` algorithms, <sref ref="[alg.unique]"/> p1 says:
+</p>
+<blockquote>
+1. Let `pred` be `equal_to{}` for the overloads with no parameter `pred`,
+and let <i>E</i> be
+<ol style="list-style-type: none">
+<li> (1.1)
+ &mdash; `bool(pred(*(i - 1), *i))` for the overloads in namespace `std`;
+</li>
+<li>(1.2)
+ &mdash; `bool(invoke(comp, invoke(proj, *(i - 1)), invoke(proj, *i)))`
+ for the overloads in namespace `ranges`.
+</li>
+</ol>
+</blockquote>
+<p>
+However for the `unique_copy` algorithms, <sref ref="[alg.unique]"/> p6 says
+that the arguments `*i` and `*(i-1)` should be reversed:
+</p>
+<blockquote>
+6. Let `pred` be `equal_to{}` for the overloads with no parameter `pred`,
+and let <i>E</i> be
+<ol style="list-style-type: none">
+<li> (6.1)
+ &mdash; `bool(pred(*i, *(i - 1)))` for the overloads in namespace `std`;
+</li>
+<li>(6.2)
+ &mdash; `bool(invoke(comp, invoke(proj, *i), invoke(proj, *(i - 1))))`
+ for the overloads in namespace `ranges`.
+</li>
+</ol>
+</blockquote>
+<p>
+This reversed order is consistent with the documentation for
+<a href="https://stl.boost.org/unique_copy.html">SGI STL `unique_copy`</a>,
+although the docs for 
+<a href="https://stl.boost.org/unique_copy.html">SGI STL `unique`</a>
+show reversed arguments too, and the C++ standard doesn't match that.
+</p>
+<p>
+A survey of known implementations shows that all three of libstdc++, libc++,
+and MSVC STL use the `pred(*(i - 1), *i)` order for all of `std::unique`,
+`std::unique_copy`, `ranges::unique`, and `ranges::unique_copy`. The range-v3
+library did the same, and even the SGI STL did too (despite what its docs said).
+Only two implementations were found which match the spec and use a different
+argument order for `unique` and `unique_copy`, Casey Carter's (cmcstl2) and
+Fraser Gordon's.
+</p>
+<p>
+In the absence of any known rationale for `unique` and `unique_copy` to differ,
+it seems sensible to make `unique_copy` more consistent with `unique`
+(and with the majority of implementations stretching back three decades).
+</p>
+</discussion>
+
+<resolution>
+<p>
+This wording is relative to <paper num="N5008"/>.
+</p>
+
+<ol>
+<li><p>Modify <sref ref="[alg.unique]"/> as indicated:</p>
+<blockquote>
+6. Let `pred` be `equal_to{}` for the overloads with no parameter `pred`,
+and let <i>E</i> be
+<ol style="list-style-type: none">
+<li> (6.1)
+ &mdash; <code>bool(pred(<del>*i,</del> *(i - 1))<ins>, *i</ins>)</code>
+ for the overloads in namespace `std`;
+</li>
+<li>(6.2)
+ &mdash; <code>bool(invoke(comp, <del>invoke(proj, *i),</del>
+ invoke(proj, *(i - 1))<ins>, invoke(proj, *i)</ins>))</code>
+ for the overloads in namespace `ranges`.
+</li>
+</ol>
+</blockquote>
+</li>
+</ol>
+</resolution>
+
+</issue>