<!DOCTYPE html>
<html><head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
     <style>
/**
 * SyntaxHighlighter
 * http://alexgorbatchev.com/SyntaxHighlighter
 *
 * SyntaxHighlighter is donationware. If you are using it, please donate.
 * http://alexgorbatchev.com/SyntaxHighlighter/donate.html
 *
 * @version
 * 3.0.83 (July 02 2010)
 * 
 * @copyright
 * Copyright (C) 2004-2010 Alex Gorbatchev.
 *
 * @license
 * Dual licensed under the MIT and GPL licenses.
 */
.syntaxhighlighter {
  background-color: white !important;
}
.syntaxhighlighter .line.alt1 {
  background-color: white !important;
}
.syntaxhighlighter .line.alt2 {
  background-color: white !important;
}
.syntaxhighlighter .line.highlighted.alt1, .syntaxhighlighter .line.highlighted.alt2 {
  background-color: #e0e0e0 !important;
}
.syntaxhighlighter .line.highlighted.number {
  color: black !important;
}
.syntaxhighlighter table caption {
  color: black !important;
}
.syntaxhighlighter .gutter {
  color: #afafaf !important;
}
.syntaxhighlighter .gutter .line {
  border-right: 3px solid #6ce26c !important;
}
.syntaxhighlighter .gutter .line.highlighted {
  background-color: #6ce26c !important;
  color: white !important;
}
.syntaxhighlighter.printing .line .content {
  border: none !important;
}
.syntaxhighlighter.collapsed {
  overflow: visible !important;
}
.syntaxhighlighter.collapsed .toolbar {
  color: blue !important;
  background: white !important;
  border: 1px solid #6ce26c !important;
}
.syntaxhighlighter.collapsed .toolbar a {
  color: blue !important;
}
.syntaxhighlighter.collapsed .toolbar a:hover {
  color: red !important;
}
.syntaxhighlighter .toolbar {
  color: white !important;
  background: #6ce26c !important;
  border: none !important;
}
.syntaxhighlighter .toolbar a {
  color: white !important;
}
.syntaxhighlighter .toolbar a:hover {
  color: black !important;
}
.syntaxhighlighter .plain, .syntaxhighlighter .plain a {
  color: black !important;
}
.syntaxhighlighter .comments, .syntaxhighlighter .comments a {
  color: #008200 !important;
}
.syntaxhighlighter .string, .syntaxhighlighter .string a {
  color: blue !important;
}
.syntaxhighlighter .keyword {
  color: #006699 !important;
}
.syntaxhighlighter .preprocessor {
  color: gray !important;
}
.syntaxhighlighter .variable {
  color: #aa7700 !important;
}
.syntaxhighlighter .value {
  color: #009900 !important;
}
.syntaxhighlighter .functions {
  color: #ff1493 !important;
}
.syntaxhighlighter .constants {
  color: #0066cc !important;
}
.syntaxhighlighter .script {
  font-weight: bold !important;
  color: #006699 !important;
  background-color: none !important;
}
.syntaxhighlighter .color1, .syntaxhighlighter .color1 a {
  color: gray !important;
}
.syntaxhighlighter .color2, .syntaxhighlighter .color2 a {
  color: #ff1493 !important;
}
.syntaxhighlighter .color3, .syntaxhighlighter .color3 a {
  color: red !important;
}

.syntaxhighlighter .keyword {
  font-weight: bold !important;
}
.syntaxhighlighter div.toolbar span a.toolbar_item{
   display: none !important;
} 
body .syntaxhighlighter .line {
    white-space: pre !important;
}
.pln{color:#000}@media screen{.str{color:#080}.kwd{color:#008}.com{color:#800}.typ{color:#606}.lit{color:#066}.clo,.opn,.pun{color:#660}.tag{color:#008}.atn{color:#606}.atv{color:#080}.dec,.var{color:#606}.fun{color:red}}@media print,projection{.kwd,.tag,.typ{font-weight:700}.str{color:#060}.kwd{color:#006}.com{color:#600;font-style:italic}.typ{color:#404}.lit{color:#044}.clo,.opn,.pun{color:#440}.tag{color:#006}.atn{color:#404}.atv{color:#060}}pre.prettyprint{padding:2px;border:1px solid #888}ol.linenums{margin-top:0;margin-bottom:0}li.L0,li.L1,li.L2,li.L3,li.L5,li.L6,li.L7,li.L8{list-style-type:none}li.L1,li.L3,li.L5,li.L7,li.L9{background:#eee}
    </style>

<title>Filling holes in Class Template Argument Deduction</title>
</head>

<body>
<p>P1021R3<br>
Mike Spertus, Symantec<br>
<a href="mailto:mike_spertus@symantec.com">mike_spertus@symantec.com</a><br>
Timur Doumler<br>
<a href="mailto:papers@timur.audio">papers@timur.audio</a><br>
Richard Smith<br>
<a href="mailto:richardsmith@google.com">richardsmith@google.com</a><br>
2018-11-26<br>
Audience: Core Working Group
</p>
<h1>Filling holes in Class Template Argument Deduction</h1>
This paper proposes filling several gaps in <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0091r3.html">Class Template Argument Deduction</a>.

<h2>Document revision history</h2>

<b>R0</b>, 2018-05-07: Initial version<br>
<b>R1</b>, 2018-10-07: Refocused paper on filling holes in CTAD<br>
<b>R2</b>, 2018-11-26: Following EWG guidance, removed proposal for CTAD from partial template argument lists<br>
<b>R3</b>, 2019-01-21: Add wording and change target to Core Working Group

<h2>Rationale</h2>
As one of us (Timur) has noted when giving public presentations on using
 class template argument deduction, there are a significant number of 
cases where it cannot be used. This always deflates the positive 
feelings from the rest of the talk because it is accurately regarded as 
artificially inconsistent. In particular, listeners are invariably 
surprised that it does not work with aggregate templates, type aliases, 
and inherited constructors.
    We will show in this paper
that these limitations can be safely removed. Note that some of these 
items were intentionally deferred from
    C++17 with the intent of adding them in C++20. 
<h2>Class Template Argument Deduction for aggregates</h2>
We propose that Class Template Argument Deduction works for aggregate initialization as one
	shouldn't have to choose between having aggregates and deduction. This is well illustrated
	by the following example:
<table border="1"><tbody><tr><th>C++17</th><th>Proposed</th></tr>
<tr><td><div><div id="highlighter_153831" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">class</code> <code class="cpp plain">T&gt;</code></div><div class="line number2 index1 alt1"><code class="cpp keyword bold">struct</code> <code class="cpp plain">Point { T x; T y; };</code></div><div class="line number3 index2 alt2">&nbsp;</div><div class="line number4 index3 alt1"><code class="cpp comments">// Aggregate: Cannot deduce</code></div><div class="line number5 index4 alt2"><code class="cpp plain">Point&lt;</code><code class="cpp color1 bold">double</code><code class="cpp plain">&gt; p{3.0, 4.0};</code></div><div class="line number6 index5 alt1"><code class="cpp plain">Point&lt;</code><code class="cpp color1 bold">double</code><code class="cpp plain">&gt; p2{.x = 3.0, .y = 4.0};</code></div></div></td></tr></tbody></table></div></div></td>    
    <td><div><div id="highlighter_642490" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">class</code> <code class="cpp plain">T&gt;</code></div><div class="line number2 index1 alt1"><code class="cpp keyword bold">struct</code> <code class="cpp plain">Point { T x; T y; };</code></div><div class="line number3 index2 alt2">&nbsp;</div><div class="line number4 index3 alt1"><code class="cpp comments">// Proposed: Aggregates deduce</code></div><div class="line number5 index4 alt2"><code class="cpp plain">Point p{3.0, 4.0};</code></div><div class="line number6 index5 alt1"><code class="cpp plain">Point p2{.x = 3.0, .y = 4.0};</code></div></div></td></tr></tbody></table></div></div></td>
    </tr></tbody></table>

For the code on the right hand side to work in C++17, the user would 
have to write an explicit deduction guide. We believe that this is 
unnecessary and error-prone boilerplate and that the necessary deduction
 guide should be implicitly synthesized by the compiler  from the 
arguments of a braced  or designated initializer during aggregate 
initialization.
<h3>Algorithm</h3>
<p>In the current Working Draft, an aggregate class is defined as a 
class with no user-declared or inherited constructors, no private or 
protected non-static data members, no virtual functions, and no virtual,
 private or protected base classes. While we would like to generate an 
aggregate deduction guide for class templates 
    that comply with these rules, we first need to consider the case 
where there is a dependent base
    class that may have virtual functions, which would violate the rules
 for aggregates. Fortunately,
    that case does not cause a problem because any deduction guides that
 require one or more arguments
    will result in a compile-time error after instantiation, and 
non-aggregate classes without
    user-declared or inherited constructors generate a zero-argument 
deduction guide anyway.
    Based on the above, we can safely generate an aggregate deduction 
guide for class templates
    that comply with aggregate rules.    


</p><p>When <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0960r1.html">P0960R0</a>
 was discussed in Rapperswil, it was voted that in order to allow 
aggregate initialization from a parenthesized list of arguments, 
aggregate initialization should proceed as if there was a synthesized 
constructor. We can use the same approach to also synthesize the 
required additional deduction guide during class template argument
    deduction as follows:

</p><ol>
    <li>Given a primary class template <tt>C</tt>, determine whether it satisfies all the conditions for an aggregate class ([dcl.init.aggr]/1.1 - 1.4).</li>
    <li>If yes, let  <tt>T_1</tt>, ...., <tt>T_n</tt> denote the types of the <tt>N</tt> elements ([dcl.init.aggr]/2) of the aggregate (which may or may not depend on its template arguments).</li>
    <li>Form a hypothetical constructor <tt>C(T_1, ..., T_N)</tt>. </li>
    <li>For every constructor argument of type <tt>T_i</tt>, if all types <tt>T_i ... T_n</tt> are default-constructible, add a default argument value zero-initializing the argument as if by <tt>T_i = {}</tt>.</li>
    <li>For the set of functions and function templates formed for 
[over.match.class.deduct], add an additional function template derived 
from this hypothetical constructor as described in 
[over.match.class.deduct]/1.</li>
</ol>

There is a slight complication resulting from subaggregates, and the 
fact that nested braces can be omitted when instantiating them:

<blockquote><div><div id="highlighter_734434" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">struct</code> <code class="cpp plain">Foo { </code><code class="cpp color1 bold">int</code> <code class="cpp plain">x, y; };</code></div><div class="line number2 index1 alt1"><code class="cpp keyword bold">struct</code> <code class="cpp plain">Bar { Foo f; </code><code class="cpp color1 bold">int</code> <code class="cpp plain">z; };</code></div><div class="line number3 index2 alt2"><code class="cpp plain">Bar bar{1, 2};&nbsp;&nbsp; </code><code class="cpp comments">// Initializes bar.f.x and bar.f.y to 1; zero-initializes bar.z</code></div></div></td></tr></tbody></table></div></div></blockquote>

In this case, we have two initializers, but they do not map to the two elements of the aggregate type <tt>Bar</tt>, instead initializing the sub-elements of the first subaggregate element of type <tt>Foo</tt>. 

<p>For complicated nested aggregates, there are potentially many 
different combinations of valid mappings of initializers to subaggregate
 elements. It would be unpractical to create hypothetical constructors 
for all of those combinations. Additionally, whether or not an aggregate
 type has subaggregate elements may depend on the template arguments:

</p><blockquote><div><div id="highlighter_611034" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code> <code class="cpp plain">&lt;</code><code class="cpp keyword bold">typename</code> <code class="cpp plain">T&gt;</code></div><div class="line number2 index1 alt1"><code class="cpp keyword bold">struct</code> <code class="cpp plain">Bar { T f; </code><code class="cpp color1 bold">int</code> <code class="cpp plain">z; };&nbsp; </code><code class="cpp comments">// T might be a subaggregate!</code></div></div></td></tr></tbody></table></div></div></blockquote>

This information is not available during class template argument deduction, because for this we first need to deduce <tt>T</tt>.
 We therefore propose simply to avoid all of these problems by 
prohibiting the omission of nested braces when performing class template
 argument deduction.


<h2>Class Template Argument Deduction for alias templates</h2>
While Class Template Argument Deduction makes type inferencing easier 
when constructing classes,
it doesn't work for type aliases, penalizing the use of type aliases and
 creating unnecessary inconsistency.We propose allowing Class Template 
Argument Deduction for type aliases as in the following example.

    <table border="1"><tbody><tr><th><tt>vector</tt></th><th><tt>pmr::vector</tt> (C++17)</th><th>pmr::vector (proposed)</th></tr>
<tr>
    <td>
<div><div id="highlighter_776658" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp plain">vector v = {1, 2, 3};</code></div><div class="line number2 index1 alt1"><code class="cpp plain">vector v2(cont.begin(), cont.end());</code></div></div></td></tr></tbody></table></div></div>
    </td>
    <td><div><div id="highlighter_343467" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp plain">pmr::vector&lt;</code><code class="cpp color1 bold">int</code><code class="cpp plain">&gt; v = {1, 2, 3};</code></div><div class="line number2 index1 alt1"><code class="cpp plain">pmr::vector&lt;decltype(cont)::value_type&gt; v2(cont.begin(), cont.end());</code></div></div></td></tr></tbody></table></div></div>
    </td>
<td><div><div id="highlighter_223547" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp plain">pmr::vector v = {1, 2, 3};</code></div><div class="line number2 index1 alt1"><code class="cpp plain">pmr::vector v2(cont.begin(), cont.end());</code></div></div></td></tr></tbody></table></div></div>
</td>
</tr></tbody></table>
    <tt>pmr::vector</tt> also serves to illustrate another interesting case. While one might
    be tempted to write
    <blockquote><div><div id="highlighter_174382" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp plain">pmr::vector pv({1, 2, 3}, mem_res); </code><code class="cpp comments">// Ill-formed (C++17 and proposed)</code></div></div></td></tr></tbody></table></div></div></blockquote>
    this example is ill-formed by the normal rules of template argument deduction because
    <tt>pmr::memory_resource</tt> fails to 
    deduce <tt>pmr::polymorphic_allocator&lt;int&gt;</tt> in the  second argument. 
    <p>While this is to be expected, one suspects that had class template argument deduction
        for alias templates been around when <tt>pmr::vector</tt> was being designed,
        the committee would have considered allowing such an inference as safe and useful in this context.
        If that was desired, it could easily have been achieved 
        by indicating that the <code>pmr::allocator</code> template parameter should be considered non-deducible:
    </p><blockquote><div><div id="highlighter_550973" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">namespace</code> <code class="cpp plain">pmr {</code></div><div class="line number2 index1 alt1"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">typename</code> <code class="cpp plain">T&gt;</code></div><div class="line number3 index2 alt2"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp keyword bold">using</code> <code class="cpp plain">vector = std::vector&lt;T, type_identity_t&lt;pmr::polymorphic_allocator&lt;T&gt;&gt;&gt;; </code><code class="cpp comments">// See P0887R1 for type_identity_t</code></div><div class="line number4 index3 alt1"><code class="cpp plain">}</code></div><div class="line number5 index4 alt2"><code class="cpp plain">pmr::vector pv({1, 2, 3}, mem_res); </code><code class="cpp comments">// OK with this definition</code></div></div></td></tr></tbody></table></div></div></blockquote>
 Finally, in the spirit of alias templates being simply an alias for the type, we do not propose
allowing the programmer to write explicit deduction guides specifically for an alias
template.
<h3>Algorithm</h3>
For deriving deduction guides for the alias templates from guides in the class, we use the following approach (for which we are very grateful for the invaluable assistance of Richard Smith):
<ol><li>Deduce template parameters for the deduction guide by deducing the right hand side of
    the deduction guide from the alias template. We do not require that this deduces all the template
    parameters as nondeducible contexts may of course occur in general</li>
    <li>Substitute any deductions made back into the deduction guides. Since the previous step may not
    have deduced all template parameters of the deduction guide, the new guide may have template
    parameters from both the type alias and the original deduction guide.</li>
    <li>Derive the corresponding deduction guide for the alias template by 
    deducing the alias from the result type. Note that a constraint may be necessary
    as whether and how to deduce the alias from the result type may depend on the
    actual argument types.</li>
    <li>The guide generated from the copy deduction guide should be given
    the precedence associated with copy deduction guides during overload resolution</li></ol>
Let us illustrate this process with an example. Consider the following example:
<blockquote><div><div id="highlighter_41752" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">class</code> <code class="cpp plain">T&gt; </code><code class="cpp keyword bold">using</code> <code class="cpp plain">P = pair&lt;</code><code class="cpp color1 bold">int</code><code class="cpp plain">, T&gt;;</code></div></div></td></tr></tbody></table></div></div></blockquote>
Naively using the deduction guides from pair is not ideal because they
    cannot necessarily deduce objects of type <tt>P</tt> even
    from arguments that should obviously work, like <tt>P({}, 0)</tt>. However,
    let us apply the above procedure. The relevant deduction guide is
<blockquote><div><div id="highlighter_368505" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">class</code> <code class="cpp plain">A, </code><code class="cpp keyword bold">class</code> <code class="cpp plain">B&gt; pair(A, B) -&gt; pair&lt;A, B&gt;</code></div></div></td></tr></tbody></table></div></div></blockquote>
Deducing <code>(A, B)</code> from <code>(int, T)</code> yield <code>int</code> for <code>A</code>
    and <code>T</code> for <code>B</code>. Now substitute back into the deduction guide to get
    a new deduction guide
<blockquote><div><div id="highlighter_147664" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">class</code> <code class="cpp plain">T&gt; pair(</code><code class="cpp color1 bold">int</code><code class="cpp plain">, T) -&gt; pair&lt;</code><code class="cpp color1 bold">int</code><code class="cpp plain">, T&gt;;</code></div></div></td></tr></tbody></table></div></div></blockquote>
Deducing the template arguments for the alias template from this gives us the following deduction guide for the alias template 
<blockquote><div><div id="highlighter_362756" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code><code class="cpp plain">&lt;</code><code class="cpp keyword bold">class</code> <code class="cpp plain">T&gt; P(</code><code class="cpp color1 bold">int</code><code class="cpp plain">, T) -&gt; P&lt;T&gt;;</code></div></div></td></tr></tbody></table></div></div></blockquote>
A repository of additional expository materials and worked out examples used in the refinement of this algorithm 
    is maintained <a href="https://htmlpreview.github.io/?https://github.com/mspertus/CTAD_POST_CPP17/master/CTAD_alias_examples.html">online</a>.
<h2>Deducing from inherited constructors</h2>
In C++17, deduction guides (implicit and explicit) are not inherited when constructors are inherited.
	According to the C++ Core Guidelines <a href="https://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#Rc-inheriting">C.52</a>,
 you should “use inheriting constructors to import constructors into a 
derived class that does not need further explicit initialization”. As 
the creator of such a thin wrapper has not asked in any way for the 
derived class to behave differently under construction, our experience 
is that users are
surprised that construction behavior changes:
<table border="1"><tbody><tr><td colspan="2">
<div><div id="highlighter_36110" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp keyword bold">template</code> <code class="cpp plain">&lt;</code><code class="cpp keyword bold">typename</code> <code class="cpp plain">T&gt; </code><code class="cpp keyword bold">struct</code> <code class="cpp plain">CallObserver requires Invocable&lt;T&gt; { </code></div><div class="line number2 index1 alt1"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp plain">CallObserver(T &amp;&amp;) : t(std::forward&lt;T&gt;(t)) {}</code></div><div class="line number3 index2 alt2"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp keyword bold">virtual</code> <code class="cpp keyword bold">void</code> <code class="cpp plain">observeCall() { t(); }</code></div><div class="line number4 index3 alt1"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp plain">T t;</code></div><div class="line number5 index4 alt2"><code class="cpp plain">};</code></div><div class="line number6 index5 alt1">&nbsp;</div><div class="line number7 index6 alt2"><code class="cpp keyword bold">template</code> <code class="cpp plain">&lt;</code><code class="cpp keyword bold">typename</code> <code class="cpp plain">T&gt; </code><code class="cpp keyword bold">struct</code> <code class="cpp plain">CallLogger : </code><code class="cpp keyword bold">public</code> <code class="cpp plain">CallObserver&lt;T&gt; { </code></div><div class="line number8 index7 alt1"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp keyword bold">using</code> <code class="cpp plain">CallObserver&lt;T&gt;::CallObserver; </code></div><div class="line number9 index8 alt2"><code class="cpp spaces">&nbsp;&nbsp;</code><code class="cpp keyword bold">virtual</code> <code class="cpp keyword bold">void</code> <code class="cpp plain">observeCall() override { cout &lt;&lt; </code><code class="cpp string">"calling"</code><code class="cpp plain">; t();&nbsp; }</code></div><div class="line number10 index9 alt1"><code class="cpp plain">};</code></div></div></td></tr></tbody></table></div></div>
</td></tr>
<tr><th>C++17</th> <th>Proposed</th></tr>
<tr><td>
	<div><div id="highlighter_155720" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp plain">CallObserver observer([]() { </code><code class="cpp comments">/* ... */</code> <code class="cpp plain">}); </code><code class="cpp comments">// OK</code></div><div class="line number2 index1 alt1"><code class="cpp plain">CallLogger&lt;</code><code class="cpp comments">/*????*/</code><code class="cpp plain">&gt; logger([]() { </code><code class="cpp comments">/* ... */</code> <code class="cpp plain">});</code></div></div></td></tr></tbody></table></div></div>
</td>
<td><div><div id="highlighter_787208" class="syntaxhighlighter nogutter  cpp"><div class="toolbar"><span><a href="#" class="toolbar_item command_help help">?</a></span></div><table cellspacing="0" cellpadding="0" border="0"><tbody><tr><td class="code"><div class="container"><div class="line number1 index0 alt2"><code class="cpp plain">CallObserver observer([]() { </code><code class="cpp comments">/* ... */</code> <code class="cpp plain">}); </code><code class="cpp comments">// OK</code></div><div class="line number2 index1 alt1"><code class="cpp plain">CallLogger logger([]() { </code><code class="cpp comments">/* ... */</code> <code class="cpp plain">}); </code><code class="cpp comments">// Now OK</code></div></div></td></tr></tbody></table></div></div>
    </td></tr></tbody></table>
Note that inheriting the constructors of a base class must include 
inheriting all the deduction guides, not just the implicit ones. As a 
number of standard library
writers use explicit guides to behave “as-if” their classes were defined
 as in the standard, such internal implementation details
details would become visible if only the internal guides were inherited.
 We of course use the same algorithm
    for determining deduction guides for the base class template as 
described above for alias templates.
<h2>Wording</h2>
At the end over.match.class.deduct/1, add an addition bullet
<blockquote><ins><ul><li>If <tt>C</tt> is defined and satisfies the conditions for an aggregate class ([dcl.init.aggr]/1.1 - 1.4), not considering dependent base classes, let <em>e<sub>1</sub>, &hellip;, e<sub>n</sub></em> be the elements of <tt>C</tt> ([dcl.init.aggr]/2), not considering dependent base classes, and let <em>T<sub>1</sub>, &hellip;, T<sub>n</sub></em> be their respective declared types. If <tt>n > 0</tt>, an additional function template, called the aggregate deduction candidate, is derived as above from a hypothetical constructor <tt>C(T<sub>1</sub>, &hellip;, T<sub>n</sub>)</tt>. Additionally, for each integer 0 &le; i &lt; n, if the types <em>T<sub>i</sub>, &hellip;, T<sub>n</sub></em> are all default-constructible, the ith parameter of this function template shall have a default argument of the form {} .</li></ul></ins></blockquote>
Add a new subsubsection after over.match.class.deduct titled &ldquo;over.match.alias.deduct&rdquo;
<blockquote><ins>When resolving a placeholder for a deduced class type (9.1.7.5) where the
	<em>template-name</em> names an alias template <tt>A</tt> 
	representing a class template specialization of a class <tt>C</tt> with template arguments <tt><em>X<sub>j</sub></em></tt>, a
	set of functions and function templates are formed as follows. For each function or function
	template <tt>f</tt> formed for <tt>C</tt> by the process in 11.3.1.8, form a function or function
	template according to the following procedure:
	<ul><li>Form a hypothetical function or function template <tt>g</tt> with the following properties
		<ul><li>The template parameters, if
			any, are the template parameters of <tt>f</tt> (including default template arguments).</li>
			<li>The argument is an rvalue reference to the result
			type of <tt>f</tt></li>
		<li>The return type is void</li></ul>
		<li>Perform template argument deduction (12.9.2) on <tt>g(C&lt;X<sub>j</sub>...&gt;{})</tt> with the exception that deduction does
	not fail if not all template parameters of <tt>g</tt> are deduced.</li>
		<li>Form the desired function or function template as follows
		<ul><li>The parameters and return type are constructed by the substituting 
			 into 
			the parameters (including default arguments) and return type of <tt>f</tt>
			the deductions
			of all template parameters that were successfully deduced in the deduction of <tt>g</tt></li>
		<li>The template parameters consist of those template parameters of <tt>f</tt> or <tt>g</tt>
			that appear in the parameters and return type constructed in the previous step</li>
		<li>The constraints temp.constr.decl (12.4.2) are the conjunction of the constraints
			of <tt>f</tt> with a constraint that the template parameters of <tt>A</tt>
			are deducible (12.2.9.5) from the return type</li>
			<li>If <tt>f</tt> is the copy deduction candidate (11.3.1.8)
			or was generated from a <em>deduction-guide</em> (11.3.1.8)
				or was generated from a constructor or <em>deduction guide</em>
				with an <tt>explicit-specifier</tt>, then
			so is the function or function template being formed here</li></ul>
		</ul>
	Initialization and overload resolution are performed as described in 9.3 and 11.3.1.3, 11.3.1.4, or 11.3.1.7 (as
appropriate for the type of initialization performed) for an object of a hypothetical class type, where the
selected functions and function templates are considered to be the constructors of that class type for the
purpose of forming an overload set, and the initializer is provided by the context in which class template
argument deduction was performed. As an exception, the first phase in 11.3.1.7 (considering initializer-list
constructors) is omitted if the initializer list consists of a single expression of type <em>cv</em> <tt>U</tt>,
	where <tt>U</tt> is a specialization of <tt>C</tt> or a class derived from a specialization of <tt>C</tt>.
	If the function or 
	function template was generated from a constructor or <em>deduction-guide</em>
that had an <em>explicit-specifier</em>, each such notional constructor is considered to
	have that same <em>explicit-specifier</em>. All such notional constructors are considered 
	to be public members of the hypothetical class type. [<em>Note:</em> The requires clause ensures that
	a specialization of <tt>A</tt> can be deduced from the return type. <em>&mdash; end note</em>]
	<br/>
	[<em>Example:</em><pre>
	template &lt;class T, class U&gt; struct C {
	  C(T, U);
	};
	template&lt;class T, class U&gt;
	C(T, U) -&gt; C&lt;T, std::type_identity_t&lt;U&gt;&gt;;
	
	template&lt;class V&gt;
	using A = C&lt;V *, V *&gt;;
	
	// <em>Possible exposition only implementation of the above procedure</em>
	// <em>f is derived (11.3.1.8) from the deduction-guide of C</em>
	template&lt;class T, class U&gt; auto f(T, U) -&gt; C&lt;T, std::type_identity_t&lt;U&gt;&gt;;
	template&lt;class T, class U> void g(C&lt;T, std::type_identity_t&lt;U&gt;&gt;&amp;&amp;);
	// <em>Template argument deduction of g(C&lt;V *, V *&gt;{}) deduces T as V *</em>
	
	//<em>The following concept ensures a specialization of A is deduced</em>
	template&lt;T&gt; void t(A&lt;T&gt;&amp;&amp;);
	template&lt;T&gt; concept deduces_A = requires { t(std::declval&lt;T&gt;()); }

	// <em>Candidate is obtained by transforming f as described by the above procedure</em>
	template&lt;class V, class U&gt; 
	auto f(V *, U) -&gt; C&lt;V, std::type_identity_t&lt;U&gt;&gt;
	  requires deduces_A&lt;C&lt;V, std::type_identity_t&lt;U&gt;&gt;&gt;;
	// <em>End exposition only</em>
	int i{};
	double d{};
	A a1(&amp;i, &amp;i); // <em>A&lt;int&gt;</em>
	A a2(i, i); // <em>ill-formed - fails to match parameter list</em>
	A a3(&amp;i, &amp;d); // <em>ill-formed - cannot deduce alias template from C&lt;int *, double *&gt; </em>
	</pre>
	<em>&mdash; end example</em>]
		</ins></blockquote>
Insert a bullet after over.match.best/1.9 as follows
	<blockquote><ins>
	<ul><li><tt>F1</tt> is generated from class template argument deduction (over.match.class.deduct) for
		a class <tt>D</tt>, <tt>F2</tt> is generated from class template argument deduction for
		a base class <tt>B</tt> of <tt>D</tt>, and for all arguments the corresponding
		parameters of <tt>F1</tt> and <tt>F2</tt> have the same type, or if not that,</li></ul></ins></blockquote>

    <script type="text/javascript">
	SyntaxHighlighter.defaults['gutter'] = false;
     SyntaxHighlighter.all()
</script>
	</body></html>