<html>
<head>
<base href="https://bugs.llvm.org/">
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW - -Wctad-maybe-unsupported fires on move_iterator"
href="https://bugs.llvm.org/show_bug.cgi?id=44857">44857</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>-Wctad-maybe-unsupported fires on move_iterator
</td>
</tr>
<tr>
<th>Product</th>
<td>libc++
</td>
</tr>
<tr>
<th>Version</th>
<td>unspecified
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>All
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>enhancement
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>All Bugs
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedclangbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>arthur.j.odwyer@gmail.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvm-bugs@lists.llvm.org, mclow.lists@gmail.com
</td>
</tr></table>
<p>
<div>
<pre>// <a href="https://godbolt.org/z/acPvrB">https://godbolt.org/z/acPvrB</a>
When manually enabling `-Wctad-maybe-unsupported` on the command line:
<source>:10:17: warning: 'move_iterator' may not intend to support class
template argument deduction [-Wctad-maybe-unsupported]
std::sample(std::move_iterator(in.begin()), std::move_iterator(in.end()),
std::back_inserter(out),
^
/opt/compiler-explorer/clang-trunk-20200209/bin/../include/c++/v1/iterator:1187:28:
note: add a deduction guide to suppress this warning
class _LIBCPP_TEMPLATE_VIS move_iterator
^
My impression is that `move_iterator` is one of those classes that's pretty
much always safe to use with CTAD. I mean it's even safer than `vector`, unless
I'm missing something, because an "vector of vector" is a different beast from
a regular "vector", but a "move iterator of move iterator" is almost
indistinguishable from a "move iterator."
So it seems backwards that Clang would issue no warning on `vector{v}` but
issue a bogus warning on `move_iterator(it)`.
(I would still like to see a general-purpose `-Wctad` analogous to `-Wvla`, but
if `-Wctad-maybe-unsupported` is going to exist, then it shouldn't warn on
`move_iterator`. I also would not object to just removing
`-Wctad-maybe-unsupported`; I'm not aware that it warns on anything I care
about, and vice versa it fails to warn on all the STL things I do care about,
such as `vector` and `optional` and `pair`. I also noticed today that it
doesn't warn on `uniform_int_distribution`, and I'm not sure yet how I feel
about that.)</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>