[PATCH] libstdc++: Document reamining C++17 implementation-defined behavior.

Jonathan Wakely jwakely@redhat.com
Fri Sep 5 14:47:16 GMT 2025


On Fri, 5 Sept 2025 at 15:17, Tomasz Kamiński <tkaminsk@redhat.com> wrote:
>
> This also covers bad_function_call::what from C++11.
>
> libstdc++-v3/ChangeLog:
>
>         * doc/html/manual/status.html: Regenerate.
>         * doc/xml/manual/status_cxx2011.xml: Add entry for bad_function_call.
>         * doc/xml/manual/status_cxx2017.xml: Add entires for bad_any_cast

"entries"

>         and nullptr_t output. Update entry for sf.cmath. Fix stable name for
>         mem.res.
> ---
> OK for trunk?

OK for trunk and gcc-13/14/15, thanks.

For the backports, if the changes to the HTML file don't apply cleanly
you can just regenerate the HTML and then git add them.


>
>  libstdc++-v3/doc/html/manual/status.html      | 18 ++++++++++++++----
>  .../doc/xml/manual/status_cxx2011.xml         |  5 +++++
>  .../doc/xml/manual/status_cxx2017.xml         | 19 +++++++++++++++----
>  3 files changed, 34 insertions(+), 8 deletions(-)
>
> diff --git a/libstdc++-v3/doc/html/manual/status.html b/libstdc++-v3/doc/html/manual/status.html
> index e1ed5bda8ee..564679bc6cf 100644
> --- a/libstdc++-v3/doc/html/manual/status.html
> +++ b/libstdc++-v3/doc/html/manual/status.html
> @@ -259,6 +259,9 @@ mainline GCC, not in any particular release.
>     </p><p>
>        <span class="emphasis"><em>20.7.2.0 [util.smartptr.weakptr]</em></span>
>        <code class="code">what()</code> returns <code class="literal">"bad_weak_ptr"</code>.
> +   </p><p>
> +      <span class="emphasis"><em>20.8.11.1 [func.wrap.badcall]</em></span>
> +      <code class="code">what()</code> returns <code class="literal">"bad_function_call"</code>.
>     </p><p>
>        <span class="emphasis"><em>20.8.9.1.3 [func.bind.place]/1</em></span> There are 29
>        placeholders defined and the placeholder types are
> @@ -1074,7 +1077,11 @@ since C++14 and the implementation is complete.
>        <code class="literal">"std::visit: variant is valueless"</code>,
>        or <code class="literal">"std::visit<R>: variant is valueless"</code>.
>     </p><p>
> -      <span class="emphasis"><em>23.12.5.2 [memory.resource.pool.options]</em></span>
> +      <span class="emphasis"><em>23.8.2 [any.bad_any_cast]</em></span>
> +      <code class="code">what()</code> returns <code class="literal">"bad any_cast"</code>.
> +   </p><p>
> +      <span class="emphasis"><em>23.12.5 [mem.res.pool.options]</em></span>,
> +      <span class="emphasis"><em>23.12.5 [mem.res.pool.mem]</em></span>
>        Let S equal <code class="code">numeric_limits<size_t>::digits</code>.
>        The limit for maximum number of blocks in a chunk is given by
>        <span class="mathphrase">2<sup>N</sup>-1</span>,
> @@ -1085,7 +1092,7 @@ since C++14 and the implementation is complete.
>        otherwise 3072 when <span class="mathphrase">S > 16</span>,
>        otherwise 768.
>     </p><p>
> -      <span class="emphasis"><em>23.12.6.1 [memory.resource.monotonic.buffer.ctor]</em></span>
> +      <span class="emphasis"><em>23.12.6.1 [mem.res.monotonic.buffer.ctor]</em></span>
>        The default <code class="code">next_buffer_size</code> is <code class="code">128 * sizeof(void*)</code>.
>        The default growth factor is <code class="code">1.5</code>.
>     </p><p>
> @@ -1120,8 +1127,11 @@ since C++14 and the implementation is complete.
>        races depends on the target C library that provides the function.
>     </p><p>
>        <span class="emphasis"><em>29.9.5 [sf.cmath]</em></span>
> -      The effect of calling the mathematical special functions with large
> -      inputs should be documented here.
> +      Calling the mathematical special functions with large inputs is supported
> +      but the results may be imprecise.
> +   </p><p>
> +      <span class="emphasis"><em>30.7.5.2.3 [ostream.inserters]</em></span>
> +      NTCTS for <code class="code">nullptr_t</code> is <code class="literal">"nullptr"</code>.
>     </p><p>
>        <span class="emphasis"><em>30.10.2.1 [fs.conform.9945]</em></span>,
>        <span class="emphasis"><em>30.10.2.2 [fs.conform.os]</em></span>
> diff --git a/libstdc++-v3/doc/xml/manual/status_cxx2011.xml b/libstdc++-v3/doc/xml/manual/status_cxx2011.xml
> index 17108e33af3..fbec6827647 100644
> --- a/libstdc++-v3/doc/xml/manual/status_cxx2011.xml
> +++ b/libstdc++-v3/doc/xml/manual/status_cxx2011.xml
> @@ -2613,6 +2613,11 @@ mainline GCC, not in any particular release.
>        <code>what()</code> returns <literal>"bad_weak_ptr"</literal>.
>     </para>
>
> +   <para>
> +      <emphasis>20.8.11.1 [func.wrap.badcall]</emphasis>
> +      <code>what()</code> returns <literal>"bad_function_call"</literal>.
> +   </para>
> +
>     <para>
>        <emphasis>20.8.9.1.3 [func.bind.place]/1</emphasis> There are 29
>        placeholders defined and the placeholder types are
> diff --git a/libstdc++-v3/doc/xml/manual/status_cxx2017.xml b/libstdc++-v3/doc/xml/manual/status_cxx2017.xml
> index 8ce714c889d..253aa4832aa 100644
> --- a/libstdc++-v3/doc/xml/manual/status_cxx2017.xml
> +++ b/libstdc++-v3/doc/xml/manual/status_cxx2017.xml
> @@ -2935,7 +2935,13 @@ since C++14 and the implementation is complete.
>     </para>
>
>     <para>
> -      <emphasis>23.12.5.2 [memory.resource.pool.options]</emphasis>
> +      <emphasis>23.8.2 [any.bad_any_cast]</emphasis>
> +      <code>what()</code> returns <literal>"bad any_cast"</literal>.
> +   </para>
> +
> +   <para>
> +      <emphasis>23.12.5 [mem.res.pool.options]</emphasis>,
> +      <emphasis>23.12.5 [mem.res.pool.mem]</emphasis>
>        Let S equal <code>numeric_limits<size_t>::digits</code>.
>        The limit for maximum number of blocks in a chunk is given by
>        <inlineequation><mathphrase>2<superscript>N</superscript>-1</mathphrase></inlineequation>,
> @@ -2948,7 +2954,7 @@ since C++14 and the implementation is complete.
>     </para>
>
>     <para>
> -      <emphasis>23.12.6.1 [memory.resource.monotonic.buffer.ctor]</emphasis>
> +      <emphasis>23.12.6.1 [mem.res.monotonic.buffer.ctor]</emphasis>
>        The default <code>next_buffer_size</code> is <code>128 * sizeof(void*)</code>.
>        The default growth factor is <code>1.5</code>.
>     </para>
> @@ -2998,8 +3004,13 @@ since C++14 and the implementation is complete.
>
>     <para>
>        <emphasis>29.9.5 [sf.cmath]</emphasis>
> -      The effect of calling the mathematical special functions with large
> -      inputs should be documented here.
> +      Calling the mathematical special functions with large inputs is supported
> +      but the results may be imprecise.
> +   </para>
> +
> +   <para>
> +      <emphasis>30.7.5.2.3 [ostream.inserters]</emphasis>
> +      NTCTS for <code>nullptr_t</code> is <literal>"nullptr"</literal>.
>     </para>
>
>     <para>
> --
> 2.51.0
>



More information about the Libstdc++ mailing list