]> git.ipfire.org Git - thirdparty/gcc.git/commit
libstdc++: Fix up <ext/pointer.h> [PR121827]
authorJakub Jelinek <jakub@redhat.com>
Mon, 8 Sep 2025 09:49:58 +0000 (11:49 +0200)
committerJakub Jelinek <jakub@gcc.gnu.org>
Mon, 8 Sep 2025 09:49:58 +0000 (11:49 +0200)
commit592bafb26eb1fd50979f6cdf2176897c4a02c281
treeed06c00ed17572c2f830bb6940128052d1428f5c
parentf6ff531d907d5de5a47d6eca503715b32962c966
libstdc++: Fix up <ext/pointer.h> [PR121827]

During the tests mentioned in
https://gcc.gnu.org/pipermail/gcc-patches/2025-August/692482.html
(but dunno why I haven't noticed it back in August but only when testing
https://gcc.gnu.org/pipermail/gcc-patches/2025-September/694527.html )
I've noticed two ext header problems.
One is that #include <ext/pointer.h> got broken with the
r13-3037-g18f176d0b25591e28 change and since then is no longer
self-contained, as it includes iosfwd only if _GLIBCXX_HOSTED is defined
but doesn't actually include bits/c++config.h to make sure it is defined,
then includes a bunch of headers which do include bits/c++config.h and
finally uses in #if _GLIBCXX_HOSTED guarded code what is declared in iosfwd.
The other problem is that ext/cast.h is also not a self-contained header,
but that one has
/** @file ext/cast.h
 *  This is an internal header file, included by other library headers.
 *  Do not attempt to use it directly. @headername{ext/pointer.h}
 */
comment, so I think we just shouldn't include it in extc++.h and let
ext/pointer.h include it.

2025-09-08  Jakub Jelinek  <jakub@redhat.com>

PR libstdc++/121827
* include/precompiled/extc++.h: Don't include ext/cast.h which is an
internal header.
* include/ext/pointer.h: Include bits/c++config.h before
#if _GLIBCXX_HOSTED.
libstdc++-v3/include/ext/pointer.h
libstdc++-v3/include/precompiled/extc++.h