| Message ID | 20200501131754.GA510077@redhat.com |
|---|---|
| State | New |
| Headers | show |
| Series | [committed] libstdc++: Replace reserved identifier _T with _Tp (PR 94901) | expand |
On Fri, May 01, 2020 at 02:17:54PM +0100, Jonathan Wakely via Gcc-patches wrote: > The libstdc++ manual documents that _T can not be used, because it's a > macro in system headers on some targets. > > PR libstdc++/94901 > * include/std/type_traits (__is_complete_or_unbounded): Replace > BADNAME _T with _Tp. > * testsuite/17_intro/badnames.cc: New test. > > Tested powerpc64le-linux and x86_64-linux, committed to master. > > This should be backported to gcc-10 too. Ok for 10 branch, thanks. > +#define _E9 _E9 is a BADNAME > + // .. > +#define _E24 _E24 is a BADNAME Just wonder about the above, IMHO it is worth spelling them all in between too, otherwise we don't catch _E10 or _E13 or _E23 in libstdc++ headers. Jakub
On 01/05/20 15:21 +0200, Jakub Jelinek via Libstdc++ wrote: >On Fri, May 01, 2020 at 02:17:54PM +0100, Jonathan Wakely via Gcc-patches wrote: >> The libstdc++ manual documents that _T can not be used, because it's a >> macro in system headers on some targets. >> >> PR libstdc++/94901 >> * include/std/type_traits (__is_complete_or_unbounded): Replace >> BADNAME _T with _Tp. >> * testsuite/17_intro/badnames.cc: New test. >> >> Tested powerpc64le-linux and x86_64-linux, committed to master. >> >> This should be backported to gcc-10 too. > >Ok for 10 branch, thanks. > >> +#define _E9 _E9 is a BADNAME >> + // .. >> +#define _E24 _E24 is a BADNAME > >Just wonder about the above, IMHO it is worth spelling them all in between >too, otherwise we don't catch _E10 or _E13 or _E23 in libstdc++ headers. OK, can do. I decided it was pretty unlikely we'd have _E10 without hitting one of the earlier ones, but I guess it's possible somebody would define _E12 to mean 10e12 or something. I'll add that to master and then backport the enhanced version.
On 01/05/20 14:45 +0100, Jonathan Wakely wrote: >On 01/05/20 15:21 +0200, Jakub Jelinek via Libstdc++ wrote: >>On Fri, May 01, 2020 at 02:17:54PM +0100, Jonathan Wakely via Gcc-patches wrote: >>>The libstdc++ manual documents that _T can not be used, because it's a >>>macro in system headers on some targets. >>> >>> PR libstdc++/94901 >>> * include/std/type_traits (__is_complete_or_unbounded): Replace >>> BADNAME _T with _Tp. >>> * testsuite/17_intro/badnames.cc: New test. >>> >>>Tested powerpc64le-linux and x86_64-linux, committed to master. >>> >>>This should be backported to gcc-10 too. >> >>Ok for 10 branch, thanks. >> >>>+#define _E9 _E9 is a BADNAME >>>+ // .. >>>+#define _E24 _E24 is a BADNAME >> >>Just wonder about the above, IMHO it is worth spelling them all in between >>too, otherwise we don't catch _E10 or _E13 or _E23 in libstdc++ headers. > >OK, can do. > >I decided it was pretty unlikely we'd have _E10 without hitting one of >the earlier ones, but I guess it's possible somebody would define _E12 >to mean 10e12 or something. > >I'll add that to master and then backport the enhanced version. Here's the extra bit. Committed to master.
diff --git a/libstdc++-v3/include/std/type_traits b/libstdc++-v3/include/std/type_traits index 75a98e38c0a..bc9a45b3746 100644 --- a/libstdc++-v3/include/std/type_traits +++ b/libstdc++-v3/include/std/type_traits @@ -197,8 +197,8 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION // Helper functions that return false_type for incomplete classes, // incomplete unions and arrays of known bound from those. - template <typename _T, size_t = sizeof(_T)> - constexpr true_type __is_complete_or_unbounded(__type_identity<_T>) + template <typename _Tp, size_t = sizeof(_Tp)> + constexpr true_type __is_complete_or_unbounded(__type_identity<_Tp>) { return {}; } template <typename _TypeIdentity, diff --git a/libstdc++-v3/testsuite/17_intro/badnames.cc b/libstdc++-v3/testsuite/17_intro/badnames.cc new file mode 100644 index 00000000000..aabd01255e2 --- /dev/null +++ b/libstdc++-v3/testsuite/17_intro/badnames.cc @@ -0,0 +1,72 @@ +// Copyright (C) 2020 Free Software Foundation, Inc. +// +// This file is part of the GNU ISO C++ Library. This library is free +// software; you can redistribute it and/or modify it under the +// terms of the GNU General Public License as published by the +// Free Software Foundation; either version 3, or (at your option) +// any later version. + +// This library is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +// GNU General Public License for more details. + +// You should have received a copy of the GNU General Public License along +// with this library; see the file COPYING3. If not see +// <http://www.gnu.org/licenses/>. + +// { dg-do compile { target x86_64-*-linux* } } + +// Names taken from coding_style.bad_identifiers in the libstdc++ manual. +// We can't test this on all targets, because these names are used in +// non-GCC system headers. Test on x86_64-linux where we know these particular +// names aren't used. + +// For Solaris: +#define _B _B is a BADNAME +#define _C _C is a BADNAME +#define _L _L is a BADNAME +#define _N _N is a BADNAME +#define _P _P is a BADNAME +#define _S _S is a BADNAME +#define _U _U is a BADNAME +#define _X _X is a BADNAME +#define _E1 _E1 is a BADNAME +#define _E2 _E2 is a BADNAME +#define _E3 _E3 is a BADNAME +#define _E4 _E4 is a BADNAME +#define _E5 _E5 is a BADNAME +#define _E6 _E6 is a BADNAME +#define _E7 _E7 is a BADNAME +#define _E8 _E8 is a BADNAME +#define _E9 _E9 is a BADNAME + // .. +#define _E24 _E24 is a BADNAME + +// Irix adds: +#define _A _A is a BADNAME +#define _G _G is a BADNAME + +// MS adds: +#define _T _T is a BADNAME + +// BSD adds: +#define __used __used is a BADNAME +#define __unused __unused is a BADNAME + // __inline (glibc uses this so can't test here) + // _Complex (glibc uses this so can't test here) +#define __istype __istype is a BADNAME +#define __maskrune __maskrune is a BADNAME +#define __tolower __tolower is a BADNAME +#define __toupper __toupper is a BADNAME +#define __wchar_t __wchar_t is a BADNAME +#define __wint_t __wint_t is a BADNAME + // _res + // _res_ext + // __tg_* + +// VxWorks adds: +#define _C2 _C2 is a BADNAME + + +#include <bits/stdc++.h>
The libstdc++ manual documents that _T can not be used, because it's a macro in system headers on some targets. PR libstdc++/94901 * include/std/type_traits (__is_complete_or_unbounded): Replace BADNAME _T with _Tp. * testsuite/17_intro/badnames.cc: New test. Tested powerpc64le-linux and x86_64-linux, committed to master. This should be backported to gcc-10 too. commit 070b4df8a06539b661c134d436ea4207099fdcf0 Author: Jonathan Wakely <jwakely@redhat.com> Date: Fri May 1 13:13:39 2020 +0100 libstdc++: Replace reserved identifier _T with _Tp (PR 94901) The libstdc++ manual documents that _T can not be used, because it's a macro in system headers on some targets. PR libstdc++/94901 * include/std/type_traits (__is_complete_or_unbounded): Replace BADNAME _T with _Tp. * testsuite/17_intro/badnames.cc: New test.