This is the mail archive of the mailing list for the libstdc++ project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] PR libstdc++/79162 ambiguity in string assignment due to string_view overload (LWG 2946)

On 20/09/17 18:59 +0100, Jonathan Wakely wrote:
On 20/09/17 17:52 +0100, Jonathan Wakely wrote:
On 20/09/17 16:36 +0100, Jonathan Wakely wrote:
On 04/09/17 16:48 +0100, Jonathan Wakely wrote:
On 30/07/17 15:01 +0200, Daniel Krügler wrote:
2017-07-28 22:40 GMT+02:00 Daniel Krügler <>:
2017-07-28 22:29 GMT+02:00 Daniel Krügler <>:
2017-07-28 22:25 GMT+02:00 Tim Song <>:
On Fri, Jul 28, 2017 at 4:10 PM, Daniel Krügler
<> wrote:
+      // Performs an implicit conversion from _Tp to __sv_type.
+      template<typename _Tp>
+        static __sv_type _S_to_string_view(const _Tp& __svt)
+        {
+          return __svt;
+        }

I might have gone for

+        static __sv_type _S_to_string_view(__sv_type __svt) noexcept
+        {
+          return __svt;
+        }

With that, we can also use noexcept(_S_to_string_view(__t)) to make up
for the absence of is_nothrow_convertible (basically the same thing I
did in LWG 2993's PR).

Agreed, that makes very much sense. I will adjust the P/R, but before
I resubmit I would like to get feedback whether the other two compare
functions also should become conditionally noexcept.

Locally I have now performed the sole change of the _S_to_string_view
declaration getting rid of the template, but would also like to gather
feedback from the maintainers whether I should also change the form of
the conditional noexcept to use the expression


instead of the current

is_same<_Tp, __sv_type>::value

as suggested by Tim Song.

I'm asking also, because I have a paper proposing to standardize
is_nothrow_convertible submitted for the upcoming C++ mailing - This
would be one of the first applications in the library ;-)

A slightly revised patch update: It replaces the _S_to_string_view
template by a simpler _S_to_string_view function as of Tim Song's
suggestion, but still uses the simplified noexcept specification
deferring it to a future application case for is_nothrow_convertible.
Furthermore now all three compare function templates are now
(conditionally) noexcept by an (off-list) suggestion from Jonathan

I've committed this, after some whitespace fixes and testing.


This change causes two regressions in C++17 mode, see

FAIL: 21_strings/basic_string/cons/char/ execution test
FAIL: 21_strings/basic_string/cons/wchar_t/ execution test

Here's a reduced version of that test, which passes in C++14 and fails
in C++17:

#include <string>
#include <assert.h>

class tstring : public std::string
tstring() : std::string() {}
tstring(tstring&& s) : std::string(std::move(s)) {}

int main()
tstring b;
tstring c(std::move(b));
assert( c.size() == 1 && c[0] == '1' );
assert( b.size() == 0 );

The second assertion fails, because this mem-initializer:

tstring(tstring&& s) : std::string(std::move(s)) {}

now prefers to use the new constructor:

basic_string(const _Tp& __t, const _Alloc& __a = _Alloc())

because tstring is convertible to string_view.

This turns a non-allocating move into an allocating copy.

This patch fixes the failure above, I'm testing it now.

--- a/libstdc++-v3/include/bits/basic_string.h
+++ b/libstdc++-v3/include/bits/basic_string.h
@@ -115,6 +115,7 @@ _GLIBCXX_BEGIN_NAMESPACE_CXX11
    template<typename _Tp, typename _Res>
     using _If_sv = enable_if_t<
       __and_<is_convertible<const _Tp&, __sv_type>,
+                __not_<is_convertible<const _Tp*, const basic_string*>>,
              __not_<is_convertible<const _Tp&, const _CharT*>>>::value,

I'm committing this to fix the regression. If the final LWG 2946
resolution does something different we can change it again.

Tested powerpc64le-linux, committed to trunk.

I'll start another test run with all variations ...

That test run revealed I forgot to make the same change to the _If_sv
helper in the old std::basic_string.

commit ab3a6ce803de3d21325bb82978f1f6d61c5cb4e2
Author: Jonathan Wakely <>
Date:   Wed Sep 20 22:48:04 2017 +0100

    PR libstdc++/79162 Fix std::string regression due to LWG 2946 (old ABI)
            PR libstdc++/79162
            * include/bits/basic_string.h [!_GLIBCXX_USE_CXX11_ABI]
            (basic_string::_If_sv): Remove from the overload set when the
            argument is derived from basic_string.

diff --git a/libstdc++-v3/include/bits/basic_string.h b/libstdc++-v3/include/bits/basic_string.h
index 8a382d5640c..a4b81137571 100644
--- a/libstdc++-v3/include/bits/basic_string.h
+++ b/libstdc++-v3/include/bits/basic_string.h
@@ -3439,6 +3439,7 @@ _GLIBCXX_END_NAMESPACE_CXX11
       template<typename _Tp, typename _Res>
 	using _If_sv = enable_if_t<
 	  __and_<is_convertible<const _Tp&, __sv_type>,
+		 __not_<is_convertible<const _Tp*, const basic_string*>>,
 		 __not_<is_convertible<const _Tp&, const _CharT*>>>::value,

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]