This is the mail archive of the libstdc++@gcc.gnu.org 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] PR78991 make __gnu_cxx::__ops constructors explicit


On 09/01/17 11:05 +0000, Kyrill Tkachov wrote:

On 09/01/17 10:47, Jonathan Wakely wrote:
On 09/01/17 10:39 +0000, Kyrill Tkachov wrote:
Hi Jonathan,

On 06/01/17 12:40, Jonathan Wakely wrote:
This solves a problem when using libstdc++ with Clang, due to Clang
more eagerly instantiating constexpr function templates during
argument deduction. G++ has some shortcuts to avoid this problem, but
Clang doesn't, and it's not clear that it's strictly speaking a bug in
Clang or if it's following the standard. By making these constructors
explicit we stop them being considered by overload resolution for
copying these functors, which stops us ending up back in the
std::function SFINAE checks.

I'm also using _GLIBCXX_MOVE to turn some internal copies into moves,
because otherwise using something like std::function with <algorithm>
results in a number of potentially expensive copies.

  PR libstdc++/78991
  * include/bits/predefined_ops.h (_Iter_comp_iter, _Iter_comp_val)
  (_Val_comp_iter, _Iter_equals_val, _Iter_pred, _Iter_comp_to_val)
  (_Iter_comp_to_iter, _Iter_negate): Make constructors explicit and
  move function objects.
  (__iter_comp_iter, __iter_comp_val, __val_comp_iter, __pred_iter)
  (__iter_comp_val, __iter_comp_iter, __negate): Move function objects.
  * testsuite/25_algorithms/sort/78991.cc: New test.

Tested powerpc64le-linux, committed to trunk.

I'll backport  the 'explicit' constructors (but not the _GLIBCXX_MOVE
changes) to the branches too.


I see this test fail on the GCC 5 branch on arm and aarch64 (error message pasted below).
Does the test need a gnu++11 guard or something on the branch?

I thought I'd changed that before committing, I'll fix it.

Thanks.
Also, I think 30_threads/thread/cons/lwg2097.cc needs something similar.

Yes. I fixed both those tests on my testing machine, but not on the
one where I committed from.

Committed to gcc-5-branch.

commit ea8019b493c1aa6b223969ec5f1153e6f381acde
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Mon Jan 9 11:06:56 2017 +0000

    Add missing dg-options to C++11 test.
    
    	* testsuite/30_threads/thread/cons/lwg2097.cc: Compile with
    	-std=gnu++11.

diff --git a/libstdc++-v3/testsuite/30_threads/thread/cons/lwg2097.cc b/libstdc++-v3/testsuite/30_threads/thread/cons/lwg2097.cc
index 3ec4325..d5d6288 100644
--- a/libstdc++-v3/testsuite/30_threads/thread/cons/lwg2097.cc
+++ b/libstdc++-v3/testsuite/30_threads/thread/cons/lwg2097.cc
@@ -1,3 +1,4 @@
+// { dg-options "-std=gnu++11" }
 // { dg-do compile }
 // { dg-require-cstdint "" }
 // { dg-require-gthreads "" }

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