[Bug libstdc++/16844] New: Explicit instantiation failing

[Bug libstdc++/16844] New: Explicit instantiation failing

[larsbj at lyx dot org]

This is adapted from code in connection.hpp in boost.

Should it work or is the code in fact wrong, and makes assumptions
that the standard does not provide?

------
#include <list>

struct bound_object {
        bool operator==(const bound_object& other) const;
        bool operator<(const bound_object& other) const;
};

template class std::list<bound_object>;
--------

g++ (GCC) 3.5.0 20040731 (experimental)

g++ -c -o alloc.o alloc.C
/opt/gcc-head/lib/gcc/i686-pc-linux-gnu/3.5.0/../../../../include/c++/3.5.0/bits/stl_list.h:
In member function `std::_List_node<_Tp>* std::list<_Tp,
_Alloc>::_M_create_node() [with _Tp = bound_object, _Alloc =
std::allocator<bound_object>]':
alloc.C:8:   instantiated from here
/opt/gcc-head/lib/gcc/i686-pc-linux-gnu/3.5.0/../../../../include/c++/3.5.0/bits/stl_list.h:460:
error: no matching function for call to
`std::allocator<bound_object>::construct(bound_object*)'
/opt/gcc-head/lib/gcc/i686-pc-linux-gnu/3.5.0/../../../../include/c++/3.5.0/ext/mt_allocator.h:106:
note: candidates are: void __gnu_cxx::__mt_alloc<_Tp>::construct(_Tp*, const
_Tp&) [with _Tp = bound_object]

-- 
           Summary: Explicit instantiation failing
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: libstdc++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: larsbj at lyx dot org
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] Explicit instantiation failing

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-07-31 23:35 -------
I can't see any reason why this shouldn't work. And indeed it does 
with my mainline snapshot from 2004-07-27 and all the releases 
I have. Can someone confirm with a newer snapshot? 
 
W. 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-01 00:16 -------
Well it looks like a regression caused by:
2004-07-28  Matt Austern  <austern@apple.com>

        .....
        * include/ext/hashtable.h: Use rebind so that allocator_type
        has correct type for a container's allocator.  Replace use of
        single-argument _Construct and _Destroy with use of allocator's
        construct and destroy methods.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |austern at apple dot com
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |rejects-valid
      Known to fail|                            |3.5.0
      Known to work|                            |3.4.0
   Last reconfirmed|0000-00-00 00:00:00         |2004-08-01 00:16:55
               date|                            |
            Summary|Explicit instantiation      |[3.5 Regression] Explicit
                   |failing                     |instantiation failing
   Target Milestone|---                         |3.5.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[austern at apple dot com]

------- Additional Comments From austern at apple dot com  2004-08-01 06:12 -------
I agree that this is almost certainly caused by my change.  This should be assigned to me.  I expect that 
the fix will be easy.  (The most interesting part of the fix will be coming up with a new test case, since 
this wasn't caught by anything in the existing test suite.)

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[pcarlini at suse dot de]

------- Additional Comments From pcarlini at suse dot de  2004-08-01 10:21 -------
Isn't just

diff -urN libstdc++-v3-orig/include/bits/stl_list.h
libstdc++-v3/include/bits/stl_list.h
--- libstdc++-v3-orig/include/bits/stl_list.h   2004-07-28 18:37:17.000000000 +0200
+++ libstdc++-v3/include/bits/stl_list.h        2004-08-01 12:10:20.000000000 +0200
@@ -457,7 +457,7 @@
        _Node* __p = this->_M_get_node();
        try
          {
-           this->get_allocator().construct(&__p->_M_data);
+           this->get_allocator().construct(&__p->_M_data, value_type());
          }
        catch(...)
          {

???

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |austern at apple dot com
                   |dot org                     |
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[larsbj at lyx dot org]

------- Additional Comments From larsbj at lyx dot org  2004-08-01 11:37 -------
The patch in #4 fixes the compile.
I have no idea if this is the correct fix.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[austern at apple dot com]

------- Additional Comments From austern at apple dot com  2004-08-01 22:53 -------
That's a correct fix, but it's not the best fix.

This is a bug in an unused private member function.  (For history buffs: it was once an auxiliary 
member function for an SGI extension.  We removed the extension itself, but neglected to get around to 
removing the auxiliary that it used.)  So the best fix is to remove it.

And, of course: the reason we are seeing a bug in an unused member function is that we're doing an 
explicit instantiation.  Conclusion: our test suite needs to do explicit instantiations, not just tests of 
correct functionality.

I'm testing a patch now, including the new test cases.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-08-02 04:40 -------
Subject: Bug 16844

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	austern@gcc.gnu.org	2004-08-02 04:40:20

Modified files:
	libstdc++-v3   : ChangeLog 
	libstdc++-v3/include/bits: stl_list.h 
	libstdc++-v3/include/ext: slist 
Added files:
	libstdc++-v3/testsuite/23_containers/deque: 
	                                            explicit_instantiation.cc 
	libstdc++-v3/testsuite/23_containers/list: 
	                                           explicit_instantiation.cc 
	libstdc++-v3/testsuite/23_containers/map: 
	                                          explicit_instantiation.cc 
	libstdc++-v3/testsuite/23_containers/multimap: 
	                                               explicit_instantiation.cc 
	libstdc++-v3/testsuite/23_containers/multiset: 
	                                               explicit_instantiation.cc 
	libstdc++-v3/testsuite/23_containers/set: 
	                                          explicit_instantiation.cc 
	libstdc++-v3/testsuite/23_containers/vector: 
	                                             explicit_instantiation.cc 
	libstdc++-v3/testsuite/ext: hash_set_explicit_instantiation.cc 
	                            slist_explicit_instantiation.cc 

Log message:
	PR libstdc++/16844
	* include/bits/stl_list.h (_M_create_node): Remove unused
	zero-argument version.
	* include/ext/slist (_M_create_node): Pass two arguments to
	allocator's construct() member function.
	* testsuite/23_containers/deque/explicit_instantiation.cc: New.
	* testsuite/23_containers/list/explicit_instantiation.cc: New.
	* testsuite/23_containers/vector/explicit_instantiation.cc: New.
	* testsuite/23_containers/map/explicit_instantiation.cc: New.
	* testsuite/23_containers/set/explicit_instantiation.cc: New.
	* testsuite/23_containers/multimap/explicit_instantiation.cc: New.
	* testsuite/23_containers/multiset/explicit_instantiation.cc: New.
	* testsuite/ext/hash_set_explicit_instantiation.cc: New.
	* testsuite/ext/slist_explicit_instantiation.cc: New.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/ChangeLog.diff?cvsroot=gcc&r1=1.2594&r2=1.2595
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/include/bits/stl_list.h.diff?cvsroot=gcc&r1=1.42&r2=1.43
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/include/ext/slist.diff?cvsroot=gcc&r1=1.23&r2=1.24
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/deque/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/list/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/map/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/multimap/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/multiset/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/set/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/23_containers/vector/explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/ext/hash_set_explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/ext/slist_explicit_instantiation.cc.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844

[Bug libstdc++/16844] [3.5 Regression] Explicit instantiation failing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-02 04:56 -------
Fixed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16844