[Bug c++/49198] New: GCC trunk and 4.6 generate spurious "has incomplete type" errors

[Bug c++/49198] New: GCC trunk and 4.6 generate spurious "has incomplete type" errors

[aaw at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49198

           Summary: GCC trunk and 4.6 generate spurious "has incomplete
                    type" errors
           Product: gcc
           Version: 4.6.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: aaw@gcc.gnu.org


Created attachment 24376
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=24376
Minimal reproduction

See the attached tst.cc file.  When compiled with GCC trunk and 4.6, it
generates the following errors:

tst.cc: In instantiation of ‘S<IC>’:
tst.cc:14:7:   instantiated from here
tst.cc:3:32: error: ‘S<T>::t’ has incomplete type
tst.cc:1:7: error: forward declaration of ‘struct IC’

However, the code is structured such that IC need not be complete.  According
to the C++ standard (14.7.1):

"Unless a class template specialization has been explicitly instantiated
(14.7.2) or explicitly specialized (14.7.3), the class template specialization
is implicitly instantiated when the specialization is referenced in a context
that requires a completely-defined object type or when the completeness of the
class type affects the semantics of the program."

In this case, we only refer to S<IC>* (pointer), which does not require a
complete type.  Therefore, the class template specialization S<IC> should not
be instantiated, and IC need not be complete.

Various trivial modifications to this program (e.g. replacing S<IC>* with IC*
or replacing FT(F) with F) do not generate an error.  Nor does GCC 4.4.3.

[Bug c++/49198] [4.5/4.6/4.7 Regression] GCC generates spurious "has incomplete type" errors

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49198

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.05.27 19:06:43
                 CC|                            |jason at gcc dot gnu.org
      Known to work|                            |4.4.6
            Summary|GCC trunk and 4.6 generate  |[4.5/4.6/4.7 Regression]
                   |spurious "has incomplete    |GCC generates spurious "has
                   |type" errors                |incomplete type" errors
     Ever Confirmed|0                           |1
      Known to fail|                            |4.6.1, 4.7.0

--- Comment #1 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-05-27 19:06:43 UTC ---
Let's ask Jason to have a look. To be safe I'm marking this as Regression.

[Bug c++/49198] GCC generates spurious "has incomplete type" errors

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49198

Jason Merrill <jason at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID
            Summary|[4.5/4.6/4.7 Regression]    |GCC generates spurious "has
                   |GCC generates spurious "has |incomplete type" errors
                   |incomplete type" errors     |

--- Comment #2 from Jason Merrill <jason at gcc dot gnu.org> 2011-05-27 21:05:25 UTC ---
The instantiation is being triggered by argument-dependent lookup.  When we see
FT(F), the associated classes for the argument F are the associated classes for
S<IC>*, or S<IC>, so we instantiate S<IC> to get any friend declarations.

If you change the call to ::FT(F), no argument-dependent lookup is done, so the
instantiation is not triggered.

[Bug c++/49198] GCC generates spurious "has incomplete type" errors

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49198

--- Comment #3 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-05-27 21:15:31 UTC ---
Thanks for the clarification. Indeed, Comeau online rejects it, current ICC,
even in strict-ansi mode, doesn't however (also using EDG, AFAIK), weird.

[Bug c++/49198] GCC generates spurious "has incomplete type" errors

[aaw at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49198

--- Comment #4 from Ollie Wild <aaw at gcc dot gnu.org> 2011-05-31 18:13:02 UTC ---
Thanks for the clarification, Jason.

I would not have figured that out on my own.