[Bug c++/11454] New: pedantic causes errors with exceptions, were warnings under 2.95

[Bug c++/11454] New: pedantic causes errors with exceptions, were warnings under 2.95

[kirkm at altera dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11454

           Summary: pedantic causes errors with exceptions, were warnings
                    under 2.95
           Product: gcc
           Version: 3.2
            Status: UNCONFIRMED
          Severity: critical
          Priority: P1
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: kirkm at altera dot com
                CC: gcc-bugs at gcc dot gnu dot org

Using -pedantic causes warnings under 2.95 to become errors under 3.2.  
Specifically, where tt.C is simply: 

int f(); 
int f() throw(); 

With gcc 2.95 
[picaso]:/il2/users/jeanb> g++ -pedantic -c tt.C 
tt.C:3: warning: declaration of `f()' throws different exceptions 
tt.C:1: warning: previous declaration here 

With gcc 3.2 
[goya]:/il2/users/jeanb> g++ -pedantic -c tt.C 
tt.C:3: declaration of `int f() throw ()' throws different exceptions 
tt.C:1: than previous declaration `int f()' 

This is causing significant headaches for our porting efforts from 2.95 to 3.2 
as 3rd party and OS headers often differ in the way they specify no exceptions 
(either without a throw clause or with a blank throw clause).

Thanks,
Kirk

[Bug c++/11454] pedantic causes errors with exceptions, were warnings under 2.95

[bangerth at dealii dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11454


bangerth at dealii dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


------- Additional Comments From bangerth at dealii dot org  2003-07-07 23:18 -------
This has nothing to do with -pedantic or not: present gcc mainline issues
an error unconditionally. The reason is simply that the language standard
says that these declarations are incompatible and the code is thus
invalid. This is, by the way, the same behavior that for example
Intel's icc shows.

The solution is and must be to leave the declaration of functions to
the system headers and not muck with it. Another possibility, if absolutely
necessary, is to include header files with -isystem instead of -I, since this
shuts down certain checks, such as this one.

W.