[Bug c++/39055] [4.3/4.4/4.5 regression] ICE with questionable default parameter of a member function

["reichelt at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org>]

------- Comment #6 from reichelt at gcc dot gnu dot org  2009-07-18 11:26 -------
GCC's history with this code is:

* GCC 2.95.3          rejects
* GCC 3.0   - 3.2.3   accepts
* GCC 3.3   - 3.3.6   rejects
* GCC 3.4.0 - 4.3.x   crashes
* GCC 4.4.0 - trunk   accepts

Without the template, only GCC 2.93.3 rejects the code, all other
versions accept it. Considering that also Intel and Comeau accept
the code (template and non-template version) I doubt that the code
is invalid.

Jason, you said:

> I was wrong; the previous paragraph says "Local variables shall not
> be used in default argument expressions."
> 3.3 was right to reject the testcase; 

IMHO "Local variables" refers to function local variables, so that
you were right in the first place (comment #3) and that the code is
indeed valid.

This would mean that we really can close the bug as fixed in GCC 4.4.0.


-- 

reichelt at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jason at gcc dot gnu dot org
           Keywords|                            |accepts-invalid
      Known to work|3.2.3                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39055