Re: c++/5050: template instantiation depth

Re: c++/5050: template instantiation depth

[Wolfgang Bangerth]

The following reply was made to PR c++/5050; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: gcc-bugs@gcc.gnu.org, <gcc-gnats@gcc.gnu.org>
Cc: georg.wild@gmx.de, <neroden@twcny.rr.com>
Subject: Re: c++/5050: template instantiation depth
Date: Wed, 22 Jan 2003 13:11:12 -0600 (CST)

 This report is about the following code (extracted by Nathanael):
 --------------
 struct ostream  {
    template<class T> ostream& foo( const T & ) 
     { return *this;  }
  };
  
  void foo()  {
    ostream os;
    (os.foo(1)).foo(2);
  }
 -----------------
 It compiles fine without optimization and -ftemplate-depth-1, since 
 ostream::foo really needs to be instantiated only once. It fails with 
 optimization -O saying that the maximal instantiation depth is exceeded.
 
 I think that this is actually a bug in the inliner, since the bug can be 
 made to go away when one specifies "-O -fno-inline". My naive 
 understanding is that the inliner just copies the body of the foo function 
 into the calling place, so the function is not instantiated, although the 
 instantiation counter is bumped up by one. When we get to the second call 
 of foo(), we check that the function has not been instantiated, so needs 
 to be instantiated now, but then the instantiation counter is already at 
 1, and we fail.
 
 The solution would probably be to not increase the template instantiation 
 depth counter when a function is inlined, rather than instantiated.
 
 Regards
   Wolfgang
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth             email:            bangerth@ticam.utexas.edu
                               www: http://www.ticam.utexas.edu/~bangerth/