[Bug fortran/53718] [4.7/4.8 regression] [OOP] gfortran generates asm label twice in the same output file

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53718

--- Comment #3 from janus at gcc dot gnu.org 2012-06-18 20:04:09 UTC ---
(In reply to comment #2)
> just as a side note: most of my bugreports come from compiling babel's
> regression test suite. Babel is a language interoperability tool that (also)
> generates bindings for Fortran 2003. It's free (LGPL) and it might be a very
> good testcase for gfortran in the future.
> 
> https://computation.llnl.gov/casc/components/

Thanks for the pointer. Sounds interesting.


Btw, here is a slightly different (but much simpler) test case, which exhibits
the same problem:


module m
  type t
  end type
end module

subroutine sub1
  use m
  class(t), pointer :: a1
end subroutine

subroutine sub2
  use m
  class(t), pointer :: a2
end subroutine


This structure is rather unusual for F03 object-oriented code, since it
involves 'naked' subroutines (F77 style), which are not part of any MODULE or
PROGRAM. Putting both routines into a module or program makes the error go
away.