New testsuite failures...

[François-Xavier Coudert]

> Mine:
> * gfortran.dg/default_format_1.f90 is failing on powerpc-darwin
> (probably a library problem)

This is due to a problem in the output of denormal numbers. I'll xfail
both this testcase and gfortran.dg/default_format_2.f90 on
powerpc-apple-darwin.

> Paul's:
> * gfortran.dg/derived_comp_array_ref_5.f90 is failing on
> powerpc-darwin (failing test for excess error, thus not a library
> issue)

I checked and can reproduce that on x86_64-linux by using -m32. This
does not appear to be a regression, and the reduced testcase (filed as
PR33636) is:

$ cat x.f90
  implicit none
  type vec3
    integer, dimension(1) :: coords
  end type vec3
  type(vec3), parameter :: v1 = vec3((/ 0 /))
  integer :: i

  i = 1
  print *, v1%coords ((/i/))
  end
$ gfortran -m32 x.f90
x.f90:9.23:

  print *, v1%coords ((/i/))
                     1
Error: index in dimension 1 is out of bounds at (1)


FX