https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122290
--- Comment #11 from Damian Rouson <damian at archaeologic dot codes> ---
Now this is truly mind-blowing. I'm using a patched version of gfortran that
Jerry built for me on a system we can both access. When I put the code from
this PR's original submission in a file named "pr.f90", the code compiles
without error using "gfortran -c pr.f90", but then when I perform a simple
string search and replacement via
sed -i'' -e "s/hyperparameters/tensor_map/g" pr.f90
the resulting code gives me an error matching the second of the two errors in
the original submission in this PR. See below. I even tried a couple of other
search-and-replacement options, e.g., using foo or foo_bar instead of
tensor_map, and those compiled without error. Something about "tensor_map" is
causing a specific problem. :O
$ cat pr.f90
module tensor_map_m
implicit none
type tensor_map_t(k)
integer, kind :: k = kind(1.)
real(k) :: learning_rate_ = real(1.5,k)
contains
generic :: operator(==) => default_real_equals
procedure default_real_equals
end type
interface
logical module function default_real_equals(lhs, rhs)
implicit none
class(tensor_map_t), intent(in) :: lhs, rhs
end function
end interface
end module
$ gfortran -c pr.f90
pr.f90:8:30:
8 | generic :: operator(==) => default_real_equals
| 1
Error: Entity ‘default_real_equals’ at (1) is already present in the interface
$ gfortran --version
GNU Fortran (GCC) 16.0.0 20251022 (experimental)
