[Bug fortran/112371] Wrong upper bound for the result of reduction intrinsics if the array is empty

[mikael at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112371

--- Comment #3 from Mikael Morin  ---
Possible culprit:
ifunction.m4 has this code:

  retarray->base_addr = xmallocarray (alloc_size, sizeof (rtype_name));
  if (alloc_size == 0)
{
  /* Make sure we have a zero-sized array.  */
  GFC_DIMENSION_SET(retarray->dim[0], 0, -1, 1);
  return;

}

[Bug fortran/112371] Wrong upper bound for the result of reduction intrinsics if the array is empty

[mikael at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112371

--- Comment #2 from Mikael Morin  ---
If dim == 3, the ubound and shape are (/ 9, 3, 7 /) as expected.
That is, the problem only arises if the resulting array is empty.

[Bug fortran/112371] Wrong upper bound for the result of reduction intrinsics if the array is empty

[mikael at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112371

--- Comment #1 from Mikael Morin  ---
(In reply to Mikael Morin from comment #0)
> i = 1
> (...)
> r = sum(a, dim=i)

If i is inlined, that is
  r = sum(a, dim=1)
the shape and ubound are (/ 3, 0, 7 /) as expected.
The difference is probably caused by inline vs library implementation.