To: J3                                                     J3/22-102
From: T. Burnus & R. Bader
Subject: Interp candidate: Lower bounds in sourced allocation
Date: 2021-December-13

Reference: 18-007r1

1. Question

Consider the following program:

program source_is_function_result
  implicit none
  integer, allocatable :: c(:)
  allocate(c, source=h(3))
  write(*,*) lbound(c,1), ubound(c,1)
contains
  pure function h(i) result(r)
   integer, value, intent(in) :: i
   integer, allocatable :: r(:)
   allocate(r(3:5))
   r = [1,2,3]
  end function h
end

What is the lower bound of c?

2. Discussion

9.7.1.2 of 18-007r1 specifies that
"When an ALLOCATE statement is executed for an array with no
allocate-shape-spec-list, the bounds of source-expr determine the
bounds of the array."

Looking at the definition of bounds in the standard, the general
idea appears to be that in this situation the lower bound should
be 1. However, for the above case of sourced allocation this is not
obvious. For all other situations where lower bounds are explicitly
needed (e.g., reallocation on intrinsic assignment, data
pointer assignment) the deduction is possible because the text
refers to LBOUND. That intrinsic's description produces the
desired answer.

3. Suggested resolution and edit:

It is suggested to modify the above specification to refer to LBOUND.

[134:27-28] replace sentence "When ... array." by
"When an ALLOCATE statement is executed for an array with no
allocate-shape-spec-list, the lower bound of each dimension of array
is the result of the intrinsic function LBOUND (16.9.109) applied
to the corresponding dimension of source-expr; the shape of array
is that of source-expr."

We have avoided a reference to UBOUND because that intrinsic also
would deliver a defined value for assumed-size objects appearing as a
source-expr, which surely should be disallowed.