To: J3                                                     J3/25-206
From: Malcolm Cohen
Subject: US25 complex/real association, edits.
Date: 2025-November-07
Reference: 24-129, 24-173, 24-188r1


1. Introduction

24-129 Feature request (use case)
24-173 Formal requirements
24-188r1 Specifications and syntax

This paper has an additional specification (syntax constraint),
and edits for the feature.

2. Modified specifications/syntax

(25S-5) When the pointer (on the LHS of the pointer assignment) is complex,
        the target shall be simply contiguous. This should be a Constraint.

However, when the pointer is real, having the target being rank one is also
not sufficient: consider
    real_pointer(1:10) => complex_target(1:20:2)
Here, there would be a gap in between every two consecutive (contiguous)
elements of the real array, which is not acceptable.

Therefore, 25S-5 is modified to

(25S-5) When the pointer and target are real and complex, or vice versa,
        the target shall be simply contiguous. This should be a Constraint.

And the syntax (new constraint) is modified to:

   "C1021a If data-target is of type real and data-pointer-object is of
           type complex, or vice-versa, data-target shall be simply
           contiguous (9.5.4). Otherwise, if bounds-remapping-list or
           upper-bounds-expr appears in a pointer-assignment-stmt,
           data-target shall be simply contiguous or of rank one."


3. Edits to 25-007r1

[xv] Introduction, Data usage and computation bullet, add sentence
   "A pointer of type real can be associated with a suitable complex
    target, and vice versa."
{Suitable mostly means "simply contiguous".}

[193:9+] 10.2.2.2 Syntax of the pointer assignment statement,
         insert new paragraph 1:
   "If a pointer-assignment-stmt has a bounds-remapping-list or an
    upper-bounds-expr, it is a \defn{rank-remapping} pointer assignment.".
{This will simplify the later wording. We're only going to use it inside
 this subclause, so not only would making it a defined term be unnecessary,
 it would be worse as it would stick the definition miles away from where
 we are using it. NB hyphenated because adjective.}
{Editor: Consider hyper-linking "rank-remapping" in the other edits to
         this paragraph.}

[193:12] 10.2.2.2 Syntax of the pointer assignment statement,
         first constraint C1016, beginning
            "(R1034) If data-target is not unlimited polymorphic..."
         change the first "If" to
   "If data-pointer-object is type real or complex, and the pointer
    assignment is rank-remapping, data-target shall be of type real or
    complex and the kind type parameters shall be equal. Otherwise, if".
Note: the rest of the constraint is:
                         "data-target is not unlimited polymorphic, data-
           pointer-object shall be type compatible (7.3.3) with it and the
           corresponding kind type parameters shall be equal."

[193:23+] Same subclause, before
   "C1022 If neither bounds-remapping-list nor upper-bounds-expr appears"
insert new constraint
   "C1021a If one of data-target is of type real and the other is of type
           complex, data-target shall be simply contiguous (9.5.4).
           Otherwise, in a rank-remapping pointer-assignment-stmt,
           data-target shall be simply contiguous or of rank one.".

[193:24] Same subclause, C1022,
         change "If neither bounds-remapping-list nor upper-bounds-expr
                 appears in a pointer-assignment-stmt"
         to "If a pointer-assignment-stmt is not rank-remapping,"
Note: the rest of the constraint is
   "the ranks of data-pointer-object and data-target shall be the same.".
{This is just an editorial improvement.}

NOTE TO J3: The next six edits describe in excruciating detail how para 8
            of this subclause will be changed. It may be easier to skip to
            the bottom where there are two paragraphs that just replace
            para 8 wholesale.

[195:15] Same subclause, paragraph 8, beginning "If bounds-remapping-list",
         delete "shall be simply contiguous (9.5.4) or of rank one, and".
{This is now a constraint.}

[195:17] Same subclause, same paragraph, before the sentence beginning
            "The number of elements..."
         insert new sentence:
   "The number of elements required by the pointer object is the number
    specified by the bounds-remapping-list, multiplied by two if the
    pointer object is of type complex and the target is of type real, or by
    half (and rounded up) if the pointer object is of type real and the
    target is of type complex.".

[197:17] Same subclause, same paragraph, sentence
   "The number of elements of the pointer target shall not be less than
    the number implied by the bounds-remapping-list."
change "implied by the bounds-remapping-list" to "required",
and split the paragraph in two after the full stop.

[197:17] Same subclause same paragraph 8 (but after the split so will be
         the new paragraph 9), next sentence,
         change "The elements of the pointer object are associated"
         to "A rank-remapping pointer assignment associates elements of the
             pointer object",
         and change "specified for the pointer object"
         to "required",
         making the whole sentence read:
   "A rank-remapping pointer assignment associates elements of the pointer
    object with those of the pointer target, in array element order; if the
    pointer target has more elements than required, the remaining elements
    are not associated with the pointer object.".

[195:19] Same subclause, same original paragraph 8 (now 9), append sentence
   "If the pointer object is of type real and the pointer target is of type
    complex, this association has two consecutive elements of the pointer
    object becoming associated with a single element of the pointer target,
    first the real part then the imaginary part; if the pointer object has
    an odd number of elements, the final element becomes associated with
    the real part of the final affected element of the pointer target."
{Semantics for real => complex.}

[195:19] Same subclause, same original paragraph 8 (now 9), append sentence
   "If the pointer object is of type complex and the pointer target is of
    type real, this association has each element of the pointer object
    becoming associated with two consecutive elements of the pointer
    target; the real part of the complex element becomes associated with
    the first of the consecutive real elements, and the imaginary part of
    the complex element becomes associated with the second of those
    consecutive elements."
{Semantics for complex => real.}

This makes the entire paragraphs 8 and (new) 9 read:

"
If bounds-remapping-list appears, it specifies the upper and lower bounds
of each dimension of the pointer, and thus the extents; the pointer target
shall not be a disassociated or undefined pointer. The number of elements
required by the pointer object is the number specified by the bounds-
remapping-list, multiplied by two if the pointer object is of type complex
and the target is of type real, or by half (and rounded up) if the pointer
object is of type real and the target is of type complex. The number of
elements of the pointer target shall not be less than the number required.

A rank-remapping pointer assignment associates elements of the pointer
object with those of the pointer target, in array element order; if the
pointer target has more elements than required, the remaining elements are
not associated with the pointer object. If the pointer object is of type
real and the pointer target is of type complex, this association has two
consecutive elements of the pointer object becoming associated with a
single element of the pointer target, first the real part then the
imaginary part; if the pointer object has an odd number of elements, the
final element becomes associated with the real part of the final affected
element of the pointer target. If the pointer object is of type complex and
the pointer target is of type real, this association has each element of
the pointer object becoming associated with two consecutive elements of the
pointer target; the real part of the complex element becomes associated
with the first of the consecutive real elements, and the imaginary part of
the complex element becomes associated with the second of those consecutive
elements.
"
{We could actually put the second paragraph after the existing p9 instead
 of in between p8 and p9, but this way works fine.}

[196:bottom-1] Insert at the end of NOTE 2:
   "A rank-remapping pointer assignment can be used to associated a REAL
    pointer with a simply contiguous COMPLEX target with the same kind type
    parameter, or vice versa. For example,

        COMPLEX(real64), TARGET :: cx(n)
        REAL(real64), TARGET :: rx(m)
        REAL(real64), POINTER :: r(:)
        COMPLEX(real64), POINTER :: c(:)
        ...\vdots
        c(1:m/2) => rx      ! c(1) associated with rx(1) and rx(2), etc.
        r(1:n*2) => cx      ! r(1:2) associated with cx(1), etc.
        c(1:m) => r         ! Invalid, because r is not simply contiguous."
{It's useful to draw attention to the invalid case, as when the data types
 are the same, rank-remapping pointer assignment is fine with discontiguous
 rank one targets.}

===END===