harrison@software.ORG (Tim Harrison) (03/09/88)
I have submitted the following code to two validated compilers. The
designated line is rejected as illegal by one of the compilers. Is the
OTHERS choice actually illegal?
----------------------------------------------------------------------
package BOOLEAN_DATA is
type BOOLEAN_ARRAY is array (NATURAL range <>) of BOOLEAN;
procedure INITIALIZE(OBJECT : in out BOOLEAN_ARRAY);
end BOOLEAN_DATA;
package body BOOLEAN_DATA is
procedure INITIALIZE(OBJECT : in out BOOLEAN_ARRAY) is
begin
OBJECT := (others => FALSE); -- rejected by one compiler
end INITIALIZE;
end BOOLEAN_DATA;
----------------------------------------------------------------------
The compiler that rejects the code gives "RM 4.3.2(5): OTHERS illegal for
unconstrained array" as the reason for rejection. Here is what I believe to
be the basis for the decision to reject the OTHERS choice in this example.
RM 4.3.2(4-5) states:
...An *others* choice is only allowed if the aggregate appears in one (4)
of the following contexts (which defined the applicable index
constraint):
(a) The aggregate is ... the expression that follows an assignment (5)
compound delimiter. Moreover, the subtype of the corresponding
... object is a constrained array subtype.
What is the subtype of the object OBJECT? Can someone provide me with a
citation that specifies the subtype of OBJECT?
I think that the following citations are relevant, but they don't state
the case clearly enough for me to believe that they are providing me with a
definition for the subtype of OBJECT.
RM 6.2(9):
"...For a formal parameter of an unconstrained array type, the bounds
are obtained from the actual parameter, and the formal parameter is
constrained by these bounds (see 3.6.1)..."
RM 3.6.1(9):
"For a formal parameter of a subprogram or entry, the bounds are
obtained from the corresponding actual parameter. (The formal
parameter is constrained with the corresponding values of the
bounds)."
RM 6.2(9):
"...the formal parameter is unconstrained if and only if the mode is IN OUT
or OUT and the variable name given in the actual parameter denotes an
unconstrained variable"
The above citations indicate that the formal parameter does have bounds
and that it is constrained, but doesn't say that the effect is a constrained
subtype.
So, what is the subtype of OBJECT? Is it constrained or not?
-- Tim H.