Re: Which compiler is correct?

[woodruff@tanana.llnl.gov (John Woodruff)]

>>>>> "bobduff" == Robert A Duff <bobduff@world.std.com> writes:
In article <EBz083.9EB@world.std.com> bobduff@world.std.com (Robert A Duff) writes:


    > In article <33A77C54.5484@bix.com>, Tom Moran <tmoran@bix.com>
    > wrote:
    >> Compilers G and J accept this code.  Compiler O gives the error
    >> message indicted in the comment, on said line.  Which is correct
    >> Ada?
    (deletia)

    > The rules for this stuff are somewhat arcane.  See 3.4 and 7.3.1.
    (tacit agreement, followed by a new case:)

Here is another inquiry of the type "which compiler is correct." It is
so similar to the issue Bob answered for Tom that I retain the thread
(and the answer Bob gave is so clear that I imagine that this one will
be equally easy to understand once I know the answer.)

Let me define a package with a tagged private type:
package Parent is
   type Obj is tagged private ;
private
   type Obj is tagged record
      I : Integer ;
   end record ;
end Parent ;

The package has just one child that extends the type.  Depending on the
specifics of how I derive the new type, I get correct code or code that
evokes an error message that I don't understand:

package Parent.Child is
   type Good_Obj is new Parent.Obj with private ;
   type Bad_Obj  is new Parent.Obj with null record ;
   procedure Do_Nothing ;  -- just to require body
private
   type Good_Obj is new Parent.Obj with null record ;
end ;

One of the compilers (compiler "r") is able to compile the body without
error; the other compiler reports an error in the body when I try to
reference the component of the parent type.

package body Parent.Child is
   good : Good_Obj ;
   Bad  : Bad_Obj ;
   procedure Do_Nothing is
   begin
      null;
   end ;
begin
   Good.I := 2 ;
   Bad.I := 2 ;  -- compiler "g" reports error
end ;

According to one compiler the component of type Bad_Obj is an "undefined
selector".  I am unable to think of a reason why the body of the child
package can reference the parent's private declaration of Obj for one,
but not for the other derived types.

--
John Woodruff	                                          N I F   \ ^ /
Lawrence Livermore National Lab                         =====---- < 0 >
510 422 4661                                                      / v \