Re: Better control of assignment

[Craig Carey <research@ijs.co.nz>]

On Sat, 10 Nov 2001 22:27:25 GMT, Jeffrey Carter <jrcarter@acm.org>
wrote:

>s only invoked for a copy (including parameter passing
>> by copy (if selected)) from one variable object of type T to another. If
>> this does not deliver enough control, you must make the type private. For
>> example:
>> 
>>    type Odd_Counter is new Integer;
>>    procedure Assign_O


The Oberon-2 language implements a feature of read-only record field
components.

Also it allows package spec variables (Module identifiers) to be
exported read-only.


A page on Oberon-2, that mentions the "read-only" "export mark"
feature of Oberon-2, is here:

http://www-cse.ucsd.edu/classes/sp01/cse131B_A/oberon2.htm

|
|[Module] Identifiers [variables] marked with " - " in their declaration
| are read-only in importing modules. 


|  Qualident = [ident "."]ident.
|  IdentDef  = ident [" * " | " - "].

...

|If a [an array] or r [a record or a component record in a record] are
| read-only, then also a[e] and r.f are read-only.
...

|MODULE Trees;  (* exports: Tree, Node, Insert, Search, Write, Init *)
|  IMPORT Texts, Oberon;  (* exports read-only: Node.name *)
|
|  TYPE
|    Tree* = POINTER TO Node;
|    Node* = RECORD
|      name-: POINTER TO ARRAY OF CHAR;
|      left, right: Tree
|    END;
...

I never saw a comment saying that a using module can't assign chars
to x.name^ [X.name.all]. 


If Ada were following closely an Oberon-2 style, then this could a
 the way to make Ref be read-only:

   type Unbounded_String is [? new Ada.Finalization.Controlled with]
      record
         Length : Natural := 0;
         Ref-   : String_Access := ...;
      end record;

That is sure to be rejected: the modifier ought be on the right of the
":" since it affects the type and not the name.

More on Modula-2: http://www.cetus-links.org/oo_oberon.html



Some selected Error messages of the "Optimizing Oberon-2 Compiler",
follow.
Error messages 244 & 245 imply that in Oberon, making X.Ref be
exported as read-only [either the type of a module variable], where
X is a record type and Ref is a pointer, does not cause X.Ref.all
(i.e. X.Ref^) to be read-only. Just as is hoped for.


http://ooc.sourceforge.net/OOCref/OOCref_19.html

|
|*201: Can only be exported with **'' 
|
|Constants, types, and procedures cannot be exported with *-'
|because the notion of a restricted, read-only access doesn't make
|sense for these objects. 
|
|
|*244: This isn't a variable designator' 
|
|This applies to the following cases: 
|  Only a modifiable variable designator may appear on the left
|    side of an assignment statement. The variable designator may
|    not refer to a variable or record field that has been
|    exported as read-only. 
|  The control variable of a FOR statement has to be an
|    unqualified (i.e., local) variable identifier. 
|Note that dereferencing a read-only pointer variable will grant
|unrestricted access to the pointer's contents. 
|
|
|*245: This is imported read-only' 
|
|This applies to the left side of an assignment statement and to
|variables passed to a VAR parameter as part of a procedure call.
|This means that the destination variable (or designator) has to be
|either
|  1. imported as read-write (and no following selector accesses
|       a read-only record field), or
|  2. the value of a dereferenced pointer variable. 
|Contrary to a pointer dereference, an element of a read-only array
|is read-only, just like a field of a read-only record variable. 
|


I have doubts about Mr Carter's assignment, with its familiar right
 hand side is read only and left hand side is written too nature.

I my proposal, the left side is read-only [i.e. writable to only by
subprocedures in its package], and the right hand side can be written
to (pointers swapped) even though an "in" mode parameter [this last
may need to be checked].

Either that or C-isms of pointers to pointers may appear.



---------

PS. I accidentally 'double posted' the same message.
In some cases the duplicate could be deleted.

PS. By sending data through a debugging TCP proxy, such accidents
ought be harder to do. 

Here is Win32 GUI debugging proxy that shows bytes passing through it:

TCP Viewer:   http://www.westbrooksoftware.com/