Re: Cobol data protection? Get a dog...

From: RKRayhawk (rkrayhawk_at_aol.com)
Date: 10/10/03 

Date: 10 Oct 2003 21:49:17 GMT

One of the more effective posters here,

"Peter E.C. Dashwood" dashwood@enternet.co.nz
Date: 10/7/03 7:27 PM EST
Message-id: <3f835b32$1_7@news.athenanews.com>

Commented in part, ...

<<
It would be foolish to abandon the network because it is perceived to be
"insecure". The fact is, it can be as secure as you are prepared to make it.

>>

In passing, let me cite

http://money.cnn.com/2003/10/09/technology/hacker.reut/index.htm?cnn=yes

for several reasons. First, although we are obligated to keep this newsgroup at
or near the subject of COBOL. The real interest, atleast generally, is
business.

Additionally, IMHO, it is useful to reset the basic interpretation of the
danger. The danger on the internet is not perceived, it is very real.

So any and how, back to the top.

I find at the incredibly generous IBM site,

http://publibfp.boulder.ibm.com/cgi-bin/bookmgr/BOOKS/igy3lr10/6.1.2?DT=20
020920180651

 the following piece of information concerning
the PROCEDURE DIVISION header, and particularly the USING clause ...

"Parameters specified in the USING phrase of a method Procedure Division
header must be passed to the method BY VALUE. "

Now that dove-tails for me. But I have some questions and would like to lean on
some of you who are most knowledgable about the standards process.

Is this BY VALUE only rule for methods the standard? Are most of the vendors
sticking to it?

If it is the standard, is consciously an attempt to keep COBOL in a specific
category of procedural languages? Is this the establishment of a definition for
the language according to a theory that we may get more protection by avoiding
the 'common' reference items?

Or is this just what can be agreed on so far?

Clearly, in an interoperable environment, resolving the addresses when
attempting to
dereference a reference (back to the invokers WS or worse back up through the
LS of a mainline flatland program which can invoke an object) is a serious
problem. So I could see vendors resisting BY REFERENCE, just for
feasibility of early product getting to market soon.

I am aware that you can pass an object reference to a method, which is a whole
nother can of worms. And indeed, because arrays are objects you can pass arrays
of object references. I need to understand that better.

But it stands to reason that an object reference
is a different kind of thing than mere reference,
and there is atleast the possibility of evolving
seriously constrictive shepherding for itl.

So, the question is, does this rule reflect a conscious design decision to keep
COBOL in
a category of procedural type languages that
prevent access by the invokee to the invoker's
data area? Has there been elaborate discussion of this in the standards
process, and does the rule reflect a fairly strong commitment to such a theory
of source coding?

Best Wishes,
Bob Rayhawk
RKRayhawk@aol.com

Relevant Pages

  • Re: How do the experts here quickly come up with references to standards?
    ... >> already familiar with the language standards enough to quickly find ... It's one big HTML file, and the conversion to HTML leaves ... > I don't know about reading it straight through, ... The same with any reference standard, ...
    (comp.lang.c)
  • Re: OT - If Sarah Palin Ever Reappears
    ... ideals and standards of the American ... After all, from *their* intellectual frame of reference, she has no serious ... attractive to the neocon movement, ...
    (rec.music.classical.recordings)
  • RE: reference to handle / handle to reference
    ... As far as I know, there's no conversion between an object reference and an Int32, unless you pin objects using GCHandles and get their associated IntPtrs. ...
    (microsoft.public.dotnet.general)
  • Re: Q: closing forms
    ... The only real difference between passing by reference and by value takes ... passing the object reference by reference ... If a Is Nothing Then MsgBox("After call using ByRef, ... Private Sub MyProcByReference ...
    (microsoft.public.dotnet.languages.vb)
  • [JavaIDL] Releasing CORBA object references in Java ?
    ... I was thinking that reference counting was the job of GC. ... But here in Java, the introduction of _releaseand _duplicate() is ... Returns a duplicate of this CORBA object reference. ...
    (comp.lang.java.programmer)