Nope.   I  certainly  agree  it should be fixed.  Historically, it was
   never  high  on  the "to-fix" list, but in today's world, it certainly
   would be advantageous.

   ______________________________________________________________________

   From: [EMAIL PROTECTED]
   [mailto:[EMAIL PROTECTED]   On   Behalf  Of  "Martin
   Phillips" <[EMAIL PROTECTED]>
   Sent: Friday, December 16, 2005 12:49 PM
   To: <[email protected]>
   Subject: Re: [U2] global catdir question - security hole

   I  think that this goes back to the issue I tried to raise a couple of
   months
   back but failed to get much interest.
   Imagine that I have an employee who has a valid user name and password
   to
   use  my  uniObjects  based  application. He is a knowledgeable sort of
   chap who
   goes home and uses uniObjects to develop his own front end that he now
   installs  on  his  PC  in  the  workplace.  His existing user name and
   password
   allow  him to connect but, because he's got his own front end program,
   he now
   has  freedom  to do anything he wants because uniObjects allows him to
   open
   files, write records, execute programs and commands, etc.
   OK,  you  can  say we should fix this with hiring practices. Is that a
   valid
   excuse under SOX when your system gets trashed or your commercially
   sensitive data is stolen?
   Or you could say that the user should not be able to install his own
   software on his PC. Yes, that's true but just how often are such rules
   actually enforced? The serious malicious user will usually find a way.
   Or  you  might  argue that permissions should control this. Not true -
   The user
   will  have the same rights as he has when he's in the real application
   and,
   as this thread pointed out, that allows him to replace programs in the
   catalogue opening the prospects of viruses and torjan horses, etc.
   Maybe  I'm  just  being  paranoid  but  I  believe that the uniObjects
   interface
   opens  a  security  hole  that needs closing. Dare I say that in QM we
   addressed
   this in the equivalent interface by providing configurable options
   controlling  what  the  client can do. In the strongest case we do not
   permit
   opening  of  files  or  execution  of commands directly by the client,
   everything
   having to be done under the control of the server side application.
   Furthermore,  we  only  allow  the client to directly call subroutines
   that are
   compiled with a special option. Why can't uniObjects provide a similar
   level
   of security? Or have I missed something?
   Martin Phillips
   Ladybridge Systems
   17b Coldstream Lane, Hardingstone, Northampton NN4 6DB
   +44-(0)1604-709200
   -------
   u2-users mailing list
   [email protected]
   To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

Reply via email to