Thread

1. RE: [QUESTIONS] The Complete and Ccomprehensive list of Newbie qu estion

   Jackson, DeJuan <djackson@cpsgroup.com> — 1998-03-17T01:41:38Z

   Is this possible?  Or is it already being done?
   
   <a lot snipped>
   > > > How do you assign function to your objects?
   <snip>
   > I don't know if tying functions to classes has been added
   > to the TO-DO list.  You may want to post to the hacker list and
   > suggest
   > it.
   > 
   Access to the function would then be gained by calling them through the
   class. (i.e. select EMP.totalPeople();, or update CORPEMP set
   CORPEMP.totals = DIVEMP.totalPeople();)
   Just an OO thought.  (*ahhhh* It's that concurrency/data-hiding/coupling
   thing again.)
   Then one could disallow select/update on a table but all execute on
   specific functions.  I do realize that this can be accomplished through
   triggers, but this has other potentials...
   
   

2. Re: [HACKERS] RE: [QUESTIONS] The Complete and Ccomprehensive list of Newbie qu

   ocie@paracel.com — 1998-03-17T01:58:56Z

   Jackson, DeJuan wrote:
   > 
   > Is this possible?  Or is it already being done?
   > 
   > <a lot snipped>
   > > > > How do you assign function to your objects?
   > <snip>
   > > I don't know if tying functions to classes has been added
   > > to the TO-DO list.  You may want to post to the hacker list and
   > > suggest
   > > it.
   > > 
   > Access to the function would then be gained by calling them through the
   > class. (i.e. select EMP.totalPeople();, or update CORPEMP set
   > CORPEMP.totals = DIVEMP.totalPeople();)
   > Just an OO thought.  (*ahhhh* It's that concurrency/data-hiding/coupling
   > thing again.)
   > Then one could disallow select/update on a table but all execute on
   > specific functions.  I do realize that this can be accomplished through
   > triggers, but this has other potentials...
   
   Looks sort of like our procedural language, except that instead of
   having a "pool" of functions, functions can be "member functions" of a
   class.  In order to do this "all the way", it seems like we will need
   all instances to appear in their parent classes -- Select
   people.average_age() should hit all instances of people, even those
   that have been subclassed into teachers, police, lawyers, etc.
   
   Ocie Mitchell
   
   
   

3. Re: [HACKERS] RE: [QUESTIONS] The Complete and Ccomprehensive list of Newbie qu

   Jan Wieck <jwieck@debis.com> — 1998-03-17T07:19:24Z

   >
   > Is this possible?  Or is it already being done?
   >
   > <a lot snipped>
   > > > > How do you assign function to your objects?
   > <snip>
   > > I don't know if tying functions to classes has been added
   > > to the TO-DO list.  You may want to post to the hacker list and
   > > suggest
   > > it.
   > >
   > Access to the function would then be gained by calling them through the
   > class. (i.e. select EMP.totalPeople();, or update CORPEMP set
   > CORPEMP.totals = DIVEMP.totalPeople();)
   > Just an OO thought.  (*ahhhh* It's that concurrency/data-hiding/coupling
   > thing again.)
   > Then one could disallow select/update on a table but all execute on
   > specific functions.  I do realize that this can be accomplished through
   > triggers, but this has other potentials...
   
       Currently, we don't have ACL's on functions. But at least for
       the ones  called  by  ExecMakeFunctionResult(),  this  really
       makes  sense  as  soon  as  we implement a uid/euid model for
       function/view execution.
   
       The uid/euid for functions, triggers and views is already  on
       my  personal  TODO  and  I hope to make it for 6.4, so we can
       offer  view  and  function/trigger  creation  (restricted  to
       trusted languages) to ordinary users.
   
       When  adding  ACL's  to  functions,  I think the ACL's at all
       should be moved into a separate catalog, containing the  Oid,
       owner and the ACL.
   
   
   Jan
   
   --
   
   #======================================================================#
   # It's easier to get forgiveness for being wrong than for being right. #
   # Let's break this rule - forgive me.                                  #
   #======================================== jwieck@debis.com (Jan Wieck) #