> Date: Tue, 16 Sep 2003 09:43:14 -0400
> From: "Ray Denenberg, Library of Congress" <[log in to unmask]>
>
> > Heavens to betsy!  A four-message thread and I can't find a single
> > statement to disagree with!  What on earth is the world coming to?
>
> Could you summarize then, because I didn't follow it.

I'm sure Matthew or Rob will shout if I misunderstood their dialogue,
but I took it to mean that (A) the mechanism described by the proposal
does explicitly offer CQL profilers the wherewithal to invent
completely new attribute types, and quite right too; but (B) if
profilers really, really want to, then the mechanism is sufficiently
flexible to make it possible for them to cheat (and quite right too).

 _/|_    _______________________________________________________________
/o ) \/  Mike Taylor  <[log in to unmask]>  http://www.miketaylor.org.uk
)_v__/\  "The body of a destructor is executed before the destructors
         for member objects.  Destructors for nonstatic member
         objects are executed before the destructors for base classes.
         Destructors for nonvirtual base classes are executed before
         destructors for virtual base classes.  Destructors for
         nonvirtual base classes are executed in reverse order of
         their declaration in the derived class.  Destructors for
         virtual base classes are executed in the reverse order of
         their appearance in a depth-first left-to-right traversal of
         the directed acyclic graph of base classes; ``left-to-right''
         is the order of appearance of the base class names in the
         declaration of the derived class." -- the C++ reference
         manual, section 12.4, reminding my Why C++ Is Not My Favourite
         Programming Language.

--
Listen to my wife's new CD of kids' music, _Child's Play_, at
        http://www.pipedreaming.org.uk/childsplay/