Re: Adding side-effects to assignment

[mcr@Sandelman.OCUnix.On.Ca (Michael Richardson)]

 [ I'm responding the the list only, the propagation on it (even to my
non-internet system) seems instantaneous anyway. I'm getting two,
sometimes three copies because of the CC:'ing ]

On Mar 13,  1:52am, Danny Jared Epstein wrote:
} What I have done in cases like this is to put the data slot (say
} fooData) in a parent (say _ fooParent*) and then override the
} assignment slot (fooData:) locally with a method:
} 
}   fooData: newValue = (
}     "do something".
}     fooParent.fooData: newValue.    "directed resend"
}   ).

  An good solution. 

} This technique lets the read operation be inherited directly and it
} avoids the naming problem.  To copy the object, you will have to copy
} the parent as well, so you should make the parent inherit from traits
} clonable.

  How will inheriting from traits clonable get me anything? I'll have
to send the fooParent a 'clone' message anyway. 
  [ Inheriting is not a solution. Many of my objects will be 'one of a
king' type things. Things that are 'just like x' except for blah. ]
  If traits clonable meant that all 'representation parents' (was that
the term you used before?) were automatically cloned when I asked for
a copy, that would be neat. Reimplementing copy in every object is not
my idea of fun...
  I'll experiment a little more with this when 1.3 is available.


-- 
   :!mcr!:            |  The postmaster never | So much mail, 
   Michael Richardson |    resolves twice.    |  so little time.
HOME: mcr@sandelman.ocunix.on.ca 	Bell: (613) 237-5629
    Small Ottawa nodes contact me about joining ocunix.on.ca!