>From dougo@ccs.neu.edu Mon Sep  1 17:25:05 1997
>Received: from vega.neu.edu (vega.ccs.neu.edu [129.10.115.77])
>	by amber.ccs.neu.edu (8.8.6/8.7.3) with SMTP id RAA01667; Mon, 1 Sep 1997 17:25:03 -0400 (EDT)
>Received: by vega.neu.edu (SMI-8.6/SMI-SVR4)
>	id RAA26004; Mon, 1 Sep 1997 17:24:50 -0400
>Date: Mon, 1 Sep 1997 17:24:50 -0400
>Message-Id: <199709012124.RAA26004@vega.neu.edu>
>From: Doug Orleans <dougo@ccs.neu.edu>
>To: Karl Lieberherr <lieber@ccs.neu.edu>
>Cc: boaz@ccs.neu.edu
>Subject: Re: example in Demeter/Java
>In-Reply-To: <199708302327.TAA14121@stockberg.ccs.neu.edu>
>References: <199708302327.TAA14121@stockberg.ccs.neu.edu>
>X-Mailer: VM 6.22 under 19.15 XEmacs Lucid
>X-Face: (4D-osoq?}7M3\EgvbWKo<JkN/8h)A`1b^S1[8/OtYE1A61B!AOmH#YD+{HKhr7}@8gMv~.
>	 tsxTzT"g.oP0dTl!q
>Mime-Version: 1.0 (generated by tm-edit 7.106)
>Content-Type: text/plain; charset=US-ASCII
>Status: R
>
>Karl Lieberherr writes:
> > Next we define a traversal/visitor combination {\tt t}, 
> > called traversal {\tt t}, 
>
>I call this a "traversal method", which is a strategy combined with a
>list of parameter types (visitor classes).
>
> > We may specialize traversal {\tt t} by 
> > creating methods using subclasses of {\tt CountingVisitor}.
>
>This statement doesn't make much sense, and seems extraneous to the
>discussion anyway.  
>
> > Finally, we define a method {\tt CountCertainEs} for class {\tt A} 
> > which defines
> > with which subclass of {\tt CountingVisitor} we want to do the work.
>
>I call this an "adaptive method".
>
> > To summarize the adaptive programming approach used above, called
> > the strategy/visitor style, we note
> > that behavior is expressed in terms of strategies and
> > visitors. Strategies and visitors are combined in two steps:
> > 1. First a family of strategy/visitor combinations, 
> > called traversals (overloading the term),
> > is defined. 2. A specific member of this family is selected by specifying
> > which specific visitor subclass to use.
>
>This notion of a "family" is kind of confusing-- a traversal method is
>exactly that, a method that defines a traversal (using a strategy).
>Also, as long as you use "traversal strategy" everywhere to mean an
>abstract path directive (i.e. not attached to a particular class
>structure or visitor class), and "traversal method" to mean a method
>attached to a class that traverses an object graph, I don't think
>there's any overloading.  However, it's still a little confusing to
>bring this up-- again, maybe using the "inline" style of adaptive
>methods directly would be clearer for the discussion?  The paper is
>about strategies, not the ability to re-use visitors (as opposed to
>your keynote speech, which is about both).  E.g. simply:
>
>    \begin{verbatim}
>    A{
>      int CountCertainEs() do S {
>	before E (@ return_val++; @)
>      }
>    }
>    \end{verbatim}
>
>Hm, now that I look at it, I wonder if "use" would be a better keyword
>than "do" for referring to named strategies?  Or "using"?  Hm.
>
>--Doug
>