Hi Doug:

thank you for your feedback. I agree with you.
Josh: please can you update the interfaces/design document based on
Doug's feedback and send it to us.

>From dougo@ccs.neu.edu Thu Jan 14 17:07:06 1999
>From: Doug Orleans <dougo@ccs.neu.edu>
>Date: Thu, 14 Jan 1999 17:07:03 -0500 (EST)
>To: Karl Lieberherr <lieber@ccs.neu.edu>
>Cc: atoosa@ccs.neu.edu, jayantha@ccs.neu.edu, johan@ccs.neu.edu
>Subject: Re: your project
>
>Karl Lieberherr writes:
> > The methods long_get, long_collect and traverse take either a 
> > Strategy-object or a TraversalGraph-object as input.
>
>Maybe TraversalGraph could be a concrete subclass of the abstract
>class Strategy?  In some sense, it is a strategy, just a very detailed
>one.  (Other subclasses of Strategy would be PathDirective and
>StrategyGraph.)

That is a very good idea.

>
> > We assume that a traversable part x has a public get_x().
>
>Maybe we should use the Bean convention and call this getX()?

I agree.

>
> > static Object longGet(Object o, Strategy s);
> > static Iteration longCollect(Object o, Strategy s);
>
>I'm not really happy with these "long" names.  How about something
>like "retrieve" and "retrieveAll"?  Or "find/All"?  (As long as we
>don't conflict with the Java 2 collection hierarchy methods.)

I like find/All but is it already used?
long* has the advantage that a conflict is less likely.
>
> > static Method getAccessor(ClassName cn, PartName pn); //Josh
> > // returns method object for accessor method of class
> > // cn and part pn
>
>These arguments should probably just be Strings, if this is something
>the Java programmer will be expected to call.

You are right.
Use primitive types in DI interface: A personality recomendation.

>
> > class ClassGraph
> > ClassGraph(File f[]);
>
>Is this an array of Java source files?

That is what I had in mind.

>
>--Doug
>

-- Karl