From dougo@ccs.neu.edu Fri Sep  5 14:29:16 1997
Date: Fri, 5 Sep 1997 14:28:58 -0400
Message-Id: <199709051828.OAA00322@vega.neu.edu>
From: Doug Orleans <dougo@ccs.neu.edu>
To: "Joshua C. Marshall" <jayantha@ccs.neu.edu>
cc: dem@carbon.ccs.neu.edu
Subject: Re: cd *extends*

Joshua C. Marshall writes:
 > Again, should the following work, or am i doing something wrong? :P
 > 
 > ::::::::::::::
 > program.beh
 > ::::::::::::::
 > Main {
 >   (@
 >   public static void main(String[] args) throws ParseError {
 >     A a = A.parse("c");
 >   }
 >   @)
 > }
 > ::::::::::::::
 > program.cd
 > ::::::::::::::
 > A       = ["c" C] ["d" D] .
 > B       = .
 > C       = *extends* B .
 > D       = *extends* B .
 > 
 > Main    = .
 > 
 > 
 > Subclasses redefine parse() with a different return val.

Yep.  Um, I dunno what to do about this.  This is why I don't generate
parse methods on alternation classes.  I guess I could put the name of
the class in the name of the parse method, but that seems kind of
tedious.  Or I could have all parse methods return Object, and the
caller would have to cast it to the right thing.  This is probably a
reasonable compromise, since it's how you have to retrieve things from
Vectors and Hashtables etc.

Comments?  Other ideas?  Too bad Java doesn't have contravariant
subtyping of functions...

--Doug