Path: news.daimi.aau.dk!olevi
From: olevi@daimi.aau.dk (Ole Villumsen)
Newsgroups: comp.object,comp.lang.eiffel,comp.lang.sather,comp.lang.beta,comp.lang.modula3
Subject: Re: Cows, Hogs & Farms
Date: 6 Jan 1995 10:12:29 GMT
Organization: DAIMI, Computer Science Dept. at Aarhus University
Lines: 37
Message-ID: <3ej52d$3va@belfort.daimi.aau.dk>
References: <3e07c4$sup@muss.cis.McMaster.CA> <1995Jan5.183534.10065@schbbs.mot.com>
NNTP-Posting-Host: angers.daimi.aau.dk
Xref: news.daimi.aau.dk comp.object:23483 comp.lang.eiffel:7116 comp.lang.sather:1388 comp.lang.beta:198 comp.lang.modula3:3296

I haven't followed (nor understood) the entire discussion.
But as far as I can see, Beta does not have the problem Sather
is claimed to have.

  herbivore: (# foodType:< plant; (* virtual type *)
                eat: (# someFood: ^foodType;
                     enter someFood[]
                     do ...
                     #);
             #);

  cow: herbivore (# foodType::< grass; (* further binding the virtual *)
                 #);

  hog: herbivore (# foodType::< corn; #);

Now we can go (assuming the variables are declared with the expected types:
someGrass: ^grass; somePlant: ^plant; etc.):

  someGrass[] -> &aCow.eat; (* o.k.; typechecked at compile-time *)

  someGrass[] -> &aHog.eat; (* error, reported by compiler *)

  somePlant[] -> &aCow.eat; (* compiler will insert a runtime type check *)

  someCorn[] -> &aHerbivore.eat; (* again, runtime type check *)

  (if somePlant## // grass## then somePlant[] -> &aCow.eat;
                  // corn##  then somePlant[] -> &aHog.eat;
  else ...
  if);

How's that? (The compiler will insert runtime type checks in the last
example, unless you instruct it not to; maybe future versions of the
compiler will be smart enough to "see" that it's not necessary.)

Ole