From etzioni@chum Mon May  6 17:13:42 1991
Received: from chum.cs.washington.edu by wolf.cs.washington.edu (5.64/7.0jh)
	id AA10815; Mon, 6 May 91 17:12:53 -0700
Received: by chum.cs.washington.edu (5.64/7.0h)
	id AA20091; Mon, 6 May 91 17:28:13 -0700
Date: Mon, 6 May 91 17:28:13 -0700
From: etzioni@chum (Oren Etzioni)
Message-Id: <9105070028.AA20091@chum.cs.washington.edu>
To: cs473@chum
Subject: Correction to what I said in class...
Status: R

   Date: Mon, 6 May 91 17:03:42 -0700
   From: vaughn@chinook
   Subject: Wildcard Binding in Prodigy...

   On coco there is a new domain "test" (only c473 can access it).  In this 
   domain I have two operators: ROLL and DRILL.

   Defined as:

   (ROLL 
	 (PARAMS (<obj>))
	 (PRECONDS (object <obj>))
	 (EFFECTS ((DEL (has-hole <obj> <*hole>)))))

   (DRILL
	 (PARAMS (<obj>))
	 (PRECONDS (object <obj>))
	 (EFFECTS ((ADD (has-hole <obj> <*hole>)))))

   If you start with initial condtions: (the "test" problem has these)

   (object plate)
   (has-hole plate h1)
   (has-hole plate h2)
   (has-hole plate h3)

   And the goal:

   (and (has-hole plate h2)
		(~ (has-hole plate h1)))

   It takes 2 operations:

   ROLL plate
   DRILL plate h2

   This is because the ROLL operator *does* delete all 3 holes, even though
   Prodigy has bound <*hole> to h1 to achieve the ~(has-hole plate h1) subgoal.
   So, as a consequence, it has to re-drill h2 to achieve the other subgoal.

   This is what I intuitively expected it to do, but it is in direct
   contradiction to what you said in class today...



From etzioni@chum Tue May  7 13:19:17 1991
Received: from chum.cs.washington.edu by wolf.cs.washington.edu (5.64/7.0jh)
	id AA23824; Tue, 7 May 91 13:18:56 -0700
Received: by chum.cs.washington.edu (5.64/7.0h)
	id AA20853; Tue, 7 May 91 13:34:32 -0700
Date: Tue, 7 May 91 13:34:32 -0700
From: etzioni@chum (Oren Etzioni)
Message-Id: <9105072034.AA20853@chum.cs.washington.edu>
To: cs473@chum
Subject: tracing in Prodigy.
Status: R

If you're operator is failing and you don't think it should
you might try the following:
1. type optrace to prodigy.
2. run the problem.
3. go into analyze and type opfail at the node where the operator
fails.  Analyze will then show you exactly at which point the match
failed.

oren


From etzioni@chum Tue May  7 14:16:09 1991
Received: from chum.cs.washington.edu by wolf.cs.washington.edu (5.64/7.0jh)
	id AA25066; Tue, 7 May 91 14:15:37 -0700
Received: by chum.cs.washington.edu (5.64/7.0h)
	id AA20937; Tue, 7 May 91 14:31:12 -0700
Date: Tue, 7 May 91 14:31:12 -0700
From: etzioni@chum (Oren Etzioni)
Message-Id: <9105072131.AA20937@chum.cs.washington.edu>
To: cs473@chum
Subject: Hanoi questions.
Status: R

Sorry for the last minute messages.  I'm responding to recent
messages.

Some people are alarmed that their encoding does not solve the problem
in less than 2500 nodes.  This is ok.

1. Don't increase Prodigy's node allowence.  Just turn in what you get.

2. In general, you'd solve this with control rules, but you don't need
to write ones for this assignment.

3. Points will not be taken off for using non-static generators THIS
TIME, but not in the future (once we've covered control rules).

4. You don't need to make any changes to your domain based on this
message.  This is just to calm people's anxieties.


From vaughn@june Tue May  7 14:57:29 1991
Received: from june.cs.washington.edu by wolf.cs.washington.edu (5.64/7.0jh)
	id AA25896; Tue, 7 May 91 14:57:28 -0700
Received: by june.cs.washington.edu (5.64a/7.0jh)
	id AA19561; Tue, 7 May 91 15:13:06 -0700
Date: Tue, 7 May 91 15:13:06 -0700
From: vaughn@june (Vaughn Iverson)
Message-Id: <9105072213.AA19561@june.cs.washington.edu>
To: eistrand@wolf
Subject: Re:  prodigy
Status: R

Gee, I've never seen that before...  My only advice is to try changing
various things in your domain (one at a time) until you get it to do
something different...

V

