Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!sri-unix!hplabs!hplabsc!marvit
From: marvit@hplabsc.UUCP
Newsgroups: comp.edu
Subject: Re: Cheating on Programming Assignments
Message-ID: <51700001@hplabsc.UUCP>
Date: Tue, 31-Mar-87 11:40:00 EST
Article-I.D.: hplabsc.51700001
Posted: Tue Mar 31 11:40:00 1987
Date-Received: Sat, 4-Apr-87 04:15:12 EST
References: <248@rruxa.UUCP>
Organization: Hewlett-Packard Laboratories - Palo Alto, CA
Lines: 34

<Concerning cheating with program assignments and testing theose programs>

I have never taught a class on a PC, so take my comments with the appropriate
filter. Bottom line: I give my assignments with a specified Input, a
specified Output, and some *sample* data.  I then run each program on a set
of test data unknown to the students.

The test data I se to verify programs is (usually) designed to test
boundary conditions, error handling, etc.  The students are welcome to use
any test data they wish during development, but the final product has to
pass mine.  This is simplified (frommy point of view) by, for example, the
program looking at the contents of a file with a particular name, or taking
preformatted input from a particular device (eg, keyboard), or have
specific entry/exit points for their code which then is linked to some
extant modules (for particularly hairy routines).

On a time-sharing system, I have (on occasion) enforced program names and
had a daemon which woke up at the due date/time, collected all programs
from the student's accounts, and whisked them away to my directory to be
graded.  No problem with straggling late comers. No problem with forged or
smudged pieces of paper.  This system had some drawbacks, tho', including some
difficulty with personal flexibilty with due dates, and (especially) the
careful construction of the problems so the solutions would be amenable to
this method.  However, I feel it was also closer to "real" life where
a programmer is given specifications for a module which must integrate into an
existing body of work and for which the interface may be well-defined.

While automated collection is more difficult with PC-based assignments, the
rest of the approach for programming tasks shouls be adaptable.  I'd be
interested in your feedback.

-Peter Marvit
 HP Labs
 marvit@hplabs.hp.com