students editing output

[render at uiucdcsb.Uiuc.ARPA]

Here at the University of Illinois, our intro programming course for the CS
majors uses the following system for grading programming assignments:

1)  the students get the assignment along with a set of test data and output;
2)  when the student is satisfied that his/her program is correct, (s)he
    run a "hand-in" program which takes a copy of the program source and
    stores it in a protected directory with a time-stamp.  
3)  all of the programs are executed by a test-harness (started up by the TA's)
    which runs the program against a set of test data.
4)  the source code for each program and the results of the test runs
    are then printed and given to graders to mark.  The graders are told
    to keep an eye out for duplicate programs.  As the hand-in program
    both time-stamps the submissions and lists the user-id of the submittor,
    there is a reduced chance of the student getting away with turning in 
    a late program or one that (s)he did not write.

Obviously, this system is not fool-proof.  Yet it has gone a long way to 
reducing the number of arguments that the TA's have with students about
programming grades.  Too, the fact that we weight the tests higher than
the programming assignments means that someone who is not doing her/his
own work on the programs will have a much harder time slipping through.

                                     Hal Render
                                     University of Illinois
                                     {pur-ee, ihnp4} ! uiucdcs ! render
                                     render at uiuc.csnet     render at uiuc.arpa