Grading programs with fatal, non-STL errors.


R. Clayton (rclayton@monmouth.edu)
Fri, 21 Jul 2000 12:12:33 -0400 (EDT)


Suppose you submit an assignment, and it has a fatal error - it dumps core -
due to, for example, some char * pointer abuse or because of some careless
storage management on your part. This message describes the method I use to
grade such assignments.

If you submit an assignment that has a fatal error due to some non-STL C++
error, I'll fix the error and test your assignment. I'll take 5 points off the
top for the repair, so your maximum possible grade is 95, but otherwise your
assignment is unpenalized. For example, if a working assignment would have
gotten you a 90, your fixed assignment will still get you a 90. You are only
penalized if a working assignment would have gotten you a 100.

Note well the following:

 1 This only applies to errors that are not directly related to the STL. If
   you make errors in your use of the STL, I won't fix them (although I'll
   point them out) and I'll take points directly from your grade.

   How can you tell if an error is STL related or not? That is actually a
   difficult question due to lots of reasons, including storage management.
   However, the following is a good rule of thumb: if you can strip the code of
   all STL-related features and the error still remains, it's not an
   STL-related error; otherwise it is.

 2 I'll only fix an error once. If you submit a later assignment with the same
   error, the error remains unfixed and the points come out of your grade.
   This applies both to STL and non-STL errors. I also reserve the right to
   grade later occurrences of the same error more severely than earlier
   occurrences.

   This rule takes precedence over the first one; that is, if your later
   assignment contains at least one old error, all errors are left unfixed.

I haven't mentioned this earlier because I didn't recognize the need for it
until now.

Let me know if you have any question.



This archive was generated by hypermail 2.0b3 on Fri Aug 11 2000 - 15:25:05 EDT