Browse Prior Art Database

Defect Isolation

IP.com Disclosure Number: IPCOM000111893D
Original Publication Date: 1994-Apr-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 84K

Publishing Venue

IBM

Related People

Luke, CA: AUTHOR [+3]

Abstract

This invention relates to defect isolation in programming sequences. The defect isolation is an extension to the test case generation work which has been done [1,2]. Defect isolation is the minimization of test case input required to detect a defect and the identification of test case input impacting the success or failure of a test case.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Defect Isolation

      This invention relates to defect isolation in programming
sequences.  The defect isolation is an extension to the test case
generation work which has been done [1,2].  Defect isolation is the
minimization of test case input required to detect a defect and the
identification of test case input impacting the success or failure of
a test case.

The reasons for defect isolation are:

1.  to determine what product inputs are processed incorrectly so
    that defects may be reported,

2.  to determine what product inputs are processed incorrectly to
    support adaptive testing (2), and

3.  to determine what product inputs are processed incorrectly to
    support correctness measurement (2) when the failure unit is
    defects.

      Defect isolation though not new is a subject which has had
little written about it.  Developers have been doing defect isolation
by minimizing the input required to detect a defect for years.  The
difference between what developers have done in the past and the work
described in this paper is that in this paper the test cases are
written by a test case generator and the minimization process uses
the test case generator to rewrite the test case.  Also, the test
case generator test cases are self-checking.

"The ideal would be to produce test programs together with the
expected trace of the results of their execution, but this is a very
difficult objective.  However, it is too early to discuss possible
approaches to this problem" [1].

"What is left out is a code-generation and run-time system.  To
complete the job, using the same formalism, we have studied (but not
yet implemented), a system for the generation and interpretation of
executable (i.e., run-time, error-free-programs) in order to test the
code generator and run-time system" [2].

      The basic ideas behind the implementation of defect isolation
methods are (1) searching and (2) the random statements created by
the generator are chains of function.

      Defect isolation with partitioning of statements takes the
first half, the last half and the middle ha...