Browse Prior Art Database

Generation of regression tests

IP.com Disclosure Number: IPCOM000012738D
Original Publication Date: 2003-May-23
Included in the Prior Art Database: 2003-May-23
Document File: 1 page(s) / 39K

Publishing Venue

IBM

Abstract

Disclosed is a process that automates the generation of regression tests.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Generation of regression tests

  Construction of regression tests for output must account for reasonable variation. This could be as simple as the date/time being identified with each output to a variable number of lines depending on timing issues. Creating a sample that accounts for these variations while still being sufficiently restrictive can be time consuming. My invention automates the generation of these regression tests. Customization and termination of these tests can be easily accomplished.

    My initial implementation was to work against hypertext markup language (HTML) pages, but it can be extended to extensible markup language (XML) documents as well. The process is simple, a number of samples from a known good source is provided to the generator. Additionally, there is a list of patterns provided that are ordered by increasing generality. The first sample is denoted as the known good. Samples are compared to the known good to determine where there is a failure. The point of failure is replaced by the items from the pattern file until a pattern is found that causes this sample and all previous samples to pass to this point or beyond, i.e., if a previous file passed, then it still passes. The known good generated will pass all the samples with a local minimum. Replacements are performed on PC data only, i.e., attributes and text. Though identifying deltas in a file is a well-studied area.

Disclosed by International Business Machines Corporation

1