By: BJ Rollison
Email:
Địa chỉ email này đã được bảo vệ từ spam bots, bạn cần kích hoạt Javascript để xem nó.
Organization: Microsoft Corp.
Title: Test Architect
Abstract
Imagine you are assigned to test a feature with 20 parameters that are interdependent. There are 5 possible variable states for each parameter. The total number of possible combinations is greater than a half trillion; which means that at one test per millisecond it would take more than 3000 years to test all possible combinations. Which combinations do we test? Pair-wise testing is a systematic procedure to effectively reduce the total number of tests by selecting a set of tests that evaluates every pair combination because historical and root cause analysis shows the majority of errors caused by the interaction of variables occurs between 2 parameters rather than interaction between the variables for 3 or more parameters. This talk compares orthogonal arrays to pair wise analysis, and then provides a detailed example of how to use a powerful, highly configurable combinatorial analysis tool that is freely available to systematically test complex interdependent parameters.
Part I will introduce the basics concepts of combinatorial fault models, and why coverage arrays may be an effective approach to help testers identify these types of bugs earlier. It will also illustrate how to model input variables that affect a common output condition, and how to use those models to generate a set of pairwise tests using a freely available tool developed by and used within Microsoft.
Part II will introduce advanced input variable modeling techniques, and how to use the tool to randomize the test combinations, increase the order (n-wise) of variable combinations, use weighting values on important variables, create a seed file of ‘must test’ combinations, and other features to produce a set of tests to increase test effectiveness and improve overall test confidence.
IMPORTANT: A laptop computer is required.