Exams

Stakeholder are given many imaginary units (100 dollar, 1000 points, etc.) Points are distrubuted among requirements to prioritize

100 points

• Req1: 40

• Req2: 20

• Req3: 30

• Req4: 10

Benefits and limitations:

• easy to use and fast

• eligible to small number of requirements, not scalable

• firewall separates the classes that depend on the class that is changed from the rest of the classes

• based on the first-level dependencies of modified components

• given 2 successive versions of an OO-system, find the difference of the two versions and identify those classes that have changed

• if any of the changed classes are in an inheritance hierarchy, also conseder descentdants of the changed class as changed

• for each class, identify all the classes that send messaged to the changed class or receive messages from the changed class and include them in the “firewall”

• testing approach where test cases are automatically generated from models

• models are the expected behavior of the system under test and can be used to represend the test strategy

• MBT usually includes following steps:

  1. design a test model

  2. select some test generation criteria: e.g. test generation cirteria should be full state coverage, if the model id a finite state machine

  3. generate the tests

  4. execute the tests

• ensure all requirements are implemented

• change impact analysis

• avoid “pold pating”

• change impact analysis

• defect impact analysis

• root cause analysis of defects

• traceability matrix

• trace tagging

• choose only subset of code that is relevant to a particular variable at a certain point

• minimize the code for most cost-efficient testing, debugging, or testing

• a slice with respect to a variable v at a certain point p in the programm is the set of statements that contribute to the value of the varaible v at p

• to establish is, all Defs and All P-uses of a variable v before a certain point p are chosen

• integration testing can be a top-down, bottom-up, sandwich, or big band

• basic ides is to use “stub” and “driver” to simulate the called function or call function

• Pros: intuitive and easy fault isolation

• Cons: need “stub” or “driver”

• integration testing is based on the call graph of neighbors or interfac matrix

• Pros: do not need “stub” or “driver”

• cons: diffocult to isolate and locate the fault

• integration testing is like an enlarged path-based unit test

• here the node is not a statement but a component

• pros: closely coupled with actual system behavior

• cons: extra effort is needed to identify message path

Regression test selection process usually includes 3 main steps:

• run the tests and establish the link between the test and the code executed

• after the code id changed, identify the changed code

• based on the link between the code and the test, choose only those tests that are related to the changed code and re-run them

• “how standards” focus on how we should work- e.g. how we should work to achieve safety or security. at present most standards are “how” standards

• “what standards” (also called goal-based standards) focus on what shall be achieved, not on how to achieve. It is up to the developers to convince the assessor that what they do will meet the standard’s requirements

• to generate test cases that are evenly spread over the entire input domain as possible

• you start with one test that is generated randomly

• then for next test case, you pick the one that is the farthest away fromm all executed test cases

  
  

• good for F-measure, i.e. it usually requires fewer test cases to detect the first failure comparing to random test

• can, at least partly, be automated

• can consistently repeat several actions

• fast and high volume

• is only a spot check

• may need several stubs and drivers

• can see the complete picture, not only a spot check

• can use all information in the team

• can be innovative and inductive

• unreliable (people can get tired)

• slow and low voume

• test how well system can be scaled up to process the increased workload

• testern can gradually increase the workload to the system

• add more serves and workload managers, in order to measure how well the system can process the increased workload through adding more servers

• test how well system can perform acceptably under worse-case condition

• system is deliberately stressed by pushing it to and beyond its specific limits for a while in order to idenfity some bugs related to memory, e.g. memory leak

• testers simulate the increased number of parallel transactions and add more servers and configure the workload manager to check how well the added servers deal the increased workload

• testers will simulate the max. number of parallel transactions for a while to see if the performance of the system is still acceptable

b, d

a,b,c