Transcript CS 8532: Adv. Software Eng. – Spring 2007 Dr. Hisham Haddad Chapter 13 CS 8532: Advanced Software Engineering Dr.

CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Chapter 13
CS 8532: Advanced Software Engineering
Dr. Hisham Haddad
Class
will
start
momentarily.
Please Stand By …
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Software Testing Strategies
Discussion of Software
Testing Strategies
Chapter 13
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing
Testing is the process of exercising a
program with the intent of finding
errors prior to delivery to the end user,
and it requires a strategy.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
What Testing Shows
errors
requirements conformance
Performance issue
an indication
of quality
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Who Tests the Software?
developer
Understands the system
but, will test “gently”
and, is driven by “delivery”
independent tester
Must learn about the system,
but, will attempt to “break” it
and, is driven by “quality”
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
The Big Picture
Testing is an element of the V&V process.
Verification: The software implements the functional specifications.
Validation: The software is traceable to customer requirements
(functional, behavioural, performance). (mapping software
components to requirements)
V&V activities = SQA (see chapter 26 on Quality Management)
The testing group works with the developers and reports to the SQA
team.
“You can’t test in quality. If it‘s not there before you begin testing, it
won’t be there when you finished testing”
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing Strategy (1)
A testing strategy is a plan (road map) that outlines the detailed
testing activities (steps, test case design, test execution, effort,
time, resources). It results in a Test Specification document.
Many testing strategies are proposed. However, common
characteristics of a testing strategies include:
- Testing starts with effective FTR
- Testing starts at the component level and work outward
- Testing is conducted by developers (small projects) and testing
groups (large projects)
- Different techniques are relevant at different points in the
development process
- Debugging is an activity of any testing strategy
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing Strategy (2)
Note 1:
For conventional software, the module (component) is the initial
focus, then integration of modules follows.
Note 2:
For OO software, when “testing in the small”, the focus changes
from an individual module (the conventional view) to an OO
class (or package of classes) that encompasses attributes and
operations and implies communication and collaboration.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Strategic Issues
• State software requirements in quantifiable manner to test for
quality characteristics
• State testing objectives explicitly
• Understand the users of the software and develop a profile for
each user category (use-cases)
• Develop a testing plan that emphasizes “rapid cycle testing”
• Build “robust” software that tests itself – exception handling
• Use effective FTRs to filter errors prior to testing
• Apply FTRs to the test strategy and test cases themselves
• Develop a continuous improvement approach for the testing
process (collect data and develops metrics)
See page 361
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing Strategy Elements (1)
unit test
system
test
integration
test
validation
test
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing Strategy Elements (2)
Unit Testing: testing functionality of individual modules
(using white-box methods).
Integration Testing: testing functionality of integrated modules
(using both white-box and black-box methods).
Validation Testing: testing the software for all established
requirements (functional, behaviour, performance, reliability, …)
(using black-box methods).
System Testing: testing the software for compatibility with other
system elements (HW, users, databases, other systems).
See figure 13.1, page 358.
When does testing stop?
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Unit Testing – Conventional Software
module
to be
tested
results
software
engineer
test cases
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Unit Testing Environment
Test driver
Tested
Module
stub
stub
interface
local data structures
boundary conditions
independent paths
error handling paths
test cases
RESULTS
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Unit Testing Errors
Some common computational errors:
- Incorrect arithmetic precedence
- Incorrect logical operators or precedence
- Incorrect initialization
- Incorrect symbolic representation of an expression
- Incorrect data type comparisons (different data types)
- Incorrect comparison of variables
- Improper or nonexistent loop termination
- Improper modified loop variables
- Improper boundary checks
- Precision inaccuracy
- Others… (see page 363)
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Integration Testing
Testing Options:
- Non-incremental approach (all at once!)
- Incremental construction strategy (one addition at a time)
Incremental approach:
- Top-Down Integration: start with main and work downward
integrating subunits (either depth-first or breadth-first order)
- Bottom-Up Integration: start with atomic unit (working
modules) and work upward integrating other units (clusters).
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Top-down Integration Testing
A
B
G
stubs are replaced one at a time,
“depth-first”.
as new modules are integrated some subset of tests
is re-run.
C
D
F
top module is tested with stubs.
E
Problem: testing upper-level unit may depend on
a lower-level unit!
Sandwich (Combo) testing may be performed!
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Bottom-up Integration Testing
A
B
G
lower modules are grouped into builds
and are integrated
C
D
F
drivers are replaced one at a time,
"depth first"
E
Cluster (build)
Bottom-up integration eliminates the
need for complex stubs!
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Sandwich Integration Testing
A
B
C
D
E
cluster
F
Top modules are tested
with stubs
G
Lower modules that unit B depends on are
grouped into builds that are tested, and
integrated
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Integration Testing - Comments
-
Possible difficulties writing stubs for top-down testing
Top-down testing allows testing control module
The entire program is not tested until the last module is added
Bottom-up testing seems easier to conduct (no stubs)
Sandwich testing is a compromise when selecting an integration
testing strategy
- Critical modules should be identified and tested as early as
possible
- “Test specifications” is a document that contains test plans,
procedures, test cases, environment, resources etc…
It becomes part of Software Configuration.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Other Integration Testing
Regression Testing: Used to test side affect each time a new
module is added.
It is re-execution of subsets of test cases that already been
conducted so that side effects (if any) are uncovered.
Smoke Testing: A top-down or bottom-up integration test for
“shrink wrapped” software applications that consist of “daily
builds” (releases).
Steps:
- Integrate new code into a “build.” ( data files, libraries, reusable
modules, and components required to implement functions)
- Design a series of tests to expose errors in the new build. (errors with
highest possibility to affect project progress)
- Integrate current build with other builds and smoke test (daily) the
entire product.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
OO Testing (1)
- OO Testing begins by evaluating the correctness and
consistency of the OOA and OOD models.
(Testing the CRC model, next slide)
- The nature of OO software changes testing strategies.
- the concept of the ‘unit’ broadens due to encapsulation
- cannot test class methods in isolation due to object
collaborations and inheritance
- class testing is driven by its methods and behavior (states)
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
OO Testing (2)
Conventional integration testing (top-down and bottom-up) are
not applicable to OO software.
Class integration options:
- thread-based integration (classes that respond to one event
(input) to the system)
- use-based integration (independent classes first, then
dependent classes)
- collaboration-based (cluster) integration (classes that
collaborate together to complete one collaboration)
Note: always apply regression testing for side effects.
Test case design draws on conventional methods, but also
encompasses special features.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing the CRC Model
1. Revisit the CRC model and the object-relationship model.
2. Inspect the description of each CRC index card to determine if
a delegated responsibility is part of the collaborator’s
definition.
3. Invert the connection to ensure that each collaborator that is
asked for service is receiving requests from a reasonable
source.
4. Using the inverted connections examined in step 3, determine
whether other classes might be required or whether
responsibilities are properly grouped among the classes.
5. Determine whether widely requested responsibilities might be
combined into a single responsibility.
6. Steps 1 to 5 are applied iteratively to each class and through
each evolution of the OOA model.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
OO Testing Strategy
Class testing is the equivalent of unit testing
- operations within the class are tested
- the state behavior of the class is examined
Integration testing applies three different strategies:
- thread-based testing: integrates the set of classes required to
respond to one input or event
- use-based testing: integrates the set of classes required to
respond to one use-case (usage scenario)
- cluster testing: integrates the set of classes required to
demonstrate one collaboration (determined from the object
relationship and CRC models)
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Validation Testing
Validation test focus on software conformance with software
requirements, and is based on the Validation Criteria, a
section of the SRS document.
behavioral characteristics, software configuration items,
performance characteristics, documentation, error recovery,
maintainability, and others…
Mainly black-box based testing.
Alpha test - acceptance test performed by the customer at the
developer’s site to validate system requirements.
Beta test - acceptance test performed by the customer at the
customer's site to validate system requirements.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
System Testing
A series of tests for system compatibility with HW, users,
databases, and other systems.
Example tests:
Recovery testing: test the system’s ability to recover from a
failure. Force the system to fails and see how it responds.
Security testing: test built-in security methods. Try to gain access
as an unauthorized user of the system.
Stress testing: test the system for abnormal conditions (resource
allocation). Try to overwhelm the system.
Performance testing: test the system’s run-time performance. Try
to cause system degradation and failure.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Testing vs. Debugging
Testing uncovers errors; while debugging removes them.
Testing is a process; while debugging is an art.
Debugging outcome is either “error cause is found” or “error
cause is not found!”
Debugging is difficult!
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
The Debugging Process
test cases
new test
cases
regression
suspected
tests
causes
corrections
identified
causes
results
Debugging
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Why Debugging is Difficult?
- cause may be due to a combination of non-errors (rounding)
- cause may be due to a system or compiler error
- cause may be due to assumptions that everyone believes
- causes may be distributed among processes/tasks
- symptom may be irregular due to both HW and SW
- symptom may disappear when another problem is fixed
- symptom and cause may be geographically separated
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Debugging Effort
time required
to correct the
error and
conduct
regression
tests
time required
to diagnose
the symptom
and determine
the cause
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Consequences of Debugging
infectious
damage
catastrophic
extreme
serious
disturbing
mild
annoying
Bug Type
Bug Categories: function-related bugs, system-related
bugs, data bugs, coding bugs, design bugs, documentation
bugs, standards violations, etc…
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Debugging Techniques
- Brut Force debugging: “let the system find the error!” (memory
dumps, run-time traces, and inserted output statements)
- Backtracking: trace the code (manually) back to the source of
the error.
- Cause Elimination:
by induction - make “cause induction hypothesis” and
use test data to prove or disapprove the hypothesis.
Or
by deduction - list all possible cause and test them for
elimination.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Debugging – Final Thoughts
- Don't run off half-cocked, think about the symptom you're
seeing.
- Use tools (e.g., dynamic debugger) to gain more insight.
- If at an impasse, get help from someone else.
- Be absolutely sure to conduct regression tests when you
do "fix" the bug.
CS 8532: Adv. Software Eng. – Spring 2007
Dr. Hisham Haddad
Suggested Problems
Consider working the following problems from chapter 13,
page 385:
1, 2, 3, 4, 7, and 8.
No submission is required for practice assignments. Work it
for yourself!