testing
Quality
Issues
 |
The quality
of software is the extent to which the product satisfies its specifications. |
 |
The Software Quality Assurance Group (SQA)
has additional responsibilities with regard to software quality. |
 |
Software Quality Assurance |
 |
ensures that software is correct at each phase |
 |
ensures that the entire product is correct |
 |
SQA also applies to the software process itself. |
 |
responsibilities of the SQA include the development
of various standards to which the software must conform, as well as the
establishment of monitoring procedures. |
 |
Managerial Independence |
 |
It is important to have managerial independence
between the development team and the SQA group. Both managers should report
to a more senior manager who could decide which of two possibly conflicting
choices should be made so that the decision is in the best interest of
everyone involved. |
 |
The additional cost of an SQA group is relatively
small compared to the resulting benefit, namely higher quality software. |
 |
Where it is not economically viable to have
a separate SQA group (company too small) it is at least best to ensure
that the specification document be checked by someone other than the person
responsible for producing those specifications. |
Nonexecution-Based
Testing
 |
The review task must be assigned to someone
other than the original author of the document. |
 |
Two types of review techniques: Walkthroughs
and Inspections |
 |
Fundamental difference between them is that
walkthroughs have fewer steps and are less formal than inspections. |
 |
Walkthroughs |
 |
A walkthrough team should consist of 4 to
6 individuals. |
 |
A specification walkthrough team should include: |
 |
at least one representative from the specifications
team |
 |
manager responsible for the specifications |
 |
a client representative |
 |
a representative of the team who will perform
the next phase |
 |
a representative of the SQA group. |
 |
the SQA group member should chair the walkthrough,
because the SQA representative has the most to lose if the walkthrough
is poorly performed and faults slip through. |
 |
The quality of the product is a direct reflection
of the professional competence of the SQA group. |
 |
Each reviewer should review the material
for the walkthrough and develop two lists: |
 |
items the reviewer does not understand |
 |
items the reviewer believes are incorrect |
 |
Managing Walkthroughs |
 |
Task of the team is to record the faults for
later correction, not to correct them. |
 |
Reasons: |
 |
A final document by the committee would be
inferior; person trained should do the corrections. |
 |
Cost for paying off team members is not an
effective use of time. |
 |
Analyze proposed faults; not everything found
is necessarily a fault. |
 |
Not enough time to detect and fix -
walkthrough should last no more than two hours. |
 |
Two ways of conducting a walkthrough: |
 |
Participant driven |
 |
participants present their lists and the representative
of the specifications team answers their questions. |
 |
Document driven |
 |
more thorough |
 |
recommended by IEEE |
 |
Person responsible for the document walks
through it |
 |
reviewers interrupt either with prepared questions
or questions triggered by the presentation |
 |
NOTE: Neither walkthroughs nor inspections
should be used to evaluate the participants |
 |
Inspections |
 |
first proposed by Fagan in 1976 for testing
of designs and code |
 |
Five Formal Steps: |
 |
Overview of Document
-
summary
of the document is given, and then copies are distibuted to all participants
at end of overview. |
 |
Preparation -
participants try to understand document; similar to a walkthrough but look
over all fault types and frequency from other documents and similar products |
 |
Inspection Meeting
-
Be sure every item is covered |
 |
Purpose - Faults are
uncovered, not corrected |
 |
Moderator productes
a report from the activities of the meeting |
 |
Rework |
 |
Faults and problems
are resolved during the rework |
 |
Follow-up: |
 |
Check all items that
were resolved or fixed |
 |
All new fixes have to
be checked and if more than 5% of the material from the original document
was changed, then you do a complete reinspection |
 |
fault statistics are
recorded by severity |
 |
major faults |
 |
minor faults |
 |
How to use this record: |
 |
compare faults with
a similar product |
 |
detect parts in subparts
of product; then look at related parts |
 |
If there are many faults
in a particular part then redo that phase |
 |
Number and types of
faults in one phase help you to prepare for the next phase |
 |
Overall, the strengths
of reviews are that they are : |
 |
effective ways to find
faults |
 |
faults are detected
early |
 |
Weaknesses |
 |
Large-scale software
is hard to review unless there are independent units |
 |
O.O. technique helps
here |
 |
Need documentation from
previous stage |
Execution-Based
Testing
What
Should Be Tested?
Who
Should Perform Execution-Based Testing?
Testing
Distributed Software
Testing
Real-Time Software
When
Testing Stops

|