26.3.2003Software Engineering 2003 Jyrki Nummenmaa 1 SOFTWARE PRODUCT QUALITY Today: - Software quality - Quality Components - ”Good” software properties.

26.3.2003 Software Engineering 2003Jyrki Nummenmaa

1

SOFTWARE PRODUCT QUALITY

• Today:- Software quality - Quality Components - ”Good” software properties


2

SOFTWARE ENGINEERINGSOFTWARE QUALITY

• Today we talk about quality - but what is quality?

• ”Suitable”• ”Fulfills requirements”• ”Customer is satisfied”• ”Other attributes than price”• ”Superiority, excellence”• ”Has required and expected features”

• It seems difficult to find a ”perfect” single definition.


3

SOFTWARE QUALITY -QUALITY COMPONENTS

• Objective quality component: properties that can be measured or approximated objectively

• Subjective quality component: customer satisfaction (”What does the product feel like?”)

• Other: features which can not be (even subjectively) evaluated at the time. This is related with future events which can not be predicted - unexpected circumstances, changes, etc.


4

SOFTWARE QUALITY -QUALITY COMPONENTS

• Objective quality component: properties that can be measured or approximated objectively

• Subjective quality component: customer satisfaction (”What does the product feel like?”)

• Other: features which can not be (even subjectively) evaluated at the time. This is related with future events which can not be predicted - unexpected circumstances, changes, etc.


5

SOFTWARE QUALITIES -PRODUCT AND PROCESS

• Product quality - the quality of the software product (including user and technical documentation).

• Process quality - the quality of the software engineering process used to produce the product.

• Users are (understandably) primarily interested in the product qualities.

• The process qualities are used to achieve the product ones.


6

SOFTWARE PROPERTIES -EXTERNAL AND INTERNAL

• External properties are the ones that are visible to the users.

• Internal properties are the ones the ones that are visible to the software developers.

• Users are (understandably) primarily interested in the external properties.

• The internal properties are used to achieve the external ones.


7

Correctness

• Last year in the Christmas issue of ITviikko- magazine professor Jukka Paakki from Helsinki University wished for at least one error-free program.


8

A delayed Christmas gift

program Hello;begin writeln(”1+1=2”);end.

#include <stdio.h>main() { printf ("1+1=2\n"); }


9

More error-free programs

program Hello3;begin writeln(”1+2=3”);end.


10




11




12




13

A PROGRAM ERROR?


14

Correct or incorrect?

program Hello;begin writeln(”1+1=3”);end.• The problem of correctness is in that it does not

depend on the program alone but also on the expectations on the program.

• So, how can we say if any of the previously seen programs was correct or incorrect?


15

How to identify correct programs?

• Sometimes this seems easy.

• But how to define correctness so that we could use the definition to identify correct programs? (and do it correctly:)


16

Mathematical proofs

• Using logic or mathematics, prove that the program has some properties.

• You can use a (hopefully correct) program to assist you in making the proofs.

• Problem: These mathematical properties may appear to be even more complex than the programs themselves.


17

QUALITY COMPONENTS - Correctness

• A program is functionally correct if it behaves according to the functional specifications.

• The functional specifications may not always be available.

• The functional specification may be very informal.• The functional specifications may contain

ambiguities.• Sometimes it is evident what is expected - is it fair

to compare the software with general expectations or its own help?

• Do we assume that the specifications are correct?


18

QUALITY COMPONENTS - Reliability

• A program is reliable, if the user can rely on the software.

• For reliability, the statistical approach could be used: What is the probability that the software fails with a given task?

• The program may be reliable in a user’s point of view even if it is not correct.


19

QUALITY COMPONENTS - Robustness

• A program is robust, if it behaves reasonably (?) well even in unexpected circumstances - i.e. it tolerates unexpected difficulties.

• Dealing with errors? E.g. program input is often different from what is expected.

• The program may be reliable in a user’s point of view even if it is not correct.

• A crucial property in some applications.


20

QUALITY COMPONENTS - Performance

• Performance = efficiency.• Efficiency: memory management, disk

management, CPU usage, ... • Asymptotic behaviour: what happens when inputs

grow larger?• Transaction processing systems:

- Throughput = how many transactions can be processed in a given time slice (average or min)- Response time = the time (max or average) needed to process a transaction.


21

QUALITY COMPONENTS - User friendliness

• A software system is user friendly if the users find it easy to use.

• A subjective quality.• Incorrect, inefficient, and unreliable systems are

not very user friendly.• A non-robust system may be user friendly.


22

QUALITY COMPONENTS - Verifiability

• A software system is verifiable, if its properties can be verified easily.

• The software properties can be verified using testing or formal analysis.


23

QUALITY COMPONENTS - Maintainability

• A software system is maintainable, if it is easy to maintain.

• Corrective maintenance - removing errors (repairability)

• Adaptive maintenance - adapting the software to new or changing environments (evolvability).

• Perfective maintenance - improving other software qualities (evolvability).


24

QUALITY COMPONENTS - Evolvability

• A software system is evolvable, if it is easy to add new functions or change old ones.

• Adding new functions or changing the old ones usually ”eats up” some of the evolvability - after the change the software is usually less evolvable.


25

QUALITY COMPONENTS - Reusability

• A software system is reusable, if it can be used to produce another software system.

• Reusability is rare in practice.• In addition to the program code, also other parts of

the software product, such as designs and documentation, can be reusable.


26

QUALITY COMPONENTS - Portability

• A software system is portable, if it can be run (or it can be made to run) in different environments.

• Portability across different hardware architectures.• Portability across different operating systems.• Portability across different hardware

configurations.


27

QUALITY COMPONENTS - Understandability

• How easy is it to understand the system’s structure and how it works?

• Some tasks are more complex: it is easier to understand an ordinary text editor than an operating system.

• There is internal and external understandability.


28

QUALITY COMPONENTS - Interoperability

• is the ability to co-operate with other systems.• Exchange of data using data files.• Exchange of data using some kind of a clipboard.• Exchange of data using network.• Standard interfaces• Open system - open interfaces


29

QUALITY COMPONENTS - Productivity

• The efficiency of the software production process (internal).

• Huge differences between teams and individuals (starting from the fact that some teams or individuals may not be able to complete some tasks at all).

• In producing new software one individual can easily be 2-4 times more productive than another.

• In maintaining old software one individual can in extreme cases be 20-40 (or even more) times more productive than another.


30

QUALITY COMPONENTS - Timeliness

• The ability to deliver a product in time.• Does not happen too often.• Result: Alpha versions, Beta versions, ”Early pre-

prototype test versions”, ...• Which is better:

to deliver a defective product in time orto deliver a better product late?(Ok, this depends on the situation.)


31

QUALITY COMPONENTS - Visibility

• The software development process is visible, if it is easy to see what has been done and what has happened.

• If all know what the state of the process is, it is easier to know when to do what.

• When personnel changes (and in long projects it does), visibility is very valuable.


32

QUALITY COMPONENTS

• Correctness• Reliability• Robustness• Performance• User Friendliness• Verifiability• Maintainability

• Reusability• Portability• Understandability• Interoperability• Productivity• Timeliness• Visibility


33

SOFTWARE METRICS

• Measurements which relate to a software system, process, or related documentation

• Examples: - size of a product in lines of code- number of reported faults- time required to produce a system component

• Control metrics measure the process• Predictor metrics are measurements of a product

attribute which can be used to predict an associated product quality.


34

PREASSUMPTIONS FOR THE USE OF PREDICTOR METRICS

• We can accurately measure some property of the software.

• A relationship exists between what we can measure and what we would like to know about the product’s behavioural attributes.

• This relationship is understood, has been validated, and can be expressed in terms of a formula or a model. (This last assumption is often ignored.)


35

SIZE AND COUNT RELATED METRICS

• Number of Lines Of Code (LOC)• Number of classes• Number of comment lines• Number of interfaces• Number of modules• Number of statements• Number of variables• There’s so many things you can count!


36

MORE SIMPLE METRICS

• Comment density: number of comment lines / number of all lines

• Fan-in: number of other classes(module,etc.) using this classes(module,etc.)

• Fan-out: number of classes(module,etc.) such that this classes(module,etc.) uses them.


37

COHESION–RELATED METRICS

• Cohesion means how well parts of some unit – say class – belong together.

• For instance, it is possible to check if methods use the same variables. If they do, they seem to have something in common.

• A number of cohesion-related metrics exists.


38

COMPLEXITY METRICS

• McCabes cyclomatic complexity:If G is the control flowgraph of program P, and G has e edges (arcs) and n nodes, then Cyclomatic number V(G) = e - n + 2

• Halstead metrics – based on the number of operators and operands.

26.3.2003Software Engineering 2003 Jyrki Nummenmaa 1 SOFTWARE PRODUCT QUALITY Today: - Software quality - Quality Components - ”Good” software properties.

Documents