diff options
Diffstat (limited to 'tex/thesis/introduction')
| -rw-r--r-- | tex/thesis/introduction/introduction.tex | 45 |
1 files changed, 40 insertions, 5 deletions
diff --git a/tex/thesis/introduction/introduction.tex b/tex/thesis/introduction/introduction.tex index f59f30f..ffc6cd2 100644 --- a/tex/thesis/introduction/introduction.tex +++ b/tex/thesis/introduction/introduction.tex @@ -2,19 +2,54 @@ \section{Motivation} -Software bugs are bad. Bugs have significant costs in terms of time, -money, security and lives. Hailpern and Santhanam state that -debugging, verification and testing can easily range from 50\% to 75\% -of the total development cost of a +%Main points: +% - software is taking on more important roles +% - software bugs are dangerous +% - software bugs are expensive +% - we have to minimise the number of bugs we have + +In today's increasingly technological world, software bugs can have +significant consequences, ranging from the relatively minor +frustration caused to average users to causing deaths. There have been +several incidents in recent years in which a bug in a software system +has led directly to injury or death. X-ray machines which provide too +high a dose of radiation, cars which continue to accelerate against +the driver's wishes and other dangerous situations have all come about +as a direct result of software bugs. + +Software bugs also have significant financial costs, with Hailpern and +Santhanam static that debugging, verification and testing can easily +comprise 50\% to 75\% of the total development cost of a system\cite{Hailpern01softwaredebugging}. +In order to limit the number of bugs we have it has become commonplace +to employ sophisticated approaches to testing. Many different types of +testing are done to attempt to ensure that software is bug-free. These +tests, although extremely useful, are inherently incapable of +\emph{guaranteeing} that software is free of bugs. This is especially +a problem in critical software systems, such as those found in +aeroplanes or large industrial machinery, where software failure can +have catastrophic consequences. + +In order to provide a guarantee that software is free of bugs we must +in some way \emph{verify} the software before running it. While it is +possible to write programs in a way that is easier to verify, it is +still a difficult and expensive process to provide verification. + +Recently there has been work done in automatically analysing programs +in order to identify bugs. This work is broadly classed \emph{static + analysis} and this thesis contributes to this work. + + + + \section{Contribution} In this thesis we present an implementation of the strategy-iteration based static analyser presented by Gawlitza et al.\cite{EasyChair:117}. Our implementation has several enhancements which significantly improve the practical performance of the analyser -on real-world data. +on real-world programs. Theoretical contribution: \begin{enumerate} |