Authors
Hugo Ribeiro
Abstract
Viewing spreadsheets as a programing language makes it the most used programming language worldwide. In fact some studies performed show that the so called "end-user" programmers surpass the professional programmers by far.
Because of this and the lack of support for abstraction, testing, encapsulation or structured programming, 90% of the spreadsheets in the real world have errors. This dissertation presents an effort to help with this problem.
The main goal of this dissertation is to create a technique that allows us to detect probable problems in spreadsheets, problems called smells (a surface indication that usually corresponds to a deeper problem).
Thus, we first introduce some theoretic concepts like metrics and smells, such as for instance the Functional Dependency Smell that was adapted from databases. We present the study we made, showing the results obtained with the tool applied to a large set of spreadsheets, the EUSES corpus.
Sample
In this spreadsheet several kinds of smells can be observed. All those smells will be automatically detected and flagged by our tool "SmellSheet Detective".
Publication
2011, Master's thesis, Universidade do Minho, November
