Thesis Idea: Programming Language for Software Engineering Research

September 16, 2008

It is difficult to make scientific conclusions in Software Engineering because of the number of variables involved in the production of software. To make a causality conclusion, we must hold variables constant and vary just one.

There are many open questions in the design of programming languages. Are dynamic languages better than static? (In what circumstances?) Are variable declarations before usage beneficial or wasteful? (In what circumstances?) Are REPLs helpful? (In what circumstances?)

A study meant to compare these today would need to use two different programming languages since every programming language currently in existence has implicitly made a decision about the questions above.

Thesis idea: Create a programming language interpreter that allows each of these features to be switched on and off, enabling more scientific research on these issues.

Advertisements

One Response to “Thesis Idea: Programming Language for Software Engineering Research”

  1. DeepBlue Says:

    Well, there are fields where you have to draw conclusions without being able to separate the variables and hold all but one constant. One good example is epidemiology (http://en.wikipedia.org/wiki/Epidemiology). Perhaps you can study their techniques for coping with the problems you describe and apply the techniques to software engineering.


Comments are closed.

%d bloggers like this: