FM Thesis

Well, I spent some time thinking about it tonight and I banged out two topic ideas. Since I'm having trouble deciding whether to go for a more theoretical/academic approach or a more research oriented topic regarding industrial applications, I worked up one topic for each.

On the industrial applications side, I'm thinking of a two-pronged analysis of various formal methods and case studies. On the one hand, I would examine various industrial projects that used formal methods, such as CICS, or SHOLIS, or whatever. Praxis has a lot of decent case study material, and I have a number of other case study papers. The idea is that I'd examine the project, the formalism used, and the results. I would then try to draw some conclusions about the efficacy of the methods used. The second prong would be to try to analyze how those conclusions would apply to other types of projects. For example, since most formal methods projects are large projects in safety- or security-critical domains, I would try to do an analysis of how they could be profitably applied to, say, small to medium sized business projects. If it was required/warranted, I could even attempt such an application, i.e. trying to design a simple business system while applying some "light-weight" formal methods.

Because my interests run towards the practical, my "theoretical" topic is actually more of a project. I have a number of papers on formalizations of programming languages and ways to formalize various language features, such as recursion or exceptions. My idea is to actually apply this to define a formal semantics for a meaningful subset of a widely used language. I'm thinking either Python or PHP might be good candidates. This adds a little novelty to the topic in that both are dynamic languages, i.e. variables are created and typed dynamically, which I've never seen before. PHP also has that array/hashtable hybrid type, which could provide for an interesting treatment. I have material on most of the basic aspects of imperative languages, so this would essentially be an attempt to aggregate that research and extend it to a language in practical use.

I'll write that up in a more formal way and send it to Roger tomorrow. For now, I'm going to get a glass of water and go to bed.