language parallax

I’ve been seeing Krzysztof Czarnecki’s name around for a couple of years, but only recenly started reading his work.

His paper “Framework-Specific Modeling Languages with Round-Trip Engineering” pdf (actually M. Antkiewicz is the first author) is something that caught my eye in particular.

From the “Related Work” section:

Domain-Specific Modeling Languages (DSMLs) and frameworks. The idea of putting a DSML on top of a framework is not new. [… But] we are not aware of any work exploring FSMLs with round-trip engineering support.

[…]

Framework instantiation. Most approaches in this category only support forward mapping to code. […]

That’s great validation that what seems like an urgent problem hasn’t had much attention to date.

Round-trip engineering. According to Sendall and Kuster the main difference between round-trip engineering and forward and reverse engineering is that round-trip engineering takes both artifacts into account with the intention of reconciling them, whereas forward and reverse engineering typically create new artifacts, potentially replacing the old versions.

Again, that’s great news. It’s good to see someone make this distinction. The latter (round-trip) is definitely the one that needs serious attention.

Finally from “Discussion and Future Work”:

FSMLs can potentially be used for semi-automatic code migration as described in Section 5. Although we do not have any practical experience with this aspect yet, we think that the specialized forward mappings can be defined for different versions of the API, or even for different frameworks. Currently, we are investigating the possibility of using FSMLs for the migration of code from the Struts framework to the Java Server Pages framework.

They’re looking at Web App Frameworks — and tackling framework-level migrations. Both of which are interesting to me and are a part of the project I’ve been working on recently.