To support various research efforts underway that involve capturing low-level developer behaviors and inferring higher level activities from them, I have put together a proposal for a new "DevEvent" sensor data type, which is intended to help us move toward common, shared infrastructure for at least the low-level side of things. In other words, it seems like we should be able to collaborate on the design and implementation of the sensors, even though the higher level analyses may differ radically.

Hongbing and I have worked together to illustrate the low-level events and high-level activities inferred from them for the case of the TDD research project. What I'd like to request is that Sebastian et. al. provide at least an initial sketch of the low and high level for debugging, and that Lorin et. al. provide at least an initial sketch of the low and high level for HPC. (Of course, submissions from others are also welcomed). I will fill out and recirculate this document as additional stuff comes in.

If we can settle on what the DevEvent SDT looks like and the kinds of data it is supposed to generate in the near future, then we can start redesigning the Eclipse/Emacs/Netbeans/Ant/etc sensors to generate conformant DevEvent data. That will enable all of us to get a big jump on generating the high level inferences we are ultimately interested in.

Please take a look and let me know what you think and when you will be able to provide a first cut at your contributions. If you find it difficult to submit conformant example data, that's extremely important for us to know ASAP!

The initial draft of the initial spec is at:

<http://hackydev.ics.hawaii.edu/hackyDevSite/doc/DevEvent.html>

Enjoy.

Cheers,
Philip

Reply via email to