Burt Wrote:
Firstly, how will this be implemented? Will this be a new SDT or a reimplementation of the issue SDT?
I would hope that the SDT should be able to stay the same. The SDT shouldn't really be coupled with one particular implementation of a sensor. Whether the sensor is sending data in snap shot mode or in any other mode it doesn't matter.
I suppose in terms of my thesis research I could just manually examine the issues in Jira so an SDT that only collects # of open issues would not be a problem. Or I could change my thesis topic but I'm worried I might not have enough time.
Manually!? I would think that one of the main points of your research is to _not_ do anything manually (except maybe the classification of the issues). I realize that your thesis could be conducted solely on manual investigation but one thing we should have all learned is that manual investigation will not be used in practice, "You can't even get them to push a button". A manual feedback loop will never be used. An automated feedback loop is better. Hackystat provides that automation.
Another concern is that this implementation seems inflexible. What if, in about 3 months, we decide that other types of unsupported status information are desirable? For example, the number of issues closed on a given day, the number of issues actively worked on by a person, the level of discussion happening on a given issue, the average number of updates on an issue, etc. It could be that we decide on a specific characteristic now (currently the number of open issues on a day) and then change out minds a little later.
You're right.. it is inflexible the way you are thinking about it. But, you should understand that the SDT should be designed to be flexible. It shouldn't matter if we are currently collecting data about open defects or closed enhancements. The SDT should be able to support both.
The information that you listed could possibly be in the SDT. In other words I think that the SDT should be flexible enough to support those. My view is that it is up to the sensor to decide whether we care about that information. For example, the FileMetrics SDT never changes yet we are or have been using it for Eclipse, BCML, LOCC and for very different information.
thanks, aaron
