After actually implementing the switch from svn commit data to DevEvent data, I ran into the following problems:
1. the data is independent in terms of owner, unlike we only care about the last committed owner for commit data, we want to know the different owner for a same data resource. that being say, if we have 2 developers, we have to match the data twice, if 3 developers, match 3 times, and so on. The way I'm thinking to match the data right now is loop thru the files and compare it with the data resouce, which also contains the file name. It seems to me this is very non-efficient.
2. sensor data is "instant", because we can only get the timestamp of when it is modified. if we want to know how long for a developer has been working on a particular file, we need a way to figure out how long the sensor data has been sent for a same file (or resource) continuously.
3. the tree we can display right now is based on the svn repository on the google hosting site. in other words, if people are working on files that has not been submit yet, their effort will not be shown (at least in the way we do it, although the sensor data will collect those info as well). maybe we need a better and faster way of constructing the file directories of the project, instead of rely on httpunit and the google hosting site. well, after all, is the directories tree even a good way of showing the project structure?
these are the problems i have right now, and they don't seem like easy problems that can be solved over night.