21.MyISERN-1.2

Distribution of system available Here

Project home page is Here

All tasks were completed.

Difficulties
There were numerous difficulties in this section of the project. At first glance, I thought it would be rather easy, yet it turned out to be quite challenging. The annoyances of many of the components made getting the final project working quite difficult.

At our first meeting, we began discussing the UI and possible cases we needed to consider that might happen. It turned out that there were so many strange cases that could occur that the UI and underlying error checking system had to be quite complex. There were so many cases to consider and messages to give the user that just the UI because quite monstrous.

Once we had a partially working system, I began playing around with some of the data and trying to run as many cases as I could. At this point I realized how unforgiving JAXB and XML can be. Several times in this process the JAXB marshalling methods corrupted the XML files. It turns out that the JAXB Marshaller will sometimes write a format that will cause the Unmarshaller to crash with an exception. Our error checking mechanisms within the system had to be very robust to ensure this type of problem never occurs.

The testing of this system was a real problem. Since so many components of the additions in 1.2 rely on user interaction from System.in, I was unable to write many JUNIT test cases. Much of my testing was done by simply using the system and trying to break it. Entering the data turned out to be an exercise in breaking the system as well. On several occasions the system created corrupt XML files after I had entered a rather large amount of data, revealing a new bug. The SVN revert option was quite handy.

Group Organization
As we had done for our earlier version, myself and my teammate Shaoxuan Zhang met at the library for a long hacking session. After this 3 hour meeting on Thursday, we had most of the system complete. After that, we simply stayed in contact with email and Skype. It turned out quite well, since most of the data entry and debugging was difficult to work on as a team.

Future thoughts
At this point, the program is already becoming very large and somewhat difficult to manage. I think the incremental development we have been doing with 1.0, 1.1, and 1.2 made the design process difficult. Now that we have a more clear idea of how the entire final system will be, we can design a much more efficient system. There are many components in our current system that may not be used in the future. The slimming down of the system will make it much less buggy and more easily manageable in the future.

Lessons Learned
JAXB can be a pain, especially since the Marshaller and Unmarshaller do not behave the same when errors are present. I also learned the importance of early and complete design to keep the final system manageable. Finally, I gained some more experience with eclipse and the debugging features it provides.