A Rocky Road Turned Hazardous

These recent blog posts seem to me nothing more of a rant about the problems faced during this current project. I believe it to be very much about problems face, but it can also be about how we've meet with the problems and resolved their issues. But the fact is every time resolve a problem, it just seems to lead into another problem. I truly am starting to wonder when the last of the problems will be resolved.

I was finally able to get the build system working just before the milestone last week, but to my surprise when I had run verify, I was slammed in the fact with countless errors. Not only was the code that my group members working on filled with checkstyle errors, but some of it also contained PMD and Findbug errors. With so little time left until the mile stone, we were only able to do so much, and the fact that our coding style was sloppy was made very clear.

To that we have overcome. With everyone now under one subversion folder we were able to quickly share code that was needed to fix some of the problems. The team worked very diligently to clear up over 100 checkstyle errors that had occurred, many of them occurring from code that was unneeded because it was imported from a package.

With the errors gone we were just tasked with one thing left. Running a complete build system for a new developer. It was my first time working with ant really. Although I had used ant before, I had only been using the predefined build system given to us, but for the next step to progress I would have to start modifying and building my own build files. To this I have almost succeeded. In our current project we have two sub projects that needed to be ran separately. To do so it was almost impossible to use the pre-exsisting build commons we had utilized which were wrapped up in a single project. I separated the two projects each with their two targeted build commons. We can now run the projects and build them separately under one unified project folder that will do the consistency checks for the whole.

So close yet, for every problem we fix, there is always another problem that arises. The current issue this week will be to be able to test the project. For some reason, the sensor portion of the project is only able to run under Edwin's computer. So far Alan and I have tried unsuccessfully to run both my build of the project and the old build (confirmed to be working) by Edwin, but to no success. The problem is a missing property file that should be created on start, whether or not this file is created by you or automatically by the code is another thing. I have traced the code to which I believe it is the source of the problem, but it appears that this portion of code is working under Edwin but no one else. Good that it works, but bad that no one can confirm it other than Edwin.