Layered development
Sorry for the lack of updates in the last few days. I know that I promised a first release of NETnotifier by the past weekend, but I wanted the code to be of good enough quality not to be thrown away immediately :)
So, while the code is 95% complete and some beta testing is happening, I've started doing some code reviews. At the beginning of this project I laid down a vague plan and sketched the overall architecture and the database schema on a couple of pieces of paper. From there I figured out the class hierarchy, wote down the classes names in a text editor and started adding methods and properties in a hierarchical way. After a bit of manual refactoring, I had a pretty clear view of the class hierarchy and started implementing it.
Of course the first method of the first class is wonderfully documented, with comments on every line, each variable painfully described, etc etc...
The second method wasn't documented that good. I got into "the zone" and couldn't stop coding because I wanted to see this fairly involved application do something.
This means that the amount of documentation decreases in a hyperbolic function from the first method implemented to the last one. AKA, the classes from the third to the last one are pretty much non commented.
So I have to comment the code. While I'm at it, I thought it would be nice to be proactive and figure out security risks right from the beginning. Problem is, I've never done this and I have no idea where to start from. Is there a methodology to follow? Or is just pure talent and luck that make up a code reviewer?
Anyways, the system is looking good and it's going to be out in a matter of days. I suppose Valentine's out of the question, but by Sunday you will be able to subscribe to a couple of services and receive alerts directly on your desktops.
This, of course, if nothing really bad happens :)