Rosetta Studio is about 5 years old. It was started as a little side project, but grew into a popular app within our niche. Right now we are quite comfortable with our release schedule, we take time before actually releasing to analyze and test the software, as a result there are fewer bugs and overall customers are happy. But it was not always this way.
When I joined RS team, well, there was no team. There was Matt, the guy who came up with the concept and there was me. Rosetta then was in v 1.2, so we did not have many features implemented, basically the app was so vanilla, we only had 2 clients. We had lots of ideas, and very little time and patience to maintain some kind of development system that would help is stay on course. On top of coding, we had to do sales and support plus we were expected to help everyone else with their computer related problems (like fixing Outlook and writing Excel macros). All this left very little time for development, and we had potential clients breathing down our necks twisting our arms to get the features they wanted implemented soon.
We were working in a fire brigade mode, there were fires, and we had to put them out fast. There was very little time to think about writing optimized and efficient code. As long as it worked, it was good enough. This might appear as a good model for someone with a new app, features are added quickly, we had releases quarterly, our client base grew and soon we realized that we don’t need to put out fires. We still have a list of suggestions (which actually more like demands then suggestions) and we are working on those, but now there is no rush to get it done, so we are doing it properly now. Matt is gone now, I lead development team, and we got enough coders to maintain a good pace and write quality code. But the stuff that was written in the past is not going away.
Right now we are rewriting Rosetta from scratch. We need to adapt to the new framework, plus Office 2007 and localization and let me tell you, it’s a lot of work. To address the issues of the old code, and believe me, I was thinking about it for quite some time I came up with this weekly code reviews. Now every week one of us takes a function from those old classes and rewrite it in the most efficient way the developer can imagine. Then, the developer has to present his work to everyone, explain briefly what the class and the function is all about, what and why was changed as well as measurable benefits of the change. I thought that first of all we will slowly start changing the code, which when done properly is a good thing, some of us get exposure to classes that we never worked with before and we would share some of or knowledge. To start it, I decided to go first.
Picked up a function, shaved some time on execution, dropped a loop, did some smaller things. Just to make sure everyone is paying attention put some goofy things with appending a space to a string and then trimming the end of it. Everyone caught it right away; I guess the guys are responding to this new idea. We’ll see how will it progress, but basically I would recommend this to everyone running a small development team as a good exercise and learning experience.