I'm getting closer to a forum release, and I'm at the point of calling it feature complete. The last item is a big refactoring thing that I'm not sure I will do yet, because it doesn't serve end users at all. I'm at the point now where I need to do a solid pass at regression testing as much as I can stand. What this usually entails is starting with a new database, going through the setup steps and then testing all of the things. I don't have a formal plan that I can do the same way each time, so that's the first problem. The other problem is that my unit test coverage still isn't very great, and there's an awful lot of stuff at the controller level that is untested (or shouldn't need to be tested, but there's stuff that has leaked in there). And I have good reason for concern, because I tried a clean start today and found that there were all kinds of problems around the new features. Yikes. I've also got a mess of front-end stuff that is less-spaghetti than it was, but I'm sure a seasoned Javascript developer would laugh at me.
I've had the fortunate experience of working with some very passionate QA people. They're really good at it, and many are really good at building QA teams, too. In cases where I've hired these folks, I've mostly given them autonomy to figure things out because I'm not passionate or good at QA. The ironic thing is they've saved my ass countless times, so you would think I would take greater interest in the sub-profession. Maybe it's the ADHD, but I can't plug in the way they do.
That's what makes being the maintainer of an open source project tough, because even if you have a number of active contributors, no one wants to QA the stuff. My QA staff is CoasterBuzz, when I deploy code there and people let me know when it doesn't work. That's not ideal, since it's hardy a systematic review of things, including non-happy-path outcomes.
So here's to my quality quality friends and colleagues over the years. Y'all are the best.
No comments yet.