Walt and I did a soft launch today for MouseZoom, our Disney World photo fan site. What a relief it is to finally see that site in production, even if it is, at the moment, a work in progress. I flipped the switch around 9:30 this morning, my time, and it mostly worked. The problems became pretty obvious pretty quickly.
The first problem was a bug where an auth filter mixed with a cache filter caused a conflict, and that tanked the home page and RSS feeds. Whoops. There were minor problems here and there with the photo stuff, but I was able to fix those issues pretty quickly. That the DeepZoom stuff works feels like a miracle now, because when I go back and look at the math involved to make it work right, I have no idea what the hell I did. That's work I started almost two years ago!
The forum was the big unknown, and naturally caused the most trouble. That was, after all, almost 10k lines of code that had never been exercised before outside of unit testing. I didn't test everything I should have, so stuff slipped through the cracks. But really, for one guy with no QA department, frankly I think it went pretty well. The two things that used to be really hard were less hard this time around: text parsing and session tracking. It was easier because of the testing.
The biggest weirdness started with the e-mail verification for new accounts. Some default encoding issues seemed to cause problems with the URL's, and some e-mail software messed them up. That was annoying. There was also a bug in redirecting to the newest post of a topic, which was caused by a test I didn't think to write. The funniest bug was one around session tracking, but found in the context of mixing the forum with the rest of the site. So for much of the day, the online user count on the home page said zero.
I've still got one major big to fix, and I'll try to sneak that in at some point tomorrow. Then perhaps Sunday I'll start going through my photo collection and really put some effort into adding photos. As time rolls on, we'll add new photos to keep it fresh.
Performance looks OK so far. It's hard to say how that will go if there's only a little bit of load (I think we peaked at around 40 simultaneous users today, which was more than enough). That was kind of the point of doing a soft launch, on a Friday. I didn't want to get slammed. Once things settle in terms of code, then we'll work to promote it harder. The forum is the biggest moving target, but it's only a few items away from feature parity to v8.x (that's what's running on CoasterBuzz and PointBuzz).
Overall, I'm pretty happy with how it went, and glad I took the day off for it. I think we even made $7 for the day! If this keeps up, once Walt and I split the profit, I'll be able to treat myself to a burrito!