An amazing first 72 hours
Initially, we didn’t give out any passwords for real FluidDB API access. That was partly to give ourselves a chance to recover, partly because we wanted to take it slowly to see how FluidDB usage would go, and partly because we really did launch early – we have a lack of monitoring and admin tools to prove it.
On Monday night I think I was more mentally exhausted than ever before. Apart from having very little sleep, the emotional side and stress factor was high. I’d told many people we would launch on the 17th, and really didn’t want to have to postpone. On the weekend Esteve and I had found two problems, one seemingly serious, and fixed them. I built the final Amazon EC2 images on the Monday morning. Esteve had been putting in long hours to roll the 0.3 version of txAMQP. Xavi was putting the final touches on the web registration process and we were moving domains and URIs around hastily after realizing there was an overlap. It was all pretty calm, but there was a lot going on.
On Tuesday I figured I’d take it easy, just keep an eye on things, and even take the kids to the pool. But it didn’t work out exactly as planned… We found that people were getting into FluidDB, and exploring using the anonymous user (the user you become if you don’t send a username). The anon user has very limited abilities, but can do a few things. Several people had shown up in the #fluiddb IRC channel we’d made and were discussing how to program to the API, and what they had learned so far.
And we were stunned to see that within 12 hours of the release, Seo Sanghyeon in Korea had blogged about us, had written a small Python library for talking to FluidDB, and had even written glue to make a FUSE filesystem backed by FluidDB. Who is that guy? we all wondered.
Since then it’s been one great surprise after another. We’ve been having 20-35 people in #fluiddb on IRC. Four client-side libraries have been written or begun (for Common Lisp, PHP, and at least a couple for Python). The people on #fluiddb have pointed out a few shortcomings of our HTTP REST interface. They’ve run into several (small, thankfully) bugs. They’ve helped to improve the documentation in multiple ways. They’ve been teaching each other. They’re sharing code. They’ve helped clarify RFC2616 and our sometimes marginal interpretation or plain misuse. They’ve discovered and intuited most of the internal organization of FluidDB. We are now using the client libraries written by these incredible people from all over the world; they’re in Korea, South Africa, Italy, the US, Uruguay, Spain, Canada, France, the UK, and probably several other countries. It’s amazing.
There’s also been a ton of support and a lot of nice comments on Twitter and Friendfeed. Some examples: 1, 2, 3, 4, 5, 6, 7, and 8. You can search Friendfeed or search Twitter to see a whole lot more.
I couldn’t have imagined a nicer way to launch FluidDB. Hanging out in IRC with a bunch of smart, enthusiastic, and energetic hackers going over the ins and outs of FluidDB is like a dream come true. It makes me smile to think of the number of times I’ve been dubiously asked why I was so sure other programmers would like FluidDB. So if you’re wondering what FluidDB is all about and how to use it, please head over to #fluiddb on irc.freenode.net and say hi. And to everyone already there: thanks so much