Launch

Long stretches of time without posts are good things. It means I'm working on coding rather than blogging.

I suppose I should've put up an entry about the launch when we launched, but a.) I was too busy and b.) there really wasn't a clear launch date...we sort of slid into it, and we're still sorta sliding into it. This just happens to be the first time I've got time to blog instead of code. So I'll try to reconstruct what's happened since the last entry, almost a month and a half ago. There's bound to be a useful lesson in here somewhere...

We went live on Tuesday, June 26. "Went live" as in "Okay, password protection is off, you can tell your friends about it." I suppose our official launch date was July 1st: that was when Mike put one of our games on Digg to test the waters.

As of the last post (May 31), we were "Just about to launch - this is the last feature we need, really." This seems to be a common feature of startups: FictionAlley was "ready to launch in a couple of days" for about a month before it actually launched, my day job has been "ready to launch" for the last two months but with little actual activity on that front, and Diffle was "one day away from launch" for about 3 weeks. So it's worth looking at the Subversion commit log to see where that time was actually spent.

On May 31, I was working on auto deployment. I didn't actually commit those changes until Jun 2, my birthday, and then it's only "beginnings of auto-deployment". I finished (with a big ? in my commit log message) on June 5, then we had to fix bugs in the main production script that only showed up after deployment. June 7 involved adding the ability to change passwords (oops! how could I forget that), June 8 was some bugfixes, and June 9 began a pretty page for handling internal server errors, which I didn't actually finish and debug until June 12 (there were no commits between June 9 and 11 - I was at Chris & Jess's wedding).

The next week was all bugfixes - there were a lot of them, apparently. I think I actually removed the password protection on June 20 (the same day I added the custom 404 not found page - there are 8 commits that day, I must've did a whole lot of other bugfixes too). There are a couple fixes there that IIRC came from suggestions by Mike's girlfriend Sandy, so it would've been open to at least her. Then I had to convert all templates to precompiled Cheetah templates to fix a unicode encoding issue - this was the last bug before launch, I discovered it about an hour before we were about to go live and decided it was enough of a showstopper to postpone the launch. It took about 2 days to convert everything to precompiled templates, then we went live on June 26, though we didn't actually tell anyone about us.

I went on vacation - no Internet access - for the 4th of July week, which was one of the main reasons we didn't actually tell anyone. Spent the time prototyping the game creation engine - I should jot down some thoughts on that soon, because they're undoubtably wrong. Mike dugg a game anyway, to see the response. We ended up getting like 700 unique visitors from that, but the site really isn't sticky. The vast majority (like 93%) came and left without checking anything else out. Since then, we've been holding steady with about 20-40 uniques a day. I suppose that's somewhat encouraging, that at least *somebody* checks us out daily, though it's not much. Mike and I both think we really need the game creation engine in order to take off.

Aside from that - we had to switch from Cheetah to Mako because Cheetah offers very poor handling of AJAX fragments. There's no real facility in Cheetah for libraries of mixin HTML fragments: basically, each def needs to be either on the page or in a parent class. Mako is much more flexible, with both inheritance and namespaces, and also seems to have more robust unicode support and an easier API.

I should start the main Diffle announcement blog so it's not just a blank page. It's at diffle.blogspot.com, and will be the public face for announcements and status updates and such. This will continue to be more internal reflections and so on.