The Bootstrap Chronicles

Imagine that a web developer was frozen in ice in 2004 and thawed two months ago.

Screencasting. HTML5, CSS3. Sprites. CDNs for the masses. Google Gears and AppEngine. Amazon EC2, S3, SimpleDB.

Flash has come and gone (pretty much). JavaScript suddenly sucks less.

Chrome, Safari, Firefox. Web standards war has largely been won. (At least Internet Explorer still sucks.)

The list goes on and on. The last time I built something on my own, the world was a very different place. I’ve been aware of the changes, but in many of the things I’ve mentioned, I haven’t remained as steeped in the details as I once was. One of the benefits of not bootstrapping alone is that there’s often a team around you to help shoulder the load of understanding all the different parts of web development. My focus of late has been on the architecture side of things, so I’m quite rusty on building client-side code efficiently.

Not anymore. The last couple of weeks, I’ve spent my evenings and weekends getting current. Somewhere along the line, I saw an article from 2006 or so that posed the same question as the one above, only the year-of-freezing was 1999. (Sorry, I’ve since lost the link to that article!)

Back in the post-Bubble 1.0 days, I spent a lot of time integrating quality OOP and unit testing practices into my personal routine. Back then, people were already saying that it was difficult-to-impossible for a very small team to build a competitive website: there is simply too much to master. Today, oddly enough, there’s as much or more to master, but people don’t seem to be talking as much about how it’s hard-to-impossible for a small shop to be successful. That’s either because no one’s even considering it anymore, or it’s somehow a lot easier.

I’m not sure it’s either. It’s certainly no easier. While a lot of the tools I mention above make the process either, there are vastly more decisions to make. Which JavaScript framework to use, which CSS version to target, whether or not to build an iPhone-sized version of the site. Google AppEngine or Amazon Web Services … or one of the many wannabes? HTML5 or XHTML 1.x? (I chose HTML5, and man, is it weird to NOT be hassled by the W3C validator about a variety of things I worked hard to break bad habits on in the HTML4-to-XHTML transition.)

Clearly it’s possible for one and two person operations to get something off the ground — too many people have been doing it lately to ignore. But navigating the choices is just as tough as its always been, if not more so.

Where have I wound up? Well:

• The dev markup is fully HTML5 compliant.
• I’m using some CSS3 stuff. IE users can switch to something real. (Note to Mircosoft: live in the now, people. I can’t believe you still suck so badly, after two major browser releases. YOU. SUCK. If you can’t see that developers aren’t going to kowtow to your non-open-standards shit by now, you should switch back to fleshing out WebTV.)
• I’m using jQuery+Appcelerator Entourage. I never could have imagined that JavaScript could be so much fun. Yes, fun.
• Google Gears LocalServer + Amazon CloudFront means this is the fastest website I’ve ever built. I finally decided on Amazon Web Services for the hosting. The auto-scaling and load balancing stuff is just too cool.
• Nearly every image (the few of them that there are) on the new site is a PNG with alpha transparency.

The timeline has shifted. Hell, the whole product has shifted. I’m NOT building what I was planning to build when this blog started last month. Turns out that the market I was originally planning to get into is just way, way too crowded. So now I’m onto something in the same neighborhood, but on a different street, so to speak.

And finally, the guy I mentioned earlier who did my logo after I dropped 99Designs has signed on as Creative Director of this venture. I couldn’t be more thrilled, as one thing that’s crystal clear is that if you want a chance to succeed, your app had better be useful and beautiful.

UserVoice is still a great decision, and we’ve got that integrated. ProtoShare, as much as I love it, proved to have too steep of a learning curve for such a small team. We opted for telephone and IM to iron out designs (he and I don’t live in the same state). Unfuddle is helping a lot, and I can’t say enough positive things about Google Apps for Enterprise. I also settled on MailChimp for managing the announcements list.

Launch plan: mid-July sometime. Stay tuned.