Sites on a Plane

I’ve just returned from a week’s getaway to Tokyo. While I was appreciating the cherry blossoms, it seems our site-specific browser concept has been replaced by the new hotness: desktop-specific sites!

I think I saw this coming when I first tried Locomotive, a self-contained application for hosting Ruby on Rails apps on Mac OS X. Locomotive is geared towards developers, but once you see how polished it is, it’s not a big stretch to imagine something like it running on end-user Macs. Then, last week, two applications appeared to seal the deal. Joyent’s Slingshot is for deploying Rails apps on end-user desktops, both Windows and Mac, and Tom Insam’s DjangoKit hopes to do the same for Django apps on Mac OS X. (Hat tip: Daring Fireball.)

So. Now that it’s possible to create desktop apps with Rails and Django, one question is inevitable: Why would you want to?

Since you can already develop real desktop apps, even real Cocoa apps, with both Ruby and Python, choosing Rails or Django boils down to one thing: you want to develop your user interface with web technology.

And why not? Ajax has brought so much interactivity to HTML-based user interfaces that it’s almost possible to mistake a polished specimen for a real desktop application. But it does not follow that Ajax is a good way to go about building a desktop UI. Hell no. As Dave Thomas says, it’s just lipstick on a pig. The lowest common UI denominator of all the computers in the world, and hellish to develop.

The only seemingly real use case for Rails and Django on the desktop is the prospect of offline versions of popular web applications — Sites on a Plane. But 37signals, the canonical web app shop, is not convinced that the demand is there. David Heinemeier Hansson mentions the demise of Boeing’s Connexion in-flight Internet service as proof of this:

Ironically, SAS killed the internet access on their transatlantic flights this January because nobody was using it. (Well, except for me saying “look, I’m online at 30,000 feet!!” in a chat room). And I think that’s a good indicator for offline web applications. The idea is cool, but the reality is that it just doesn’t matter. You don’t need access to all your stuff all the time. We’re already overloaded with connectivity. Cherish the few remaining strongholds for offlineliness!

While the example is not good (Boeing killed the service because they failed to sign on enough airlines — not users — to break even), David is right. While web access is still not as ubiquitous as it could be, the situation does not call for moving apps from an online platform to a less connected one.

The right way to improve the usability of web apps in connectivity-poor environments is to enhance the way browsers deal with unreliable and low-bandwidth network connections. The next major Firefox release should be a good first step in this direction.