The idea is simple: take a browser, cut out the tabs, the URL bar and all the rest of the window chrome and instead load one website at a time. Hence the colloquial name: “site specific browser”.
Michael McCracken deserves credit for inspiring interest in this idea early on with his Webmail app, that in turn lead to Ben Willmore’s Gmail Browser application. Both literally took WebKit — Apple’s open source rendering engine (like Mozilla’s Gecko engine) — had it load Gmail.com and released their apps. It doesn’t get much more straight-forward than that.
For my own part, I’ve been chronically the development of Site-Specific Browsers from the beginning, setting up the WebKit PBWiki and releasing a couple of my own apps, most recently Diet Pibb. I have a strong belief that a full featured rendering engine coupled with a few client side tweaks is the future of browsers and web apps. You can see hints of this in Dashboard Widgets and Adobe’s AIR framework already, though the former’s launching flow conflicts with the traditional “click an icon in my dock to launch an application” design pattern.
This is what Creammonkey did for Safari. And thank god it was open source or else we never would have ended up with today’s release of the successor to Creammonkey called GreaseKit.
Let me step back a little.
When I found out about Creammonkey, I contacted Kato Kazuyoshi, the developer, and told him how excited I was about what he had created.
“But could I use this on Site-Specific Browsers?” I wanted to know.
In broken English he expressed his uncertainty and so I went about hacking it myself.
I ended up with a crude solution where I would recompile Creammonkey and aim it at a different application every time I wanted to make use of a different Greasemonkey scripts. It was tedious and didn’t really work the way I envisioned, but given my meager programming skills, it demonstrated the idea of Site-Specific Browsers with Site-Specific Hacks.
I called this collection of site-specific scripts with a properly crafted Input Manager a “GreaseKit” and let the idea sit for a while.
Some time later, Kato got in touch with me and we talked about rereleasing Creammonkey with the functionality that I envisioned and a new name. Today he released the results of that work and called it GreaseKit.
I can’t really express how excited I am about this. I suspect that the significance of this development probably won’t shake the foundations of the web, but it’s pretty huge.
For one thing, I’ve found comparable solutions (Web Runner) clunky and hard to use. In contrast, I’m able to create stand-alone WebKit apps in under 2 minutes with native Apple menus and all the fixins using a template that Josh Peek made. No, these apps aren’t cross-platform (yet), but what I lose in spanning the Linux/PC divide, I gain in the use of Xcode, Apple’s development environment, and frankly a faster rendering engine. And, as of today, the use of Greasemonkey scripts centrally managed for every WebKit app I develop.
These apps are so easy to make and so frigging useful that people are actually building businesses on them. Consider Mailplane, Ruben Bakker’s Gmail app. It’s only increments better than McCracken’s WebMail or Willmore’s Gmail Browser, but he’s still able to charge $25 for it (which I paid happily). Now, with GreaseKit in the wild, I can add all my favorite Greasemonkey scripts to Mailplane — just like I might have with Firefox — but if Gmail causes a browser crash, I only lose Mailplane, rather than my whole browser and all the tabs I have open. Not to mention that I can command-tab to Mailplane like I can with Mail.app… and I can drag and drop a file on to Mailplane’s dock icon to compose a new email with an attachment. Just add offline storage (inevitable now that WebKit supports HTML5 client-side database storage) and you’ve basically got the best of desktop, web and user-scriptable applications all in one lightweight package. I’ll have more to write on this soon, but for now, give GreaseKit a whirl (or check the source) and let me know what you think.