So this week I met with someone who was very excited about working with Google Dart, a web programming language aimed at being a better JavaScript running both on the server and ultimately the client where the VM could run in browsers (only chromium for now). When Google first announced Dart in September '11 I thought "cool, too bad it will never work" and basically dismissed the project as doomed since it would presumably face too much pressure from competitors and open standards. After the failures of Flash, Java (applets) and most recently Silverlight it's hard to imagine Google having an ability to convince the world to abandon JavaScript in favor of Dart. Even GWT seems to be losing Google's attention somewhat. Having just been burned my MSFT on Silverlight I am overly wary of any proprietary tech these days.
But what I overlooked when I initially dismissed dart was the dart2js compiler and the fact that Dart doesn't really need the Dart VM to succeed for the language to get some traction. So skepticism temporarily put on hold I decided to port a small existing project from jQuery to Dart. The project was pretty simplistic, about 175 lines of Javascript served up statically and allowing users to navigate a full page grid with vim-like keybindings, entering free flowing text in "contenteditable" divs within the grid.
Some thoughts on the experience:
.....
But what I overlooked when I initially dismissed dart was the dart2js compiler and the fact that Dart doesn't really need the Dart VM to succeed for the language to get some traction. So skepticism temporarily put on hold I decided to port a small existing project from jQuery to Dart. The project was pretty simplistic, about 175 lines of Javascript served up statically and allowing users to navigate a full page grid with vim-like keybindings, entering free flowing text in "contenteditable" divs within the grid.
Some thoughts on the experience:
- Dart editor based on eclipse is handy and the integration with chromium made the whole experience of getting up and running with the SDK really easy
- Classes! Optional Types! (or ?) Libraries! Isolates! Lexical Scoping! All very nice and immediately familiar in the resulting code.. although I did find myself momentarily confused by:
- lack of object notation (use Maps instead)
- lack of dot access to those map properties
- lack of explicit private keyword (use _prefix to indicate private)
- lack of string concatenation (use string interpolation, which I actually really like)
- handling of concurrency is really interesting not just in the Isolates but also in the "Future" type which makes for some really clean chaining of asynchronous calls such that you are not nesting N levels of callback functions. Didn't use them but they looked cool.
- There are equivalents in the "dart:html" library for a lot of what you'd commonly reach for in jQuery, but in general I found myself writing more verbose code than what I was doing equivalently in jQuery. That said...
- I had WAY fewer error cycles in Dart, and the semi-static typing made the whole conversion process take only a couple hours including all the time spent looking things up..
- ... and looking things up was pretty easy, you can tell it's largely from Google but they are investing enough in their docs that it was always pretty easy to get what I needed
- A nagging feeling... While dom manipulation was straight forward I found myself cut off from Javascript. It looks as though calling between them is totally doable but feels hacky as soon as you go down that path. Dart itself assumes just enough responsibility that leaving the core libraries feels like foreign territory effectively isolating you off from all of the innovation happening out there in Javascript land. I may be making a mountain out of a molehill here but having to choose between Dart native libraries and Javascript community for common tasks is not pleasant.
So all in all it was a really positive experience, the tooling is solid, the language itself is very comfortable for someone with my background and I can see clear advantages especially on a larger project. But I'm now at crossroads on this little project, do I go with the comfortable tools and language with questions around the long term viability and community support for the proprietary Dart (to be fair it's open source, but open source doesn't remove the fact that this is very much a Google initiative) or do I stick with the "open web" and just deal with all the pains?
After a couple days and writing this out I decided I just can't commit to Dart yet. There's too many exciting developments happening in the open web and Dart feels too much like unnecessary fragmentation of the collective efforts to improve things. Will be keeping an eye on this though, just because it's cool. The project's ambition is admirable and it feels like they are just getting started.