While I'm not a UI person, I like html/css/js more than most other GUI toolkits I've used. The notion of using these for presentation in desktop apps gets me pretty excited. While there is something to be said for conforming to a platform's GUI standards, most of the GUIs I've been impressed with, e.g., Picasa, Lightroom, many web-apps, and Chrome use non-standard GUIs.
If SeedKit programs are easily portable to windows/mac, this might be a nice open-source replacement for Adobe AIR. Either way, I'll definitely give this a look next time I build a Linux GUI.
The kit, it is hoped, will lower the ‘entry barrier of Gnome development to web developers’ making the Gnome platform a more attractive proposition.
I might be the only one, but I totally facepalmed after reading this.
Maybe this kit will help them to realize that GTK isn't actually that hard if you just RTFM. Even more so if you use any of the Python, Perl, Ruby, PHP, . bindings that make it ludicrously easy as is.
Good work though. Imagine it was pretty tough work getting this piece of kit out the door.
I've been working on hybrid web/desktop apps for a while now. I use Qt's webkit to make the application window a HTML page (QWebView) and load pages bundled with the app or from a remote server (or mix content from both). Whenever I need to reach out to the client platform--in my case, for bluetooth and for app.exit()--I use Qt's javascript/native bridge mechanism for great justice.
As with everything there are tradeoffs. Benefits include reusing libraries for both browser-based and native-based applications portably across client platforms. I have a large library of server code for the kinds of applications I write, so knocking up a new application in that domain is easy. I find writing clients in HTML/javascript easier for my iterative development style. Client libraries like jQuery help too, although I don't use them much as my apps run on an embedded device (333MHz) where they slow down page rendering too much. I use them for mocking up and prototyping and rewrite to inline javascript for deployment.
The biggest drawback is the native client, and I try to avoid having to use one as much as I can. And while Qt is good for Windows/Linux/Mac, you need something else for Android (I can't comment on iPhone/iPad because you can't program the bluetooth on them, so it is useless for my particular requirements). So you'll be maintaining separate clients which can be a bitch, but such is the nature of tradeoffs.
On the whole I much prefer this style of application, playing around mostly in HTML/javascript/perl, dropping into C++ when I need to touch the platform. Originally I thought using a browser might be too slow on my constrained device, but webkit is quite pleasingly snappy--faster than VB for an application one of my hybrid apps replaced.
That's a great concept! User interface design has to enable rapid turnaround in order to be improved iteratively. I wonder what are the possibilities of coding the business logic API in C, and communicating with it from an HTML5 front-end.
