Google SoC Idea: web-interface library

[Matej Svejda]

> Why create a new server? There are plenty of web servers around.
Well, I think that creating a really small server myself would be easier than 
trying to integrate Kross into an existing server (but I might be wrong...).

> >   * connect this web server to use Kross for executing scripts. This
> > would give it instant ability to write code in Python, Ruby,
> > JavaSciprt (?) and Java
> What would be the usage scenario for this kind of language connection? How
> would somebody wanting to provide a web interface for his KDE application
> make use if it?
Developing web-apps in scripting-languages like Ruby or Python seems like 
the "natural" thing to do (that's how most of the current web-interfaces 
work). But the problem here is how to communicate between C++-software and 
those languages.
This is where the whole integrating-Kross-into-the-server-idea kicks in: 
Accessing C++-objects in those scripting languages would become as easy as it 
is now when you write scripts for Kross-applications. You would just need to 
call something like...
	serverInstance->addObject(object, "ObjectName");
or ...
	Kross::Action* action = new Kross::Action(this, "WebInterface");
	action->setFile("script.py");
	action->addObject(object, "ObjectName");
	serverInstance->setAction(action);
	serverInstance->trigger();

Then you could do anything you like with that object in your web-interface 
script.

> What would adding objects and module actually mean?
Adding them to the script so they can be accessed as if they were "native" 
objects in the scripting language that is being used.

> I understand the idea as a way to provide a way for existing KDE
> applications to get a web interface, not the other way around, coupling
> existing web applications to KDE. So in what way would PHP be relevant
> here?
It would give the programs the ability to use frameworks such as CakePHP or 
Prado. But that's not really required.

> It depends on the concrete goal of the project and how it's approached. As
> a generic solution it sounds like a pretty tough project to do in the
> time-frame of the SoC. This of course also highly depends on your
> background and how much you would have to learn to be able to become
> productive.
Well, I have moderate C++ skills but I have lots of experience with 
web-development. I think having a proof-of-concept implementation and a port 
of one of the existing interface to this library by the end of the summer 
(without the PHP-backend for Kross) should be doable. But again, I could be 
wrong...

Matej