Hello all,
The following writing is my attempt to suggest a real-world application for Genode that "philosophically" gives more "responsibility" to the OS/framework at the expense of application-dominance.
Starting in the late-90s, I felt that the OS should be closer to the applications (such as web-, data-, and logic-servers in opposition to the movement, which was to disassociate applications from OSs). My job was maintaining both OSs and applications (as an administrator) and I saw efficiency by blurring the differences between them.
My OS model (called the Thinman) was a single OS/application that would have libraries/modules that would be "called upon" and brought to windows on the desktop that had "browsed" for material such as pictures and edited the material using tools (libraries and modules). Of course, new material could be initiated with the "New" button.
More recently, I have personally concluded that the SVG, or Scalable Vector Graphics page, should be at the heart of the future Web, and the WC3 seems to agree by proscribing full HTML capability within SVG pages. Thus, nearly anyone can easily make an exceedingly attractive page (far more so than FaceBook or Google+), and then paste in the written material as basic html text.
Given this, I tend to use Inkscape (IS) SVG editor a lot. Previously, I used GIMP, but GIMP is an image editor, not a page editor. So, today, I use both, primarily to cover IS's weaknesses in graphics manipulation. As you might predict, the interrelation between the two systems is awkward. You can "cut" from GIMP using the mouse or "ctl" key combinations, and "paste" to IS, and the pasted image can be manipulated, but cannot be saved as an SVG page. (I use snipping tool to make web images).
In light of my "native" OS beliefs, I "fantasized" a solution that has an OS-based graphic image editor that I call Object Manipulator (OM). OM takes the input stream from the mouse (or ctl-keys), creates from it an object, and then applies the libraries of tools and filters from GIMP (and other sources) directly to the object, including all the cutting/selecting tools. The image, when fully altered, is then selected and "pasted" back into IS, or other format-specific editor, to be made into a presentable product.
Given this model, there is no reason the "pasted" object has to be a graphic; it could be a poem, and the libraries brought to it might include a thesaurus. Thus, OM has no specific purpose, except to bring together objects and libraries (which are technically also objects, but that is a different discussion).
OM has no connection to "reality" as its space is "an abstraction." It only works in the context of the OS and has the desktop as its "background" in ways no different than a repairman's bench-top is the "background" for fixing/developing some device such as an appliance. It does not even intend to use the filesystem to communicate with applications (though of course it can by converting to established file formats or by creating a native one).
The object might not even need a background, or even an application called OM to call it; it may be defined in of itself as a selected object, perhaps defined in the context of, or possessed by, the mouse.
I am hoping that this tangible example (that I emulate using IS and GIMP) can help provide a real-world basis, or context, for the Object OS layer that you are attempting with Genode. Implementing this idea involves subtle philosophic/holistic issues that are at the heart of an "agnostic" OS framework; that is to say that given circumstances either technical or political, a kernel can be switched out at the bottom, or, likewise, and object manipulator can be switched out. This suggests that the OS framework must be impartial and democratic --a sort of United Nations of OSs and desktops.
I hope that this writing is comprehensible, and I will continue to attempt to simplify it especially for a wider audience.
Regards, John