This paper presents a window manager whose purpose is to enlarge the windows space available.
The main focus is therefore Windows, not files. Windows represent documents and activities though.
The authors want first to understand the reasons that constraint desktop use. Their analysis concludes that:
(1) the high overhead of moving and reshaping windows that users must often suffer with overlapped window systems derives from a severe screen-space resource contention lying just below the surface, but that
(2) we can partially overcome this resource contention by designing a virtual workspace manager that exploits the statistics of window reference.
They compare windows contentions on the screen to page faults in the memory.
"Evidence for locality of window reference seems clear enough from protocols we have been able to examine of users interacting with an overlapped window system. In the first place we have informally observed such behavior: Users are often seen to use a group of windows for a while, then delete or shrink most of them, and then begin building another set. More systematically, the distribution of window interreference intervals (the number of window references between two references to the same window) is shifted heavily to the low end (compared with what would be expected for random window referencing), indicating that the user in this protocol often went back and forth among a small set (five or fewer) of windows. [...] Four indicators [...] suggest in unison that users exhibit locality of window reference and that, therefore, as in the case of virtual-memory operating systems, mechanisms that exploit this statistical property of user behavior might be developed."
The locality may be identified with the execution of a task: the user gathers a number of tools to do it: as the user moves back and forth among the tools, the tools windows form a locality.
The windows management systems provides several screen sized rooms: each room contains a set of windows placement. Since they expect users to use windows in a locality consisting in a task, each room aggregate the windows belonging to a task.
They provide an overview and doors among rooms to navigate among different rooms.
The paper then describe different problems and solutions relative to navigation, sharing of the windows, duplications, moving and so on.
Disclaimer: I wrote these summaries to help me remember the content and the main ideas of the paper. Since I am interested in certain aspects, I may leave out others.
If you have any comment, contact me at