OS theory - LeFreq/Singularity GitHub Wiki

Operating systems theory asks, are there idealized operating-systems architectures for different categories of computation?

For example, an OS for developing a simulation engine (for CAD or 3-d graphics) is architected differently and requires different underlying hardware to be efficient versus a data-organizing computer and is for building knowledge (databases, spreadsheets, perhaps word processing, object-oriented environments).

The Church-Turing thesis says that all computation is effectively the same, but this says nothing about the efficiency and elegance of the work done.

It might be that we're in a nascent area of digital computing where one-size-fits-all approach is used because of an absence of real computer architecture to understand how to build efficient machines for mundane tasks or perhaps the market is spoiled. Whatever the case, it is poor design and inefficient.

An ASIC, for example, could be all one will ever need to perform secretarial duties and the rest of the GUI and such is really a distraction and source of technical difficulties.

A system like Unix, for example, is good for building apps. But most computer use involves using just a handful of them.

This project suggests that besides traditional OS theory structured around the execution process and securing data (through storage, CRC checks, cryptographic security, etc.), that there is another element: data architecture. That is important. Many OSs must patch together their file systems and the relationships between data in an ad hoc and poorly structured manner. Do config files, for example, go under the application directory or user directory? These problems never get solved or improved because these OS's lack a unified model for data and lack a data philosophy.

So this project identifies four main foci for new operating system theory:

• process execution
• data storage
• system security
• data model
• how the OS relates to other computers