The user's view

The end-user of the system is of course not supposed to think in terms of byte-code, assemblers or node-servers. We want to hide these uninteresting (to the end-user at least) technicalities. The way I would like to see this system evolve, we will be able to present users with a friendly front-end and a well designed language for expressing numerical problems. How the parallelization and scheduling is actually happening, should not be a concern of the user. The system as I envision it, will in time evolve into something that - to the end user - will look much like Figure 2.2.

Figure 2.2: The user's view of the system.

Actually most of what the user sees, will be integrated into one client tool, that will provide source-code editing, visualization and job submission capabilities. The user will not need to bother whether the code is being run on 1 or 100 nodes.

A well functioning back-end is - as I see it - the most vital part of this system, and it is what this report is all about. The second most important thing will likely be a good language for the end-user. The current VM-oriented assembly language is a reflection of the inner workings of the virtual machine, not a reflection of the way users will want to easily express computing problems.

The development of the TAL (TONS Algorithmic Language) language, the compiler, and the front-end application, is a huge task in itself, and way out of the scope of this report. There is some work in progress with defining the language, but it will take years before we see a really polished well integrated end-user interface to the TONS system. The current TAL development is headed in the direction of a language that combines ideas from Pascal, MatLab and Cobol. I am not taking part in this development currently, it is done by people I know, who share the interest in developing a better computing environment. So far, I have been the sole developer of the code discussed in this report.