Speaker: Peter Lee
Building Systems in Standard ML
Date: March 9, 1995
Abstract: A tremendous amount has been learned about the theoretical principles underlying the design and implementation of programming languages. In some areas, such as the type theory of the ML language, progress has been nothing short of spectacular. Despite this, such languages are rarely used to implement ``real-world'' systems. Why is this so? And what are the prospects for modern ideas in programming languages in the real world?
For the past two years, I have participated in an experiment to build systems software (in particular, a suite of network communication protocols) in an extension of the Standard ML programming language. One result of this experiment is tangible evidence that SML can be used to build practical systems, and that the type theory underlying the language provides many benefits, not just in programming, but also in system design and structuring. This is not to say, however, that the current design and implementation of the language is satisfactory. In this talk, I will describe our motivations for the project, the experiment, our experiences in carrying it out, and the results. Then, I will describe several interesting problems in SML that were exposed in the process. Some preliminary thoughts on how to approach some of these problems will be presented.
This is work with Robert Harper and Edo Biagioni.