Search

Many language implementations, like C++ and Java, check the types of values and operations to be sure each operation is supported for the types of its operands. An important feature of Standard ML is the type inference system which is somewhat like the type checkers of C++ and Java, but a bit more powerful. A type checker checks the types written by the programmer to be sure each type declaration is consistent with the operations being performed, values being passed to functions, and the values being returned. Compilers for Java and C++ even infer the types of some expressions when polymorphic operators are used. For instance, the addition operator has multiple result types depending on the types of its operands.

The ML in the name Standard ML stands for meta-language. SML was designed as a language for describing languages when it was used as part of the Logic for Computable Functions (LCF) system [9]. Two tools were designed to work with Standard ML for language implementation, ML-lex and ML-yacc. The pattern matching, ease of defining recursive datatypes, the functional nature of the language along with these two tools make Standard ML an excellent language choice for implementing interpreters and compilers.

This paper provides a description of a framework for programming language implementation that is accessible to students in a one semester course focused on programming languages. Rather than concentrating solely on language features, this paper describes a framework where imperative, functional, and logic programming languages are all used to develop a compiler for a non-trivial subset of a functional programming language. Provided by the framework are a virtual machine, a disassember of Python programs, a partial implementation of a Standard ML compiler with suggested tests, and a partial implementation of a type inference system implemented using Prolog, again with suggested tests. Classroom experience gained while using this framework is also shared. A suggested fourteen week sequence is provided. Real reactions from students and reflections on coverage of the ACM 2013 Curricula guildelines conclude the paper.

When using formal methods to generate compilers it is desirable for all levels of the compiler to be formally specified. Typically, register allocation has been thought to be equivalent to graph coloring. Since graph coloring is NP-Complete most algorithms for register allocation have been ad-hoc. This paper presents a framework for register allocation that has been formally verified using an inductive theorem prover.

This documentation was generated from the Python documentation available by typing from turtle import * help (TurtleScreen) in the Python shell.

This documentation was generated from the Python documentation available by typing help(list) in the Python shell.

This documentation was generated from the Python documentation available by typing help(dict) in the Python shell.

This documentation was generated from the Python documentation available by typing help(int) in the Python shell.

This documentation was generated from the Python documentation available by typing help(float) in the Python shell.

This documentation was generated from the Python documentation available by typing help(str) in the Python shell.