Interpreter Generators: A Brief Look at Existing Work · Stefan-Marr.de
Motivated by Tiger, a tool for generating interpreters, being mentioned on Twitter, I had a brief look at vmgen, Tiger, eJSTK, Truffle DSL, and DynSem. What follows are my rather rough notes and pointers. So, this is by no means a careful literature study, and I welcome further pointers.
vmgen [1, 2] might be one of the seminal projects on interpreter research. It has been used to evaluate optimizations such as superinstructions and stack caching.
It provides a domain-specific language (DSL) to specify the semantics of bytecodes as operations on a stack. These specifications are then used to generate the interpreter, possibly as threaded code interpreter, and allow vmgen to apply optimizations to the bytecodes or construct superinstructions.
The first line encodes the stack operations, and says that two integers are popped of the stack, and one is return as result. The variable names can then be used in the second line to define the actual operation using C code. The language also supports defining register machines and superinstructions.
Overall the DSL is a relatively thin layer over the C target code, which gives a lot of control while enabling a high degree of automation.
