Interpreter not Compiler Issues

I'm thinking on working on an interpreter project for a language that
supports multitasking, but basically a 3rd generation language.
The language includes constructs that involve runtime access to the
symbol table, and source code, as well as runtime construction and
evaluation of code like LISP (eval ...)
The language also includes a GOTO, and semaphores for interprocess
communication.


What are the issues that distinguish implementing this kind of language
versus implementing a compiler?


I would like to implement as high a performance interpreter as I could,
are there known techniques that slow down interpreters, but are acceptable
for compilers?


I'm thinking of using PCCTS to implement the parser. Are there other
systems which are known to make interpreter writing more efficient?


Finally, how hard is type inferencing? The source language has
only a string datatype, and multiple dimensional strings-subscripted arrays
as its sole data structuring operator.


Appreciate input:


David (whitten@netcom.com) (214) 437-5255.
--

Return to the comp.compilers page.
Search the comp.compilers archives again.