Programming Language Development - via Virtual Machines

I don't know about you, but programming language design and development has always interested me. Why and how a language handles Objects, Functions, Pointers, References, and more. At first it seemed like only the Coding Wizards could do something so complex. But then I learned about VM's and Assembly, it has changed my entire outlook on developing my own programming language and "Interpreter" and/or Compiler.

Let's take a couple hours and implement our own VM, which runs our very own Assembly like machine code. We will run through developing a Stack Based Virtual Machine, and then a quick Assembler to convert our text based Assembly to our VM's Binary Machine Code format.

Why bother?

First, it is crazy cool and then the usual. Looks good to have in your github account, you will learn a lot, and there are actually very good use cases for a Domain Specific Language (DSL) think about SQL for a little. That is a very successful DSL. What if you have a need to enable scripting in your project? Is reaching for Lua or Python really the best choice? Why not just bake features into your VM and language, specific to your project. That is what Godot did after all, and it turned out to be very successful.

What if it is to advanced for me?

It most likely isn't. If you have a basic understanding of a C-like language or even Python, with some intuition about what Assembly is, you will do just fine. Maybe this will help calm your nerves: We are going to start with creating an Array of Bytes (which we will load from a file), then we will create another Array of Bytes (all zeroed). This is our Program Memory (PM) and Storage Memory (SM), respectively. SM will contain both Stack and Heap, while PM will only contain the instructions (ByteCode) of the program (resembling a Harvard Architecture). We will create a function for each Instruction our VM understands (i.e. ADDI, SUBI, MULI, DIVI, MODI, JEQI, JMP, CALL, etc.) which will look something like

void vmi_jmp(VMState* state) { /* Code to update vm state */ }

and a main loop function that starts from the first byte of PM and reads bytes until it reaches the end, calling the individual functions for each Instruction encountered.

If this sounds like fun, please consider joining me, and get your first VM based DSL up and running in about 2 hours.

Possible Bonus Round

If time permits, we can even fly through the implementation of my NymphVM and Nymph language compiler (Lexer/Parser/Generator). This is a VM I am developing for use in my FOSS Marketing Automation project Sociomauto, for inter-module communication and querying. Designed to be used with multiple Languages from Lisp variants to BASIC inspired languages.