RoadMap

Phase 1: Minimal Viable Product

• Write a Lexer
• Write A Grammer
• Determine which parser i would like to do
• Write an analysis pass
  • determine function variables and closures
  • determine function flags
• Write a module linking pass
  • determine what needs to be linked
  • determine how linker will inject code
  • determine how to deal with modules that require other modules
  • write main linker
  • ensure no cyclic dependency
• change the grammar for imports to support names imports
• perform macro running phase
• Write a type checking pass
  • Implement a basic type checker
  • add type checking of function calls.
• Write a validity check pass
• write an optimization pass
  • Remove Comments from parsetree
  • Remove Compiler Flags From parsetree
  • perform constant propagation
  • perform dead code removal
  • call known functions directly
• Replace commander in cli with a custom solution
• Unit Tests
  • write a basic unit testing library
  • write unit tests for lexer output
  • write unit tests for parser output
  • write unit tests for module linking output
  • write unit tests for type checker output
  • write unit tests for all intermediate phases
  • write unit tests for compiler output
• perform a phase for gc instruction insertion
• write a compiler phase
  • add support for 64 bit pointers
  • wasm imports
  • strings
  • wasm datatypes, i32, i64.
    • add basic support for wasm datatypes
    • add a way to specify that the datatypes are wasm in function call parameters
  • Numbers
    • negatives
    • positives
    • floats
    • 64 bit numbers
    • Rationals
    • Arbitrary precision integers
  • Closures
    • implement closures
    • allow closures to reference the function they are assigned too.
• Add in functions that correspond to wasm instructions
• write a parser

Phase 2: Making the Language Easy to use

---

• Write A runtime in Brisk
  • Write memory management
  • write garbage collection
• Write A standard library
  • memory access
    • memory.load(ptr: i32, offset: i32) -> i32;
    • memory.load(ptr: i32, offset: i32, data: i32) -> Void;
  • wagi calls
  • io
  • process
    • process.args() -> String[]
  • Print function
    • println
    • print
  • Control flow
    • if if(condition: boolean, ifTrue: Function, ifFalse: Function) -> Void
    • looploop(body: () -> boolean) -> Void
      • When you return true the loop will continue otherwise returning false will cause a break
  • Math
    • Addition
    • Subtraction
    • Division
    • Multiplication

Phase 3: We Have Been Succesful

---

• Write A Lexer in Brisk
• Write A Parser in Brisk
• Port the rest of Brisk from ts to js
  • write all other passes, basically just changing the syntax and optimizing
  • write a new compiler that does not depend on binaryen: probably the hardest part to do well.

========

• Version 1: Determine what needs to be done and get everything working
• Version 2: Rewrite Version 1, this is good because it means things are more organized because I understand what I need at which point instead of snaking stuff through
• Version 3: Remove Binaryen
• Version 4: Rewrite in itself