Skip to content
/ tarlox Public

Tarbetu's Lox - My toy interpreter with parallelism capabilities

License

Notifications You must be signed in to change notification settings

Tarbetu/tarlox

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

93 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Tarbetu's Lox

What is Lox

Lox is a programming language described and crafted in the book, Crafting Interpreters. It's dynamically typed, object-oriented and has a C-like syntax. An example Lox program:

fun some_function(x) {
  if (x / 2 == 0) {
    return x * 5;
  } else if (x / 5 == 0) {
    return x * 10;
  } else {
    return x * 2;
  }
}

for (var i = 0; i <= 100; i = i + 1) {
  print some_function(i);
}

Some differences from JavaScript are:

  • Lox has stronger typing than JS and has only == for equality checking
  • Lox has a built-in print statement
  • Lox is class-based while JavaScript is prototype-based.
  • Lox don't have some interesting features like modules, bitwise operators, arrays, for-each loop, modules etc.

You can get further introduction here: The Lox Language

What is Tarbetu's Lox

"Tarbetu's Lox" or "Tarlox" is my toy implementation of Lox. It has some features which Lox don't have in its original implementation.

No numeric overflows

Tarbetu's Lox has 256-bit precision floats for number type. You can go beyond as you can.

var x = 100000000000000000000000000000000000000000000000000000000000000000000 * 9999999999999999;

Note that the printing would format the number as the scientific notation if the number is too big.

print x; // 9.999999999999999000000000000000000000000000000000000000000000000000000000000000e83

Values of variables are calculated parallelly

In Tarbetu's Lox, variable declarations are processed in parallel threads; it won't block the main thread until it's accessed.

var x = a_function_that_took_seconds();
print "The variable x is calculating, as the main thread goes on.";
var y = an_another_function_that_took_seconds();
print "The variable y is calculating like x. Now, two threads are running besides the main thread";
var z = an_another_function_with_parameters_took_seconds(y);
print "The main thread still isn't blocked, but the variable z will be calculated after variable z is ready.
var r = 666 * 333;
print "Variable r is calculated parallelly"
print x;
// Main thread is blocked until the variable x is ready. At this moment, y and z are still calculating in another thread.

To avoid this behaviour, you can declare your variables with await_var.

await_var x = a_function_that_took_seconds();  
// Thread is blocked until x is ready.
print x;
// No blocking, because x has been calculated.

You might check the variable if it's ready; this can be done with is_ready operator:

var x = a_function_that_took_seconds();
if (is_ready x) {
	print "x is ready. Here is your x: " + x;
}

Function Return Value Memoization

If a function is called with the same arguments, the function will quickly return the value which is cached before unless executing the function body.

fun power_of_two(x) {
  print "Calculating power of " + x;
  return x * x;
}

print power_of_two(2); // Prints the text from the function body before returning a value;
print power_of_two(2);
// Returns the value, but does not execute it so the print statement from the function body doesn't work.
print power_of_two(4); // Prints the text from the function body before returning a value;
print power_of_two(4);
// Quickly returns the value so the print statement from the function body doesn't work.

Always the cache doesn't work and the function call will execute the body in this situation:

  • If callable returns nil
  • If callable don't take any arguments
  • If callable is a method or native function (Implemented in the host language, Rust)

You can return functions in Lox, but the cached values will be cleared.

Tail Call Optimization

Tail Call Optimization is an optimization technique which eliminates additional calls in recursion if the return statement only consists of function calls.

fun factorial(x) {
	fun factorial(acc, n) {
		if (n < 2) {
			return acc;
		}
		return factorial(acc * n, n - 1);
	}

	return factorial(1, x);
}

print factorial(666);

(This example also demonstrates that Lox allows inner functions)

Lambda

You can declare lambdas like this:

var power_of_two = lambda(x) { return x * x };
print power_or_two(2);

Like other functions, memoization will work in the same scope.

Issues and Caveats

About Paralelism

Sometimes the program goes into a deadlock while playing around with variables. The behaviour is not consistent, so I can't reproduce. It's not common, and the interpreter may execute the same code without no issue.

Direct call on Lambda

Currently, you can't do that:

lambda(x) { print x * 2; }(4);

It's odd, I know.

Tail Call Optimization is only triggered when the return expression is a call.

This won't be counted as a tail call:

fun factorial(acc, n) {
	if (n < 2) {
		return acc;
	}
	var x = factorial(acc * n, n - 1);
	return x;
}

Memoization causes memory leaks in the global scope

How will the cached values cleared? If you leave them in a scope, they will be cleared.

However, what happens when you implement them in a global scope like dozens of functions? Dozens of memory leaks.

Check this:

fun factorial(x) {
	fun factorial(acc, n) {
		if (n < 2) {
			return acc;
		}
		return factorial(acc * n, n - 1);
	}

	return factorial(1, x);
}

print factorial(10000000);

All results of function calls in recursion were stored. It is pure waste!

You can avoid wasting your memory like this:

class Factorial {
	init(x) {
		fun factorial(acc, n) {
			if (n < 2) {
				return acc;
			}
			return factorial(acc * n, n - 1);
		}

		this.result = factorial(1, x);
	}
}

print Factorial(42).result;

Since methods don't do memoization while nested functions do, there will be no memory leak after this point.

Wishlist

To be honest, I don't have lots of free time to implement these but I would like to do them in my free time. If you implement this for me, you will get a place in my heart.

  • Direct call on Lambda and environment capturing
  • Changing await_var x = 0; as await var x = 0;.
  • Importing mechanism
  • A good standard library
  • Arrays and Hashmap
  • An iterator protocol and for-each syntax
  • Implementing module and bitwise operators
  • Better approach for storing variables to avoid deadlocks
  • A decent GC for preventing memory leaks of memoization

How to Run

First of all, you will need Rust tools to execute the interpreter. The best way to install them is to install Rustup on your machine.

To get the REPL:

cargo run

To run a file:

cargo run -- ~/Code/Lox/my_script.lox

To run an example:

cargo run -- example/tail_call.lox

You can install "Tarbetu's Lox" to your system like this:

cargo install --path .

The executable will be available in your $PATH, you can call the interpreter like this:

tarlox my_script.lox

To remove:

cargo uninstall tarlox

About

Tarbetu's Lox - My toy interpreter with parallelism capabilities

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages