Manual | Discord | Paper | X | BlueSky | Email

Why use Morloc?

• Universal function composition: Import functions from multiple languages and compose them together under a unified, strongly-typed functional framework.

• Polyglot without boilerplate: Use the best language for each task with no manual bindings or interop code.

• Type-directed CLI generation: Write concrete function signatures once and automatically generate elegant command-line interfaces with argument parsing, validation, help text, and documentation.

• Composable CLI tools: Morloc CLI programs can be composed by simply importing them into a new Morloc module and re-exporting their functions.

• Seamless benchmarking and testing: Swap implementations and run the same benchmarks/tests across languages with consistent type signatures and data representation.

• Design universal libraries: Build abstract, type-driven libraries and populate them with foreign language implementations, enabling rigorous code organization and reuse.

• Smarter workflows: Replace brittle application/file-based pipelines with faster, more maintainable pipelines made from functions acting on structured data.

Below is a simple example, for installation details and more examples, see the Manual.

A Morloc module can import functions from foreign languages, assign them general types, and compose new functions:

-- Morloc code, in "main.loc"
module m (vsum)

import root-py
import root-cpp

source Py from "foo.py" ("pmap")
pmap a b :: (a -> b) -> [a] -> [b] 

source Cpp from "foo.hpp" ("sum")
sum :: [Real] -> Real

--' Input numeric lists that will be summed in parallel
--' metavar: LISTS 
type Lists = [[Real]]

--' Sum a list of numeric lists
--' return: Final sum of all elements in all lists 
vsum :: Lists -> Real
vsum = sum . pmap sum 

The imported code is natural code with no Morloc-specific dependencies.

Below is the C++ code that defines sum as a function of a standard C++ vector of doubles that returns a double:

// C++ code, in "foo.hpp"

#pragma once

#include <vector>
#include <numeric>

double sum(std::vector<double> xs) {
    return std::accumulate(
       xs.begin(), xs.end(), 0.0);
}

Below is Python code that defines a parallel map function:

# Python code, in "foo.py"

import multiprocessing as mp

# Parallel map function
def pmap(f, xs):
    with mp.Pool() as pool:
        results = pool.map(f, xs)
    return results

This program can be compiled and run as below:

$ morloc make main.loc

$ ./nexus vsum -h
Usage: ./nexus vsum LISTS

Sum a list of numeric lists 

Positional arguments:
  LISTS  Input numeric lists that will be summed in parallel
         type: [[Real]]

Return: Real
  Final sum of all elements in all lists

$ ./nexus vsum [[1.2],[0,0.1]]
1.3