mo

Mo is a programming language under development, aiming to provide full C++ interoperability, lisp-like meta-programming and a high level of built-in safety checks, without compromising performance or ease-of-use. At the moment, it’s undergoing rapid development of the specification. Read more…

abstract

Introduction to mo

syntax

The required basic syntax and features of the language which cannot effectively be wrapped inside a library are considered syntax. This layer could perhaps be likened to the grammar of a spoken language; it provides the structure from which meaning can be derived, but no actual vocabulary.

It includes the import and alias commands since these create new symbols outside of the usual symbol-creation syntax, and because import must be available for use before even the core library has been imported, and likewise alias aught always to be available.

It also includes an assignment operator, since assignment cannot really be implemented with the expected syntax in the same way as other operators are implemented.

lang (most content still needs to be separated into cagegories)

Features which must be provided as built-ins because they are either atomic or act as atoms for performance or other reasons are provided as symbols under lang, and could be called the language’s semantics.

Atoms consist of conditionals, storage types, arrays(?), and fundamental operations. These include no operators; for example lang.int32.add is a built-in integer-addition function.

lang is potentially not the best name for this component, but fits reasonably well in that when programming in mo, all symbols under lang are clearly built-ins.

core (no content)

Since many of the features of the lang namespace cannot so simply be used for standard programming tasks, a low-level library, dubbed core, is provided to wrap these with convenient syntax for every-day use. Thus this library provides standard data types with operators: int, string, etc., and flow-control constructs like if and for, as well as wrapping built-in functions like lang.float64.exp with templates to select the correct version based on expected return-type.

The reasons for providing this behaviour in a library and not as built-ins are two-fold:

the compiler can be simplified as much as possible, which allows more to be written in mo’s own syntax and hopefully reduces the differences between independent compiler implementations
core is just a standard library, and so people wishing to experiment with their own higher-level syntax may develop alternative implementations without braking binary compatibility with libraries based on core. However, I expect most such attempts will be built as extensions on top of core to ease code compatibility and avoid duplications.

In any case, higher-level functionality such as hash-maps and IO-streams are not included in this library.

Distributed under the Boost Software License, Version 1.0. (See accompanying file licences/BOOST_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)

mo

Contents