The Order Metalanguage for C Preprocessor Metaprogramming

From the preface I wrote years ago:

Order grew out of my interest in the design of programming languages and metaprogramming as well as desire to make something original, fascinating and beautiful. My work on the Boost Preprocessor-library had left me unsatisfied. While it was clear to me that it could theoretically be used to implement useful non-trivial metaprograms, it would have been quite tedious work, full of nasty little details and limitations to trip you up. I considered the prospects for further enhancements to the library, but it didn't seem possible to provide a programming model that I would have considered pleasant to think in. It occured to me that instead of attempting to provide semimodular enhancements on top of the C preprocessor in a bottom-up fashion, it might be possible to lift the limitations by designing a complete monolithic interpreter for a high-level source language in a top-down fashion. After some experiments at interpreting lambda-calculus and a couple of prototype interpreters for early versions of the Order language, I was left with a feeling that it would be too inefficient to be practical. The most serious limitation seemed to be the inability of the early prototypes to generate arbitrary output efficiently. I decided to forget about the interpreter and move on, but the work proved too interesting for me to forget completely. Over a long period of time I found several small incremental improvements to the interpreter until it suddenly started to look like more than a theoretical curiosity. Then I knew I just had to finish the work. Hopefully everyone will be able to find some beauty in the result. At least I know I've had a lot of fun designing the language.

Vesa Karvonen