A demo implementation of a simple dependently-typed language for OPLSS (Used in 2014 and 2013) Stephanie Weirich
The goal of this project is to bring up the design issues that occur in the implementation languages like Agda, Coq, Epigram, Idris, etc. Of course, it can't cover everything, but this code is a good starting point for discussion.
As its main purpose is didactic, the code itself has been written for clarity, not for speed. The point of this implementation is an introduction to practical issues of language design and how specific features interact with each other.
Furthermore, this code based includes a number of features (unit, booleans, sigma types) which are all subsumed by the general mechanism for datatypes. These are included to give examples before diving into the more general, and much more complicated, code.
- PTS representation (uniform syntax for all levels)
- bidirectional type checking
- erased arguments (forall)
- propositional equality
- indexed datatypes
- termination & inductive datatypes
- effects
- co-induction
- type inference & unification
- general constraint solving
This code is open source. Feel free to extend or adapt it for your own project. The definitive version will be uploaded to github after the summer school is complete. https://github.com/sweirich/pi-forall
There will be bugs. There will be design bugs. The language is probably not sound. Or may not be so in the presence of 'innocent' axioms like classical logic. Even in this little language there are many semantic pitfalls. C'est la vie.
pi-forall requires GHC and cabal, most easily available from the Haskell Platform (http://www.haskell.org/platform/)
in the top level directory type 'cabal install'
pi-forall/ README (this file) LICENSE pi-forall.cabal test/.pi example files src/.hs source code
Some of this code was adapted from the 'zombie-trellys' implementation by the Trellys team. The Trellys team includes Aaron Stump, Tim Sheard, Stephanie Weirich, Garrin Kimmell, Harley D. Eades III, Peng Fu, Chris Casinghino, Vilhelm Sjöberg, Nathan Collins, and Ki Yung Ahn.
This material is based upon work supported by the National Science Foundation under Grant Number 0910786. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.