This is a project for a course in functional programming and type systems (MPRI 2-4). Its goal is to add type classes to a ML-like language using a source-to-source translation, and to support type inference for typeclasses.
The language for this project is supposed to be called Mini-Haskell; but it doesn't even have Monads, Haskell's most famous feature! In order to support them, we have added (limited) support for constructor classes, which includes polymorphism with type variables of kind * -> * and the ability for methods to be polymorphic in more type variables than just the parameter of their class.
To make this fit into the pre-existing framework, we had to resort to piling up ugly hacks on top of the code base. For example, if you try to put something complicated into an instance declaration for a constructor class, it will break; you are advised to define your methods separately and then just put an equality in the instance definition.
To enable this feature, use the --fts flag, which also lifts a
restriction on namespaces in the spec. A sample file is given in
test/functor-applicative.mlt. Test using
./src/front.native test/functor-applicative.mlt --fts --compile-with-ocaml
You will be able to witness the usage of ML modules and functors to translate constructor classes.
- OCaml >= 4.00
- The Menhir parser generator
cd src/
make
This results in an executable named joujou in src/.
To clean the build files, just do make clean.
The build system is now completely based on ocamlbuild, which makes life a lot easier!
To use the tests, first enter the directory test, which contains
the script run-tests.hs.
runhaskell run-tests.hs elaboration bad
This runs the program on every .mle file in the directory and
displays a formatted log of the test results.
When the compiler has returned with exit code 0, [[Success]] is
displayed, otherwise [[Failure]] is displayed. In the summary [OK]
means test successful i.e. the test file was rejected (since these are
negative tests). So [[Failure]] -> [OK] and [[Success]] ->
[KO].
To test inference, run
runhaskell run-tests.hs inference bad
Note that when testing inference, --inference-only is activated: the
program won't bother with elaboration.
runhaskell run-tests.hs elaboration good
runhaskell run-tests.hs inference good
Pretty much the same thing, except [[Success]] and [OK] match.
runhaskell run-tests.hs (inference|elaboration) (good|bad) clean
Deletes the generated .mle/.mlr files.
ocamldoc -html -d doc -I elaboration/ -I inference/ -I common/ **/*.mli