Implementations of type inference algorithms with constraints?

[Please respond directly to Howard.  Howard, can you summarize
responses here in a few days?  --BCP]

We are looking for any implementations of constraint-based type inference
algorithms, such as for a language with subtyping. So far, we've seen the
Illyria language from Alex Aiken. Are there other implementations, and are
they publicly available? In particular, we're interested in how large
constraint sets are in practice, and how the constraints might be

Thanks very much,
--Howard Huang