22 Sep 2008
This paper, presented at ICFP'08, describes ML type inference in an entirely graphical setting, and shows that MLF type inference is only a very simple generalization of the ML case. Our approach is constraints-based; we give some equivalence rules on constraints, as well as an algorithm to solve them. We also show that type inference for MLF has linear complexity in practice, as in ML.
The full abstract for the paper can be found in my publications page. Compared to the published version, this version corrects some small mistakes in Definitions 9 and 11, and in Figure 17. The long version is available as Part 2 of my PhD dissertation.
Some talks I gave about this work: