Practical fallout from types and type inferencing

[Shriram Krishnamurthi <address@hidden>]

Dan Sugalski wrote:

> Are there any good papers on *implementation* benefits from type checking
> or type inferencing (especially the latter)? I can draw some reasonable
> assumptions on wins I can get from doing things with types, but practical
> papers (or discussion here, that'd be cool) of things to do, not do,
> places to cheat, high and low cost/payback techniques and whatnot'd be
> quite interesting.

Just off the top of my head:

Let's see: what are the stages of a typical implementation?

- scanner

None I can think of.

- parser

None I can think of.

- semantic analysis

Obvious.

- code generation, optimization

Typed-intermediate and -assembly languages, and the corresponding TIL
and TAL optimizing compilers.  (Also Shao's work along similar lines.)

- run-time systems

Tag-free garbage collection, Benjamin Goldberg.  (Much trickier in the
presence of parametric polymorphism.)

Shriram