James

I agree with @CaviarChen here in that we should remove some definitely unneeded params from continuation procs. They arent that difficult to approximate, as they're inputs to whatever the continuation's...

This one isn't an issue of type checking, but of expansion/inlining. The generically typed code comes from the inlined list constructor, which is generically typed. Fixing this will probably require...

Linked structures are likely a good candidate also, as having the structures allocated close by in memory could be good for cache performance.

I suppose this example could be solved with the use of a generalised `break`, taking an optional integer constant, `n` argument that breaks `n` loops e.g. ``` def product(matrix:list(list(int)), ?p:int)...

One potential issue I see with `failure_reason` as an automatic resource is the ultimate idea of it being some type class. If two different procedures set the `failure_reason` resource, there's...

One issue with a structure such as `[1,2]` is that the structure is not flat, and would have internal pointers, ala `?t0 = [2 | []]; ?t = [1 |...

Another very simple optimisation that we could implement is to pull multiple heap allocations together, allocating one large structure, and using segments of the allocated memory for each of the...

Interesting. That means that something like `disuse` is semantically equivalent to simpler constructs. i.e., ``` def foo use a, b { disuse b { !bar } } ``` is equivalent...

Why couldn't we handle HO resources as we currently do, in that we can't stop a HO term from using a resource. In essence, a HO term can always use...

I believe one alternative that works currently is ``` head(lst, ?h, _) ``` Which reifies the test, ignoring the succeed/fail `bool`. I think the program should segfault in many cases...