Nathaniel Wesley Filardo

Should lists not use `nil/0` and `cons/2` but rather `$nil/0` and `$cons/2`, too?

This is really only a problem for `/0` atoms, right? In particular, the fact that we're co-opting initial-lower-alpha-alphanum-strings (ETA: without parens) for `for/2` `in/2` `is/2` `new/1` `whenever/2` `with_key/2` (and maybe...

Whoop, I reverted this without thinking about it as part of the QPT branch, which did away with booleans-as-primitive-values. I'll put those back, I guess.

I'm not really sure what's going on in the second failure -- that seems really bad and I am pretty sure it's not my fault. :) Can you run "ghci...

The planner is picking `alpha(U), beta(V), edge(L,U,V)` as each such query binds two free variables (including the result of `is`). The alternative, `alpha(U), edge(L,U,V), beta(V)`, requires running `out is edge(out,in,out)`,...

The current heuristic (exponentially) prefers steps that bind fewer variables to ones that bind more. Thus the binding of three variables in `out is edge(out,in,out)` is much worse than the...

Working on making the compiler understand. Amusingly enough, the parser framework does not seem to have this right; investigating.

OK, so I figured out what's going wrong. I added a trivial example to the forward-backward program and set num_iterations to 1. (Jason, as an aside, it would be nice...

> Why this kind of backward error propagation? count_transition/3 didn't do anything wrong. As discussed at length before, it's all we have: we're standing at the source of a hyperedge...

Hm. Sorry, you've caught the project in a pretty sad state right now; I've been trying to revamp some of the theory here for... entirely too long. In any case,...