The talk was not one of my best. I do this thing where I get the talk to a usable level quite early, planning to improve it. But then the thought of changing it makes me nervous. It could have done with animations and better use of the presentation hardware, and more planning to leave room for the interesting stuff at the end.
I've decided to give up on space as the separator in explicit Lists, though cute. So comma is always used for operations that create a compound value, overriding the comma operator that creates Tuples.
Had an interesting discussion of subtyping and equality. I can see that it is useful to also have a system where you take a type and add some extra data and methods, and have a forgetful procedure/coercion to go back to the original type, and some automatic inheritance of properties. However I'm wary of introducing the confusion that eventuates in OO where parent and child have the same property with different meanings, e.g. equality.
Another interesting discussion with the guy advocating weak algebraic effects. That only works with left to right evaluation. Haskell was forced to monads because of lazy evaluation. I hope to avoid that with the TimeOrder type. To reduce the need for !s I will introduce EffectfulProcedures (`!{ }) which force left to right evaluation (i.e. calls to them have an implicit ! unless there is an explicit one).
I was disappointed not to get any useful discussion on proofs.
I'm preparing two docs: Wombat Tour and Wombat Operators to cover my ambitious implementation plan.
