News for hackers. Stuff that matters.
Re:Homotopy Type Theory (2012) [pdf] by curryhoward on Tuesday October 27, 2020 @08:56PM (#24913808)
> it has canonicity, which I'd argue is necessary for a general-purpose programming language

MLTT, even without HoTT/CTT, already features canonicity, no? (At least for positive types.)

I thought the main advantage of HoTT is that you can define non-trivial equalities, and the advantage of CTT in particular is that univalence can be derived in it (rather than introducing it as a non-computational axiom).

Re:Homotopy Type Theory (2012) [pdf] by remexre on Friday October 30, 2020 @03:33PM (#24945999)
Right, I think I'd put univalence in my list of "I want this for my dream general-purpose dependently-typed programming language" features; being able to prove properties on simple representations of data and transport those properties to more optimized forms is pretty important, imo.