Type systems don't eliminate all testing, but they do eliminate a large swath of possible errors that need to be tested for, like the one that has (had?) been afflicting Sinatra.
"Some programming language theorists would have us believe that the one true path to working systems lies in powerful and expressive type systems which allow us to encode rich constraints into programs at the time they are created. If these academic computer scientists would get out more, they would soon discover an increasing incidence of software developed in languages such a Python, Ruby and Clojure which use dynamic, albeit strong, type systems. They would probably be surprised to find that much of this software—in spite of their well-founded type-theoretic hubris—actually works, and is indeed reliable out of all proportion to their expectations."
YMMV, but I've never seen the Ruby project that didn't suffer from stupid bugs even the most basic type system would have caught.
One of the biggest advantage of type-systems for me is that they don't suffer from No True Scotsman. If you don't have a test that covered this bug, people are happy to tell you you're doing it wrong. The type-system doesn't have that problem. It doesn't "rot". It expresses only what you tell it to. It enforces maintenance and rigor in a way that tests, as non-integral code, fundamentally can't.
But that's just my opinion man.
Anyway, I'm no PL theorist, only a schlub that's a better programmer because of ML. Yeah yeah, sample size of one...
I've written lots of useful stuff in Python and Ruby, but I've also sunk lots of time into tests that I would have preferred not to need to write.