This is an excellent point, and I didn't mean to imply that it's impossible to implement TLS securely in Haskell (though my original comment did seem to; I plan to fix that shortly). Only that there are more natural choices that give the same primary advantage (a strong algebraic type system).
Here's an example of a type definition that specifies strict evaluation https://github.com/vincenthz/hs-tls/blob/master/core/Network...
See? They don't have to switch languages after all.