Looking at ReasonML I can’t help but remember Alan Perlis’ “Syntactic sugar causes cancer of the semicolon” epigram. ;)
Seriously though, I like some of the ideas, but I have a big problem with the function definition syntax.
let f (x, y) and let f ((x, y)) are different type. That’s bringing the most counter-intuitive parts of S-expression based languages into one that isn’t. In the original syntax, parens are just expression delimiters, no more, no less, they have no syntactic significance other than that. Also, the syntax for curried anonymous functions ends up more verbose even than desugared syntax in the original.