Ambiguity in proposed pattern syntax

[marijnh]

Colons will allow specifying the types of pattern elements (let x: int = 6), but are also used as separators between field names and sub-patterns in record patterns ({x: ?binding}). Since we allow the sub-pattern to be omitted ({x} is interpreted as {x: ?x}) this introduces an ambiguity—the parser can't tell whether a colon in a record pattern indicates a sub-pattern or a type.

Possible solutions include using another character for type assertions or simply specifying that the patterns has to be written out in full if you want to type it (resulting in the rather ugly {x:?x:int}).

[marijnh]

Graydon clarified on IRC that he intended type annotations for patterns to only occur at the top level of the pattern, not nested inside. That's probably just as powerful, and solves this problem.