Is the documentation for optional parameters incorrect?

[John Watson]

Excellent package and excellent documentation! I wonder, though, whether the description of an optional String parameter for Feed in the documentation is slightly incorrect? You have in the cleanest representation (as well as the less clean ones) :

"Feed": "feed" ? { search: optional }

but for me, this fails to compile, whereas this works:

"Feed": "feed" ? { search: optional <<< string }

[Nathan Faubion]

What error are you getting? I think the former is how it ideally works, but it might not work because of how the compiler currently handles polymorphic things in row types along with constraints. The optional <<< string likely works because it makes it monomorphic (Maybe String instead of Maybe a).

[John Watson]

I'm still using PS 0.12.5 and get:

No type class instance was found for
                                                                                                                       
    Routing.Duplex.RouteDuplexBuildParams (Cons "predicate" (RouteDuplex t3 t4 -> RouteDuplex (... ...) (... ...)) Nil)
                                          ( predicate :: RouteDuplex t3 t4 -> RouteDuplex (... ...) (... ...)          
                                          )                                                                            
                                          ( predicate :: Maybe String                                                  
                                          )                                                                            
                                          ()                                                                           
                                          ( predicate :: Maybe String                                                  
                                          )        

[Nathan Faubion]

Yeah, you can also try a type signature.

optionalString :: RouteDuplex' String -> RouteDuplex' (Maybe String)
optionalString = optional

[John Watson]

OK - to make the function monomorphic. Thanks, Nate.