You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
At the moment, method expressions must be of the form T.m or (*T).m . There's no good reason not to generalize the notation to T.m where T is any kind of type that has a method m. The
only options are the ones we already have, as well as unnamed interface types and struct
type embedding fields with methods.
These extra cases are largely academic but there's no real reason for the restriction;
it only adds complexity to the spec and requires extra checks in the compiler.
This would be a backward-compatible language change w/o impact to users. It would
simplify the spec. It simplifies the type-checker. It may simplify the rest of the
implementation.
See also issue #8605.