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
A new issue has arisen with this as of #45. Where the following was (maybe?) working before, the type system is now unable to determine if A constitutes the value of a nullary constructor (comparison against null here):
This extends to subtyping constraints. Changing A on f to for example A extends string makes no difference as, as far as the type system is concerned, A could be string & null.
Could this be solved by representing nullary values with a symbol and switching out to null when (de)serialising? Depends how the typechecker treats unique symbols.
A tracking issue for if this ever becomes viable. In the meantime we should document the challenges faced.
The text was updated successfully, but these errors were encountered: