fromMaybe is working, but stuff feels a little messy/fragile

newt/Zoo1.newt

@@ -48,8 +48,26 @@ length (Define env _) = S (length env)
 
 lookup : Env -> Nat -> Maybe Val
 lookup (Define env v) Z = Just v
-lookup (Define env _) (S k) = Just (lookup env k)
--- bug in unify? are the meta args backwards? It seems to quote back right..
--- we're getting `Maybe Val` as meta3, and comparing:
--- Maybe ?m3  =?= Maybe Val
+-- If I write "Just (lookup env k)" on RHS, it's wrong, but the error message is unusable (mainly due to FC)
+-- The FC is fine if I write lookup {Val} env k
+lookup (Define env _) (S k) = lookup env k
 lookup (ENil) x = Nothing
+
+eval : Env -> Tm -> Val
+
+cApp : Closure -> Val -> Val
+-- If I put Closure instead of MkClosure, it reports missing case, fix that (should be bad constructor or something)
+cApp (MkClosure env t) u = eval (Define env u) t
+
+eval env (Var x) =
+  case lookup env x of
+    -- case doesn't use the new code.  We've got a wildcard here that
+    -- is forced to {Val}, but we don't have forcing/dotting
+    -- I guess we see what Jesper says about dotting
+    Just x => x
+
+eval env (App t u) =
+  let tv = eval env t
+      tu = eval env u
+  in ?
+

newt/testcase3.newt

@@ -0,0 +1,25 @@
+module Scratch2
+
+data Nat : U where
+  Z : Nat
+  S : Nat -> Nat
+
+data Maybe : U -> U where
+  Just : {a : U} -> a -> Maybe a
+  Nothing : {a : U} -> Maybe a
+
+
+-- failed to unify _:1 with Val
+-- Legit on RHS, IMO. On LHS, we should be dotting?
+-- I either need to unify and collect constraints or figure out how
+-- other systems manage this.
+
+-- Note that an unused
+-- variable may stand for either a wildcard or a forced pattern. In the latter case our algorithm
+-- treats it as a let-bound variable in the right-hand side of the clause.
+
+-- we need let-bound in environment but we do have define.
+
+fromMaybe : Maybe Nat -> Nat
+fromMaybe (Just x) = x
+fromMaybe Nothing = Z