为什么在这个Haskell程序中没有使用尾调用优化？

Question

以下程序打击堆栈：

__find_first_occurrence :: (Eq b) => b -> [b] -> Int -> Int
__find_first_occurrence e [] i = -1
__find_first_occurrence e (x:xs) i
    | e == x = i
    | otherwise = __find_first_occurrence e xs (i + 1)

find_first_occurrence :: (Eq a) => a -> [a] -> Int
find_first_occurrence elem list =   
    __find_first_occurrence elem list 0

main = do
    let n = 1000000
    let idx = find_first_occurrence n [1..n]
    putStrLn (show idx)

失败

  

堆栈空间溢出：当前大小为8388608字节。使用`+ RTS -Ksize   -RTS'增加它。

但是，据我所知，对__find_first_occurrence的可能递归调用是__find_first_occurrence评估的最后一件事，因此应该可以进行尾调用优化。

Answer 1

使用了尾部调用优化，但(i+1)表达式被thunked，并在最后评估它们导致堆栈溢出。