如何结合两个Maybe类型？自定义Semigroup实例遇编译错误求助

首先，你编写的instance Semigroup (Maybe Int)编译报错的核心原因是：Haskell标准库已经为Maybe a提供了通用的Semigroup实例（要求a本身是Semigroup类型），自定义的具体类型实例会和这个通用实例冲突，导致编译器无法解析。

要实现仅当两个Maybe都是Just时相加，否则返回Nothing的行为，且直接用<> 运算符，最佳方案是通过newtype包装Maybe类型，避免和标准实例冲突：

-- 用newtype包装Maybe，明确语义是数值相加的Maybe
newtype SumMaybe a = SumMaybe (Maybe a) deriving (Show)

-- 为SumMaybe定义Semigroup实例，约束内部类型是Num
instance Num a => Semigroup (SumMaybe a) where
  SumMaybe (Just x) <> SumMaybe (Just y) = SumMaybe (Just (x + y))
  _ <> _ = SumMaybe Nothing

-- 测试用例
m1 :: SumMaybe Int
m1 = SumMaybe (Just 1)

m2 :: SumMaybe Int
m2 = SumMaybe (Just 11)

m3 :: SumMaybe Int
m3 = SumMaybe Nothing

main :: IO ()
main = do
  print $ m1 <> m2  -- 输出 SumMaybe (Just 12)
  print $ m1 <> m3  -- 输出 SumMaybe Nothing
  print $ m2 <> m3  -- 输出 SumMaybe Nothing

为什么不能直接定义Num a => Semigroup (Maybe a)？ #

标准库中已经存在Semigroup a => Semigroup (Maybe a)的实例，对于同时满足Num a和Semigroup a的类型（比如Int），两个实例会产生重叠，编译器无法确定使用哪一个，因此这种写法会触发重叠实例错误，不推荐使用。

对比原Monad实现 #

这个SumMaybe的<> 行为和你用do语法实现的combine函数完全一致，都能达到：两个Just则相加，任意一个是Nothing则返回Nothing的效果。

内容的提问来源于stack exchange，提问作者toni057