Question

I copied the Euler.11 solution given on ( http://www.haskell.org/haskellwiki/Euler_problems/11_to_20#Problem_11), but it fails with an indexing error: " (Array.!): undefined array element". Of course firstly I d like a better error message(!), perhaps even giving the failed index, but failing that I tried to debug it.

这些数据是正确的,其印刷版显示了正确的界限和数据。

So I added a few trace messages, to both the result expression, and the body of the comprehension. I get lots of trace results from the final expression, but none from the body computations. Why?

prods :: Array (Int, Int) Int -> [Int]
-- trace ("xs: " ++ show xs) (product xs)
prods a = [trace ("xs: " ++ show xs) (product xs) | i <- range $ bounds a,
                        s <- senses,
                        let trace1 = check "i: " i,
                        let is = take 4 $ iterate s i,
                        let trace2 = check "is: " is,
                        all (inArray a) is,
                        let xs = map (a!) is]

-- Doit
-- euler = print . maximum . prods . input =<< getContents
euler eData = maximum . prods $ input eData

-- Debugging tracecheck :: String -> a -> a
check msg v | trace (msg ++ (show v)) True = v

Answer 1

首先,使错误信息包括失败指数将需要<代码>。显示限制在阵列指数中,这些限制可能未经许可。

第二,正如罗马人所说的那样,这些电文不是由于zy评价而印制的。 Bang-patterns (let !trace = 核对“i: “i)可能是绕过它的最方便的方法,但我不知道在名单上的行文中如何做这些工作。

其次,<代码>未界定阵列要素的电文告诉你,阵列的构造是不正确的(有些内容没有界定),因此,你需要缩减构造功能,而不是使用它的功能。

Answer 2

Haskell is a lazy language(*). You do not use trace1 and trace2 in your computation, hence they are not evaluated, and nothing gets printed.

For example, if you replace

let is = take 4 $ iterate s i

iii

let is = take 4 $ iterate s trace1

然后将使用<代码>trace1,并导致跟踪信息。

(*) 更确切地说,GHC是今天最常用的Haskell执行工作,它利用zy评价实现Haskell的非限制性杂质。如果这样做,就永远不会感觉到。

友情链接