If you actually evaulate [1,1..1] == [1..] in a GHCI prompt it will tell you False immediately. :)

[1..] evaluates to  [1, 2, 3, 4, 5, 6, ... etc ...

[1,1..1] evaluates to [1, 1, 1, 1, 1, 1 ... etc ...

The syntax where you give 2 elements before the .. is intended to resemble this style of shorthand for lists: [1, 2, 3, 4, ...10]. The idea is that Haskell should take the two elements as "example elements" showing the pattern you want. So [1,1..] is saying you want a list that starts at 1, then has 1, and continues on like that; you're specifying a step size of *zero*, not of 1.

You can step 1 by 0 an infinite number of times without exceeding 1, so [1,1..1] should not be finite.

-- Ben

----- Original Message -----
From:
"Krisztian Pinter" <pin.terminator@gmail.com>

To:
<haskell-cafe@haskell.org>
Cc:

Sent:
Thu, 17 Mar 2016 03:58:06 +0100
Subject:
[Haskell-cafe] Odd list comprehension behaviour


Hello,

I noticed some odd behaviour with list comprehensions.

[1..1] == [1]
BUT
[1,1..1] == [1..]

I noticed this while writing a Clean program, but it seems Haskell inherited this as well.
In the case of integer lists with step size >= 0 the up_list function[1] is used:

up_list :: Integer -> Integer -> Integer -> [Integer]
up_list x0 delta lim = go (x0 :: Integer)
                    where
                        go x | x > lim   = []
                             | otherwise = x : go (x+delta)

In the case of [1,1..1] x0 == lim, so go will recurse infinitely, producing an infinite list.

I think the reasonable behaviour would be [1,1..1] == [1]. Is there a reason it doesn't work like this?

[1] http://hackage.haskell.org/package/base-4.8.2.0/docs/src/GHC.Enum.html#up_list

Thanks,
Krisztián