[Haskell-cafe] Order of Map.fromListWith

[Patrick Redmond]

Correction: "the combiner has type 'k -> a -> a -> a'"

On Thursday, March 3, 2016, Patrick Redmond <plredmond at gmail.com> wrote:

> It looks like fromListWith is indeed implemented with a left fold over
> insertWithKey, with the Map as the accumulator. However, in insertWithKey
> the value-combiner function has type 'k -> a -> k -> a' which means it can
> combine in either order (it doesn't have to do with the fold or iteration
> order at all). The docs say:
>
> If the key does exist, the function will insert the pair @(key,f key new_value old_value)@.
>
>
> This doesn't really answer the "why" part of your question, but explains
> why you get [2, 1] instead of the reverse.
>
> On Thursday, March 3, 2016, Niklas Hambüchen <mail at nh2.me
> <javascript:_e(%7B%7D,'cvml','mail at nh2.me');>> wrote:
>
>> Does anybody know why for fromListWith, the arguments to the combining
>> function seem flipped?
>>
>>   > import Data.Map
>>   > fromListWith (++) [('a',[1]),('a',[2])]
>>
>>   fromList [('a',[2,1])]
>>
>> I often use it to group things by some key, e.g.
>>
>> postsByUserId :: Map Int [Int]
>> postsByUserId =
>> fromListWith (++) [ (userId, [postId]) | (userId, postId) <- posts ]
>>
>> and regularly get tricked by the postIds being reversed in the result.
>>
>> This is especially unintuitive to me since:
>>
>>   > foldl (++) [] [[1],[2]]
>>   [1,2]
>>   > foldr (++) [] [[1],[2]]
>>   [1,2]
>>
>> Any ideas?
>> _______________________________________________
>> Haskell-Cafe mailing list
>> Haskell-Cafe at haskell.org
>> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-cafe/attachments/20160303/1e5f09e2/attachment.html>