Fix haddocks of Data.IntMap.compose

[tomsmeding]

The words "former" and "latter" in the documentation of compose on IntMap were swapped. This PR puts them the right way round.

[Mikolaj]

LGTM

[alexfmpe]

Mind fixing this in Data.Map.compose as well?

[meooow25]

The docs say

-- | Relate the keys of one map to the values of
-- the other, by using the values of the former as keys for lookups
-- in the latter.
...
compose :: IntMap c -> IntMap Int -> IntMap c

which looks correct to me:

-- | Relate the keys of one map (IntMap Int) to the values of
-- the other (IntMap c), by using the values of the former (IntMap Int) as keys for lookups
-- in the latter (IntMap c).

[tomsmeding]

Ah, my bad. You are both correct:

1. The hadocks for Data.Map could to be improved too
2. The issue is not actually with former/latter, but with the fact that the "former" is the second argument to compose and the "latter" is the first argument to compose, which confused me, @Mikolaj and @alexfmpe -- @meooow25 was the first to notice.

I've force-pushed a new suggestion. Thanks for reading critically!

[Mikolaj]

Aha! "Relate" as in "map", as opposed to as in "identify". I think this proves the text needs to be rephrased totally.

[tomsmeding]

I think "relate" can be justified, because a Map (and an IntMap) is, after all, a relation.

[alexfmpe]

Maybe "link" ?

[tomsmeding]

Perhaps:

Given maps bc and ab, build a new map that relates keys of ab to values of bc, by using the values of ab as keys for lookups in bc.

This still uses "relates", but uses it in a way that makes it fairly unambiguous what it means, I think. @Mikolaj?

[Mikolaj]

Yes, this helps.

[meooow25]

Thanks, this does seem like an improvement.