Adds support for Enum.sum/2

[tiagonbotelho]

This PR adds a mapper function argument to the sum function argument.

There are at least two main benefits with this addition:

1. sum becomes available to a wider range of enumerables (e.g. maps)
2. Added expliciteness without requiring an Enum.reduce/3 or Enum.map/2 |> Enum.sum/1 to be done

[tiagonbotelho]

I probably should've opened a discussion in the mailing list but the change felt so small that it almost didn't seem worth it. If that'd be preferred I'm more than happy to create it 😊

[tiagonbotelho]

This is probably the bit I'm more concerned about, from my testing it seems like the approach of using the fun on the "edges" of the range and then coercing it back into a range, seems like the way to go. But I'm open to feedback on this 🙌

[josevalim]

This will only work if the function is linear, which we cannot guarantee. For example, &:math.sin/1 wouldn’t work as callback.

[josevalim]

Hi @tiagonbotelho! 👋

As you noted, a custom sum can be achieved using reduce, as you noted, and the other aggregation function with arity 2 that we have, count/2, has the second function working as a filtering. So this would add some inconsistency and it can be easily achievable in other ways, for this reason I am 👎 on this change. :)

[tiagonbotelho]

Ah @josevalim I wasn't aware of count/2 that is a good point 👍 I'd still prefer to have a straightforward Enum.sum/2 with an explicit mapper function (similar to what we have in Enum.sort_by/3) rather than an Enum.reduce/3 which requires a bit of parsing in order to understand what is really going on.

It is just a matter of preference though, if we are confident this would lead to some inconsistency in the public API and don't feel like this is a strong enough reason to move forward with this change, I'm more than happy to drop it 😊

[josevalim]

Right, sum_by may make more sense then but I am not strongly in favor of the function.

[tiagonbotelho]

Right, sum_by may make more sense then but I am not strongly in favor of the function.

Agreed 👍 , as I mentioned this might come down to a matter of preference, so I'm happy to close the PR if we think this shouldn't be added to the Enum public API

[josevalim]

Hi @tiagonbotelho, I am closing the PR due to the points above. Feel free to open up a discussion in the mailing list if you want a more general discussion!