Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

SIP - Allow multiple implicit parameter lists #520

Closed
wants to merge 6 commits into from
Closed
Changes from 4 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
@@ -0,0 +1,65 @@
---
layout: sip
disqus: true
title: SIP-NN - SIP Allow multiple implicit parameter lists
---

**By: Miles Sabin and Merlin Göttlinger**

## History

| Date | Version |
|---------------|---------------|
| Apr 21st 2016 | Initial Draft |

## Motivation

Currently, the implicit parameters of a function are limited to being placed in a single, final, parameter list of a function. This prevents later parameters from using dependant types of former ones, as this is only allowed between, but not inside a single parameter list.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/dependant/dependent


This restriction also means that if you want to explicitly specify one of the implicit parameters you have to specify all of them by hand (or write `implicitly` in place of every other parameter that should be filled by the compiler).

## Motivating Examples

### Examples

For regular parameter lists it is possible to write the following:
{% highlight scala %}
trait A { type B }

def works(a: A)(b: a.B) = "works"
{% endhighlight %}
Due to the restriction of being contained in a single parameter list, implicit parameters can't work in the same way. A common workaround is to pull the type member `B` into a type parameter of a new `Aux` object and make this a type parameter of your function:
{% highlight scala %}
trait A { type B }
object A { type Aux[D] = A { type B = D } }

def works[B](implicit a: A.Aux[B], b: B) = "works"
{% endhighlight %}
While this works it introduces unnecessary complexity because our method shouldn't need to have type parameters and we shouldn't need the `Aux` object just to use the type member of `A` for implicits.

With the proposed change of allowing multiple implicit parameter lists the example would again be reduced to the following concise and readable example:
{% highlight scala %}
trait A { type B }

def worksWithImplicits(implicit a: A)(implicit b: a.B) = "pls make me work"
{% endhighlight %}

Another benefit mentioned before would be that implicit parameters, that are expected to often be filled in manually can be put in separate lists.
{% highlight scala %}
def now(implicit a: Int, b: String) = b * a
now(1, implicitly)

def then(implicit a: Int)(implicit b: String) = b * a
then(1)
{% endhighlight %}

## Implementation

The implementation of this change is as simple as removing the restriction of having only one parameter list in the `Parser`. A pull request (#5108) from Miles Sabin already implements this change as well as an accompanying test file. The change should not affect any existing code as every Scala statement that was valid before will still be valid after the change. There will even be binary compatibility in both directions.
The desugaring ob context and view bounds will be added on the left of the leftmost implicit parameter list to go with the general expectation from the current implementation.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/ob/of


## References

1. [Miles' PR][1]

[1]: https://github.com/scala/scala/pull/5108 "PR#5108"