Minutes: SIP Meeting December 2017


#1

Dear Contributors,

Some nice advancements were made in the last SIP meeting. If you are curious to watch the meeting, here is the YouTube link

We had a pleasure to welcome a new Committee member Miles Sabin (@milessabin) to the SIP team!
On top of that, the SIP 23, lead by Miles got accepted so did the SIP-33 (by @soronpo).
As usual, the minutes are published on our website but do scroll down to find out more (o:

Wishing you a great holiday period!
Darja


title: SIP Meeting Minutes - 6th December 2017


Minutes

The following agenda was distributed to attendees:

Topic Reviewers Accepted/Rejected
Discussion and voting on Miles Sabin (Typelevel representative) joining the Committee Accepted
SIP 23: Literal-based singleton types Adriaan Moors Accepted
SIP-33: Priority-based infix type precedence rules Josh Suereth Accepted
SIP-NN: Adding prefix types Josh Suereth Pending
SIP-35: Opaque types Sébastien Doeraene Not discussed
Discussion about the future of Scala 2.13 and 2.14 Not discussed

Jorge Vicente Cantero was the Process Lead and Darja Jovanovic as secretary.

Date and Location

The meeting took place on the 6th December 2017 at 5 PM CEST via Google Hangouts at EPFL in Lausanne, Switzerland as well as other locations.

Watch on Scala Center YouTube channel

Minutes were taken by Darja Jovanovic.

Attendees

Proceedings

Opening Remarks

Jorge opens up the meeting by announcing Miles Sabin as a new Committee member, Miles joins the Committee meeting.

SIP 23: Literal-based singleton types

YouTube time 6’44’’

Miles introduces the SIP, giving a progress overview after taking charge of the SIP and about its implementation in Typelevel Scala.
He is reasonably confident that “the documentation matches the PR and that PR matches the people’s expectations”.
Adriaan clarifies that this PR finally brings to the language users features that have been, so far, commonly used inside of the compiler. Proper documentation and implementation on how type inference interacts with these features were crucial in the process.

Jorge asks for a comparison between Miles’ implementation and the Dotty one.
They are now aligned in Miles’ opinion, but he noticed some implementation differences in Dotty, notably “the use of the Singleton bound on a type variable to allow singleton type to be inferred”

Sébastien YouTube time: 15’41’’ - 21’30’’ challenges the meaning of asInstanceOf in the text of the SIP, asking for the clarification between what it does and what it should do.
Martin points out that it is might be misunderstood, and continues by explaining that asInstanceOf never does an equality test and in general it does the minimum amount of work to satisfy the underlying platform “JVM” or “JS”.
The consensus is that asInstanceOf should be corrected to which
Adriaan adds that “spec says that asInstanceOf is a pattern matching” and that’s where the change needs to happen.
The Committee members agree the SIP is ready to be voted for, given the track record and it’s actual use in the community.

Conclusion : The SIP-23 is accepted by unanimity. The “asInstanceOf” should be changed in the SIP text.

See also:
Brief explanation about the "The presence of an upper bound of Singleton on a formal type parameter 3rd point in the SIP YouTube time 14’ to 15’35’’ and YouTube Time 17’42’’ to 18’20’’

SIP-33: Priority-based infix type precedence rules

YouTube time from 1’34’’ - 6’45’’

Jorge shortly introduces the SIP and notifies the Committee that the author has amended all the changes as per Committee suggestions. The SIP-33 was split in two SIPs as follows:

a) SIP-33: Priority-based infix type precedence rules

b) SIP-NN: Adding prefix types

Seth asks about the implementation status in Dotty and if there are any crucial differences in Scala 2 and Dotty?
Martin and Sebastien agree there are none in regards to this SIP.
The members are all in favour for this change and proceed to voting.

Conclusion : The SIP-33 is accepted by unanimity.

SIP-NN: Adding prefix types

YouTube time: 25’00 until the end

Jorge introduces the SIP’s development, based on the idea
(in Oron’s words) “it is easier to reason about the language when mathematical and logical operations for both terms and types are expressed the same”; goes over the motivation examples Oron proposed since the last SIP (splice prefix types for meta programing; singleton-ops library and DFiant library example) and opens the discussion about the recent use-cases.

Martin starts with by introducing his PR, the use-case in Dotty, “Principled Meta Programming”. Sébastien argues that Martin’s use-case is not really related to what the SIP-NN is aiming to achieve, but in this context he would rather agree on special-casing the ˜ for macros and splices.

However, Sebastian gives his preference to the SIP itself.
Martin, on the other hand, is “dubious” about the SIP, stating that in Scala those 4 operators were defined originally because the syntax in Java. He disagrees with now making a step even further - elevating them to the principal. He is also sceptical because he foresees the “end-operator misuse” to which Adriaan adds the issue between annotations and variants that could make the confusion even deeper.
Seth and Eugene agree it could be too confusing, and even though there is a potential in unifying the language features, this particular SIP doesn’t seem to address it in a clear and persuasive way.

Miles proposes to let this SIP have its implementation in Typelevel Scala. He believes the arguments raised in this discussion could be tested and eventually even answered or “shaped” by the user’s experience. He asks to defer the discussion until the use-case is ready. Adriaan supports the idea but underlines that it is important to know that implementation should not be considered as a guarantee leading to be a part of the language.

The Committee proceeds with voting on numbering the SIP.

Conclusion: The SIP-NN is numbered, from now SIP-36, it will be discussed once results of the Typelevel implementation are ready.

To be discussed

Other announced agenda items were not discussed in this meeting because of the lack of time. They will be addressed in the next meeting.