Episode 40

thescalawags's podcast

Endless Summer

YouTube link: https://www.youtube.com/watch?v=T6A3FpfEJG8

Surfboards, suntans, and... Scala?

Your hosts this episode: Dick Wall, Daniel Spiewak, Heather Miller, Seth Tisue

Join us during and between episodes for web-based Scalawags chat on Gitter.

Intro (0:00)

• Daniel utterly disavows any familiarity with pop culture
• Heather is AWOL in Kazakhstan or maybe Kyrgyzstan
• Josh is AWOL because shhhhhhhhh. let the rumors fly!
• Dick Wall isn't welcome in Canada
• Seth escapes giant bird, lives in San Francisco now
• Daniel expounds on the Laurentian Divide

Dick's summer (6:30)

• Dick's messing with Scala and GIS at CIBO
• also, he will sell no wine before its time
  • "how far things have come, how easy it's got" to build and deploy a whole app/site yourself (thank you Heroku)
  • yes Lutz he's using Scala and Play (and play-json)
• also working with Josh writing their Scala World talk about for comprehensions
  • prepare for an onslaught of puns

Scala and Sublime Text (11:40)

• Daniel's improved Scala mode for Sublime "literally fixes everything" and is "definitively second best" (behind Emacs)
  • info: tweet 1, tweet 2, tweet 3
• Dick's still making ENSIME/Sublime improvements, using that combo "80 or 90 percent of the time" when writing Scala
  • but make sure you allow ENSIME huge amounts of RAM

Daniel's summer (14:50)

• surprise, Daniel has been at a "million conferences"
• and moving (just within town)
• and working at Verizon now (since January)
  • with Scala notables such as Rúnar (Bjarnason), Tim (Perrett), Stew (O'Connor), Cody (Allen), ...

Conference calendar (16:20)

• Scala World, September 12-13 in the UK's Lake District
  • Dick defends England's climate, in vain
• Scala by the Bay, November 11-13
  • in San Francisco this time, not Oakland, at Twitter HQ
• Scala Symposium, Oct 30-31 in Amsterdam, co-located with Splash
• and Dick's revving up early for Devoxx's first US conference, March 21-23 in San Jose
  • he's heading the languages track. Scala will be "well represented", but how about up-and-comers like Elm, Pony, Rust? Daniel suggests PureScript.

Seth's Scala news roundup (21:35)

• Scala 2.12.0-RC1 imminent (M5 already out)
• Scala Native is taking off; strong community participation
• Scala Exercises site launched
  • contributions wanted both for Scala, and for 3rd party libraries
• ScalaTest 3.0 is out, with Scala.js support and async testing
• ScalaBridge launched: "introductory programming workshops for women", modeled after RailsBridge
• recent Scala Center activities:
  • several engineers hired
  • Scaladex library index launched and open sourced
  • three MOOCs launched on Coursera
  • SIP (Scala Improvement Process) revised and relaunched by Jorge Cantero: July meeting notes, August meeting notes
  • work on a Scala code migration tool has started

Dotty (24:10)

• Daniel is thinking about the Scala-to-Dotty migration path (Scala 2.x to 3.x)
• Dick's realized that Scala 2 isn't standing still, Dotty features are being backported
• IntelliJ has some level of Dotty support now
  • Dick praises JetBrains for not sacrificing Scala support to Kotlin
• Daniel: getting started with Dotty is a lot easier than it used to be (REPL, sbt support...)
• ENSIME and Dotty? it would need a presentation compiler. this came up at the recent Scala Center advisory board meeting
  • Dotty had this use case in mind from the start, unlike scalac

SI-2712 (33:05)

• "Add support for partial unification of type constructors"
  • SI-2712 got addressed by this PR by Miles Sabin
  • the PR has a lot of discussion and technical details, including an explanation written by Daniel
• what is this SI-2712 thing anyway? Daniel expounds
• Scala would refuse to infer only one type parameter of a higher kinded type with multiple type parameters, because it didn't know which one you mant; now (if you enable -Ypartial-unification) it assumes you meant the first one (left to right)
  • which Daniel describes as "what almost everybody wants almost all of the time"
• do you have to worry whether a library you use was compiled with this or not?
  • binary compatibility: no
  • as an API user: perhaps
• Seth: hey, does somebody want to SIP the kind-projector plugin, which gives a much nicer syntax for type lambdas?
  • Daniel says there are some subtleties and limitations that would need to be addressed (danger, don't get "pessimistic and jaded")

Biased Either (46:10)

• Either is now right-biased (see Simon Ochsenreither's PR)
• is everyone completely pleased? will alternatives such as Cats's Xor go away?
• the change will be in 2.12.0-RC1 and we'll learn more then
• let's fight about how to pronounce everything!

2016 is a very good year (52:25)

• Seth: everything's moving again: 2.12, Dotty, the Scala Center, SIPs...
• and not so much bad blood? we'll see once the SLIPs come back, perhaps
  • including the notorious JSON SLIP
  • and the concept of a "Scala platform", or multiple platforms
• Daniel is "placated" and has "absolutely no objections anymore" to the JSON proposal in the context of a platform

SIP-23: Singleton types (57:00)

• initial implementation by Adriaan Moors and George Leontiev
• Miles Sabin is now trying to polish and finish it on this PR
• Dotty already has this
• Daniel thinks this is "not actually as useful as it seems", especially since you need to use macros to take full advantage
  • there was further discussion on this point on our Gitter channel
• how much actual typelevel programming does this enable?
• the PR will probably make Scala 2.12.1 (behind -Yliteral-types flag or -Xexperimental?)
  • it may appear sooner in Typelevel Scala
• syntax discussion: syntax in PR differs from Dotty syntax? we're not sure
  • we can relax, .type is definitely not required (Daniel's correction)
• semantics discussion: Dotty is far more willing to hang on to singleton types

GIS (Geographical Information Systems) (1:01:45)

• Dick's wanting to use singleton types in a GIS context
• Daniel: Shapeless (or Daniel's own micro-Shapeless called "Shapely") can help you with this

Conclusion (1:06:00)

• play the breakdown, because we're broken-down lamers without Josh