typelevel/general

General topics relating to the the Typelevel project http://typelevel.org/ | See also individual project channels | Code of conduct https://www.scala-lang.org/conduct/

Rob Norris

@tpolecat

Is there a reference for what -language:Scala2 does?

Guillaume Martres

@smarter

nope

also that was renamed Scala2Compat recently, I should fix that.

we'll need to make an exhaustive list at some point

Rob Norris

@tpolecat

cool

Guillaume Martres

@smarter

currently it allows some deprecated stuff to compile (like procedure syntax), changes the semantics of implicit search to mimic scala 2 more closely (cf http://dotty.epfl.ch/docs/reference/changed-features/implicit-resolution.html), and loosens some checks like allowing some unsound variance usage, overriding def foo() with def foo and vice-versa, etc

pretty soon it's going to affect whether context bounds are desugared as implicit parameters or using the new syntax for implicit parameters, the only difference being that at call-site one needs a special syntax to pass implicit parameters explicitly

@travisbrown you might hit that last point in cats btw ^

Dale Wijnand

@dwijnand

Just to put it explicitly -language:Scala2 means (IIUC) "accept more things", so quite different to the flag I was referring to, which is to restrict dotc to a subset of the language.

Rob Norris

@tpolecat

Oh ok I completely misunderstood then.

Dale Wijnand

@dwijnand

Maybe it needs to be -language:+Scala2 and -language:=Scala2 :trollface:

Rob Norris

@tpolecat

I want to convert my library project to Scala 3 and compile it in such a way that I can use Scala 3 features, yet retain Scala 2 compatibility.

If I can't do this then I'm stuck on Scala 2 until EOL

Travis Brown

@travisbrown

@tpolecat Why not do the same thing we do for every new Scala 2.x release and have some version-specific source where it's necessary / high-value?

Dale Wijnand

@dwijnand

Yeah, I think for that you need to wait on the flag coming into existence, and also @bishabosha's work, which is adding TASTy reader support to Scala 2.

Sorry :(

Rob Norris

@tpolecat

@travisbrown yeah I think that's what I'll end up doing, but it's delicate work. Really my concern is about positioning. This isn't what people think of when they hear "two-way compatibility".

I should be able to cross-compile atto so I'll start with that one and see how it goes. I think probably it will work fine with no changes because it doesn't do anything fancy. That will help my confidence.

I guess it's more of an issue for library maintainers and there aren't many of us so maybe it's really not a huge deal.

Ross A. Baker

@rossabaker

I assume these interpolators support the verify-literals-at-compile-time pattern? I've had a hard time keeping up with Dotty.

Guillaume Martres

@smarter

yes, that's the whole point of the f-interpolator

@tpolecat so the two-way thing isn't done yet, the scala 3 to 2 way is wip

Rob Norris

@tpolecat

gotcha

Christopher Davenport

@ChristopherDavenport

Is sbt 1.3.7 seemingly really flaky on artifact downloads. Travis has been all over the place on the projects I’ve upgraded it on.

Ross A. Baker

@rossabaker

I've never seen a straight answer on what's going on, but have wondered if it's rate limiting. I don't know that 1.3.7 made it worse.

som-snytt

@som-snytt

I wonder if related: "message" : "'scala-ci' is blocked due to recurrent login failures, please try again in 4 seconds"

Jakub Kozłowski

@kubukoz

1.3.7 has been pretty good for me

Dale Wijnand

@dwijnand

I think Maven has gotten stricter on rate limiting. I think the release of sbt 1.3.7 is just coincidental.

Travis Brown

@travisbrown

Anyone have objections to publishing discipline-scalatest 1.0.0-RC4 as 1.0.0 today? https://github.com/typelevel/discipline-scalatest/releases/tag/v1.0.0-RC4

It's been in Cats and Circe since last week, and I just tried it out in several other projects without running into anything unexpected:

Michael Pilquist

@mpilquist

RFC on scodec-bits 2.0: scodec/scodec-bits#141

Ross A. Baker

@rossabaker

Looks like maybe Travis CI is adding IPs faster than Sonatype is whitelisting them? https://travis-ci.community/t/continuous-maven-repo-403/3908/37

There's a mirror proposed at the bottom, but it only refreshes every couple hours, which I presume would be additive to the Sonatype sync lag.

@travisbrown No objections on discipline-scalatest release.

Travis Brown

@travisbrown

Okay, kicking off discipline-scalatest 1.0.0 now!

Travis Brown

@travisbrown

https://github.com/typelevel/discipline-scalatest/releases/tag/v1.0.0

Jakub Kozłowski

@kubukoz

🎉🎉🎉🎉🎉🎉

Michael Pilquist

@mpilquist

Any plans to update ScalaCheck so that Shrink doesn't use the deprecated Stream type?

Looks like it was punted to 1.15 back in typelevel/scalacheck#411

Seth Tisue

@SethTisue

I highly doubt anybody is working on it

Ross A. Baker

@rossabaker

Is that in the public API? The thought of a breaking scalacheck change makes me dizzy.

Michael Pilquist

@mpilquist

yeah, it's public api of Shrink

Ross A. Baker

@rossabaker

So fixing that would be a breaking change to scalacheck, and therefore discipline, and therefore cats-laws and cats-effect-laws, right?

Would also roil specs2, though I think there are less public testing packages built on that.

Michael Pilquist

@mpilquist

Yep!

Ross A. Baker

@rossabaker

So fixing that deprecation sounds a lot like forcing a Cats 3 and rebuilding the world again.

What pain is that causing now? Do we have examples of projects that can't turn on fatal warnings because of it?

I'm in favor of fixing it for The Next Round Of Change, but I'm not excited for that to be the cause of The Next Round Of Change.

Michael Pilquist

@mpilquist

any project that provides a Shrink instance

Eg scodec-bits

Ross A. Baker

@rossabaker

So what should we do about it? Just turn off fatal warnings on those modules and be sure this gets cleaned up in the Cats 3 Era?

Where communities thrive

People

Repo info

Activity