Paweł Marks, VirtusLab
Hello from the Scala 3 team! It has already been six weeks since we have announced the release candidate for the first minor version after the initial release of Scala 3. Now, after two more RCs, we can confidently promote Scala 3.1.0-RC3 to a new stable release of the language. In other words, Scala 3.1.0 is officially out!
Compatibility notice
Scala 3 follows major.minor.patch
versioning scheme (unlike Scala 2, which uses epoch.major.minor
). This means that this is the first minor release after Scala 3.0.0. This has the following consequences:
- Scala 3.1 is backward binary compatible: you can use dependencies compiled with Scala 3.0 in 3.1 projects.
- Scala 3.1 is not forward binary compatible: you cannot use dependencies compiled with Scala 3.1 in Scala 3.0 projects.
Although we cannot guarantee full source compatibility between minor versions, we have put a lot of effort into assuring that all code that was working in 3.0.2, except in some rare cases, will also work in 3.1.0. This means that if you are an application developer, you can confidently update the compiler version to take advantage of the newest improvements. You will still be able to use dependencies compiled with Scala 3.0.
If you are a library maintainer, updating to 3.1.0 will force all of your users to update to 3.1.0 as well. We understand that the current state of binary compatibility may be unsatisfactory. We are actively working on technical solutions to support forward-compatibility in 3.2.0. In the meantime, we recommend testing your library with Scala 3.1.0 and 3.0.2, but publishing it with 3.0.2. This will allow downstream users to use your library even if they can’t update to Scala 3.1.0. You can find an example of build that follows this recommendation here.
What’s new in 3.1
New experimental feature: safer exceptions
A new experimental feature that allows declaring and checking which exceptions can be thrown. It relies on the effects as implicit capabilities pattern.
You can opt-in using the following import.
import language.experimental.saferExceptions
Now it is possible to mark types with the type of exception that can be thrown during the evaluation:
def f(x: Double): Double throws LimitExceeded =
if x < limit then f(x) else throw LimitExceeded()
You can read more in the document proposing this feature.
Efficient bytecode for matching over strings
Scala, since 3.0.0, has been able to optimize pattern matches with Int
as a scrutinee type by emitting bytecode containing very efficient instructions: lookupswitch
or tableswitch
.
Now in Scala 3.1, similar behavior was implemented for matching over String
literals.
fruit match
case "apple" => 1
case "orange" => 2
case "banana" => 3
case _ => 0
will now generate lookupswitch
over hash codes of String
literals instead of a chain of conditional expressions. This behavior is consistent with what Java does and more performant than the old approach.
Support for -Wconf
and @nowarn
We have added a -Wconf
compiler flag that allows filtering and configuring compiler warnings (silence them, or turn them into errors).
We have also brought back @nowarn
annotation from Scala 2.13, originally inspired by the silencer plugin. It can be used for suppressing warnings directly in the code, in the places where they are expected to occur.
-Wconf
and @nowarn
work largely the same way as in Scala 2, but some filters for selecting warnings are different. For the more details see the output of -Wconf:help
flag.
Simplified Manifest synthesis
For the sake of compatibility with Scala 2, the Scala 3 compiler now generates given instances for the Manifest
trait. The new implementation is a slightly simplified approximation of the Scala 2 implementation. It guarantees that any of the expressions:
manifest[A] == manifest[B]
manifest[A].runtimeClass == manifest[B].runtimeClass
optManifest[A] == optManifest[B]
optManifest[A].asInstanceOf[ClassTag[A]].runtimeClass == optManifest[B].asInstanceOf[ClassTag[B]].runtimeClass
that were true in Scala 2, will also remain true when compiled with Scala 3.
Mirrors for Hierarchical Sealed Types
Consider the following sealed hierarchy:
sealed trait Top
case class Middle() extends Top with Bottom
sealed trait Bottom extends Top
Previously you would not be able to summon an instance of scala.deriving.Mirror.SumOf
for Top
, as its child Bottom
is not a case class. In Scala 3.1 this is now possible, enabling compiletime reflection over nested hierarchies of sealed types.
The only change necessary for users is that they should now consider that one of the MirroredElemTypes
for a Mirror.SumOf
could also be a sum type, not just product types.
Other changes
- Scastie was integrated into Scaladoc to make snippets interactive.
@experimental
annotation is no longer itself experimental. The users are no longer required to use a nightly build of the compiler to be able to define experimental APIs. See more detail in the design document- Now
TastyInspector.{inspectTastyFiles, inspectTastyFilesInJar, inspectAllTastyFiles}
return a boolean value indicating whether the process succeeded - A
Wildcard
was made a subtype ofIdent
in the reflection API TypedOrTest
was added as a supertype ofTyped
in the reflection APIUnapply.apply
was added to allow contruction ofUnapply
trees from macros- Unreductible match types now raise type errors
- Scala 3.1 targets Scala.js 1.7.x+. This means that users must upgrade to Scala.js 1.7.0 or later to use Scala 3.1.
Beside that Scala 3.1.0 introduced multiple small improvements and fixed a handful of bugs. You can see the detailed changelog on GitHub.
What’s next
During the Scala 3.1.0 stabilization period, which took the last six weeks, we haven’t stopped improving the language and fixing the bugs in the compiler. You can already test the results of our work, using the published release candidate: Scala 3.1.1-RC1. The full changelog is as always available on GitHub.
You can expect the stable release of Scala 3.1.1 in early December.
Contributors
Thank you to all the contributors who made the release of 3.1.0 possible 🎉
According to git shortlog -sn --no-merges 3.0.2..3.1.0
these are:
61 Martin Odersky
51 tanishiking
50 Nicolas Stucki
39 Kacper Korban
33 Andrzej Ratajczak
22 Dale Wijnand
22 Olivier Blanvillain
21 Lukas Rytz
17 Filip Zybała
16 Yichen Xu
15 Rikito Taniguchi
14 Sébastien Doeraene
12 Krzysztof Romanowski
11 EnzeXing
9 Guillaume Martres
9 Jamie Thompson
8 Tom Grigg
6 Seth Tisue
5 Alec Theriault
4 Michał Pałka
4 Phil
3 Fengyun Liu
3 Paweł Marks
2 Vadim Chelyshov
2 Dmitrii Naumenko
2 Julien Richard-Foy
2 Kevin Lee
2 Liu Fengyun
2 Stéphane Micheloud
2 Tomasz Godzik
2 Adrien Piquerez
2 adampauls
2 noti0na1
1 Kai
1 Jeremy Smith
1 Som Snytt
1 Anselm von Wangenheim
1 Boris
1 Arnout Engelen
1 odersky
1 Anatolii Kmetiuk
1 PJ Fanning
1 Matthieu Bovel
1 Performant Data
1 Mario Bucev
1 Ruslan Shevchenko
Library authors: Join our community build
Scala 3 has a set of widely-used community libraries that are built against every nightly Scala 3 snapshot. Join our community build to make sure that our regression suite includes your library.