Scala 3's -Yexplicit-nulls

[alexandru]

Hello,

I noticed that version 0.5.0 doesn't enable -Yexplicit-nulls on Scala 3.

Any reasons for it?

[armanbilge]

Is it ready for use? e.g. @SethTisue on Discord a few months ago:

seconded that -Yexplicit-nulls isn't really that usable yet, but I just want to note that it's being worked on: https://github.com/lampepfl/dotty/pull/18112

https://discord.com/channels/632150470000902164/632150470000902166/1131611679566872698

[SethTisue]

Anyway, in general -Y flags should be considered guilty until proven innocent. -Y is a signal that something is internal and/or not-ready-for-prime-time

[alexandru]

I'd enable it 🤷‍♂️

sbt-tpolecat is a plugin that enables errors in case of unused imports or unused variables, and there's nothing more annoying than that.

The “platform types” in Kotlin are well implemented, as the compiler is also looking at @NonNull and @Nullable annotations from Java's ecosystem. I hope that Scala implements that as well, but meanwhile, while we're waiting, I want explicit nulls for protection against Java APIs, and I don't mind using .nn all over the place.

[armanbilge]

IMHO this issue also seems like a major impediment to using explicit nulls.

• https://github.com/lampepfl/dotty/issues/15891

This use of Option is idiomatic for handling possibly-null values, so it seems ironic that it isn't available when using explicit nulls.

[alexandru]

That untagged unions aren't used yet in the stdlib is such a pity, as it would improve type safety by a lot.

[kubukoz]

Anyway, in general -Y flags should be considered guilty until proven innocent. -Y is a signal that something is internal and/or not-ready-for-prime-time

honestly the compiler/language doesn't do a great job at marketing this, I've worked in scala for 9 years and it's the first time I'm hearing that these aren't prime time 😅