You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
You may want to use an experimental feature in an isolate module without allowing it to infect the entire crate. This is also currently needed for rust-core to use macros, because until we have both static linking and whole program optimization it's going to be a module.
The text was updated successfully, but these errors were encountered:
We had originally explicitly decided to not have module-level control over the feature set, but that decision didn't really have a reason other than "it feels right".
One thing to consider with module-level control is whether modules should the have the ability to deny features? Do we want something like #[feature(+feature1, -feature2)]? I don't think we'd want to do that, but I can imagine that someone will at some point want to do that.
We still don't have the ability for this. It's possibly useful in terms of restricting which parts of a the crate are experimental (e.g. I could imagine most of a crate being stable, with certain parts pushing the boundaries of the type system etc. with feature flags). On the note of infection, rust-lang/rfcs#475 means these will be forced to infect the entire crate with only-usable-with-nightly-itis.
Quite a bit of time has passed since this was opened, and with the current design of feature gates it's intentional that it's only crate-level, so closing.
You may want to use an experimental feature in an isolate module without allowing it to infect the entire crate. This is also currently needed for rust-core to use macros, because until we have both static linking and whole program optimization it's going to be a module.
The text was updated successfully, but these errors were encountered: