Is StdoutLock guranteed to be reentrant?

[FeldrinH]

Location

https://doc.rust-lang.org/std/io/struct.Stdout.html#method.lock
https://doc.rust-lang.org/std/io/struct.Stderr.html#method.lock
https://doc.rust-lang.org/std/io/struct.Stdin.html#method.lock

Summary

Currently code like this is fine and a convenient way to lock stdout to prevent other threads from interrupting while printing is happening (this is especially useful when the printing happens deep inside some library/component, where StdoutLock can't be easily passed as an argument):

let stdout = io::stdout().lock();
println!("a");
println!("b");
println!("c");
drop(stdout);

As far as I understand, this works because Stdout::lock() is reentrant. However, the documentation does not mention that Stdout::lock() is reentrant.

Is Stdout::lock() guaranteed to be reentrant in all future Rust versions, or is this an implementation detail that should not be relied on?
Either way it would be nice if the documentation of Stdout::lock() made it clear what the expected behavior is and how much of it is an implementation detail.

The same question also applies to Stderr::lock() and Stdin::lock().

[workingjubilee]

As far as I understand, this works because Stdout::lock() is reentrant

Yes, it is a thin wrapper on a ReentrantLockGuard.

It seems reasonable to document this guarantee given that we broke a couple crates to maintain this:

• Add manual Sync impl for ReentrantLockGuard #125527
• regression: RefCell<LineWriter<std::io::stdio::StdoutRaw>> cannot be shared between threads safely #127340