cli: handle EPIPE for version subcommand

[taspelund]

EPIPE handling was added recently via {e}println_nopipe() macros, but wasn't used for the "version" subcommand. This replaces println() calls with println_nopipe().

Before:

➜  oxide.rs git:(main) ./target/debug/oxide version | head -1
Oxide CLI 0.6.1+20240710.0
➜  oxide.rs git:(main) ./target/debug/oxide version | a
zsh: command not found: a
thread 'tokio-runtime-worker' panicked at library/std/src/io/stdio.rs:1021:9:
failed printing to stdout: Broken pipe (os error 32)
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
thread 'main' panicked at cli/src/main.rs:105:10:
called `Result::unwrap()` on an `Err` value: JoinError::Panic(Id(9), ...)
➜  oxide.rs git:(main)

After:

➜  oxide.rs git:(trey/epipe_version) ./target/debug/oxide version | head -1
Oxide CLI 0.6.1+20240710.0
➜  oxide.rs git:(trey/epipe_version) ./target/debug/oxide version | a
zsh: command not found: a
➜  oxide.rs git:(trey/epipe_version)

Fixes: #773

[ahl]

@wfchandler do we know why you lint pr didn't turn this up?

[ahl]

Nailed it

[wfchandler]

@wfchandler do we know why you lint pr didn't turn this up?

@ahl the lint PR isn't merged yet.

[ahl]

@wfchandler do we know why you lint pr didn't turn this up?

@ahl the lint PR isn't merged yet.

Yes, but the code exists in that PR and wasn't flagged: https://github.com/oxidecomputer/oxide.rs/blob/wc/enforce-nopipe/cli/src/cmd_version.rs#L30

[wfchandler]

@wfchandler do we know why you lint pr didn't turn this up?

@ahl the lint PR isn't merged yet.

Yes, but the code exists in that PR and wasn't flagged: https://github.com/oxidecomputer/oxide.rs/blob/wc/enforce-nopipe/cli/src/cmd_version.rs#L30

Oh interesting, this seems like a clippy bug. In my testing this only happens in #[async_trait] impls, maybe something about the proc macro expansion is preventing clippy from recognizing the println!. Will investigate further.

[wfchandler]

@wfchandler do we know why you lint pr didn't turn this up?

@ahl the lint PR isn't merged yet.

Yes, but the code exists in that PR and wasn't flagged: https://github.com/oxidecomputer/oxide.rs/blob/wc/enforce-nopipe/cli/src/cmd_version.rs#L30

Oh interesting, this seems like a clippy bug. In my testing this only happens in #[async_trait] impls, maybe something about the proc macro expansion is preventing clippy from recognizing the println!. Will investigate further.

More specifically, async_trait functions with a return type will fail to trigger the lint. I've opened rust-lang/rust-clippy#13228