Unexpected `cfg` condition value: `no-std`

[trappitsch]

I wanted to use this crate to do some unitful returns of instrument queries and wanted to add torr and millitorr (still used heavily in vacuum gauges) as a unit. However, even just running a check or tests on the main branch results in this error (which is elevated from a warning due to the cfg). Is there something I'm doing wrong or is there something I'm missing here? Thanks!

❯ cargo check
warning: no edition set: defaulting to the 2015 edition while the latest is 2024
    Checking measurements v0.11.0 (/home/reto/Documents/code/embedded/rust-measurements)
error: unexpected `cfg` condition value: `no-std`
  --> src/measurement.rs:39:7
   |
39 | #[cfg(feature = "no-std")]
   |       ^^^^^^^^^^^^^^^^^^
   |
   = note: expected values for `feature` are: `from_str`, `regex`, `serde`, and `std`
   = help: consider adding `no-std` as a feature in `Cargo.toml`
   = note: see <https://doc.rust-lang.org/nightly/rustc/check-cfg/cargo-specifics.html> for more information about checking conditional configuration
note: the lint level is defined here
  --> src/lib.rs:9:9
   |
9  | #![deny(warnings, missing_docs)]
   |         ^^^^^^^^
   = note: `#[deny(unexpected_cfgs)]` implied by `#[deny(warnings)]`

error: unexpected `cfg` condition value: `no-std`
  --> src/measurement.rs:42:7
   |
42 | #[cfg(feature = "no-std")]
   |       ^^^^^^^^^^^^^^^^^^
   |
   = note: expected values for `feature` are: `from_str`, `regex`, `serde`, and `std`
   = help: consider adding `no-std` as a feature in `Cargo.toml`
   = note: see <https://doc.rust-lang.org/nightly/rustc/check-cfg/cargo-specifics.html> for more information about checking conditional configuration

error: could not compile `measurements` (lib) due to 2 previous errors

[thejpster]

It's a bug that old rustc ignored and new rustc complains about - referring to features that do not exist.

[trappitsch]

Ah, okay. I first thought that these lines do nothing, but just removing them let something else fail and I was unclear why until now. Turns out, there's a second issue.

The doctest at the beginning of measurements.rs fails. However, if those lines are below it, the formatting is not good so it doesn't get picked up by cargo test. Removing these lines makes cargo test pick the doc test up, which means the tests now fail, and I got confused.

I got it mostly figured out, will prepare a PR for your consideration :)

Also: I noticed there's no CI setup. If of interest, I could get the going and start another PR. LMK.

Thanks!

[trappitsch]

So, running that doctest in fact opens up a whole other can of worms. First it fails since there's no std library. Adding use core as std; solves that, but now it's complaining that there's no panic handler... Two ideas:

• make it not a ![no_std] doc test and note otherwise that use core as std; is required to run macros in a ![no_std] environment.
• add use core as std; to the doctest, mark it as ignored (plus document why it is ignored)

What do you think?

[thejpster]

I noticed there's no CI setup. If of interest...

Yes, please. We should definitely have CI running. Although if you set up clippy, please pin clippy to a fixed version because I'm tired of "new clippy, new CI breakage" day every six weeks.

make it not a ![no_std] doc test

Yeah, I don't think there's a particular need to have no_std doc-tests. We could set up a no_std example and run it on qemu-system-arm if we really wanted to.