ffi struct produces warnings on Rust 1.79.0 or +nightly #270

sax · 2024-04-24T19:05:17Z

I have a struct that I use to convert wrapped errors to strings that can be sent to Swift:

#[swift_bridge::bridge]
mod ffi {
  #[swift_bridge(swift_repr = "struct")]
  pub struct BridgedError {
    pub inner: String
  }
}

This compiles without any errors in Rust 1.77.2, but when I compile it with nightly it produces the following warning:

> cargo +nightly build -Zbuild-std --package my-bridge --target aarch64-apple-darwin

error: field `0` is never read
  --> rust/my-bridge/src/lib.rs:17:14
   |
14 | #[swift_bridge::bridge]
   | ----------------------- field in this variant
...
17 |   pub struct BridgedError {
   |              ^^^^^^^^^^^^
   |
note: the lint level is defined here
  --> rust/my-bridge/src/lib.rs:2:38
   |
2  | #![cfg_attr(feature = "strict", deny(warnings))]
   |                                      ^^^^^^^^
   = note: `#[deny(dead_code)]` implied by `#[deny(warnings)]`
help: consider changing the field to be of unit type to suppress this warning while preserving the field numbering, or remove the field
   |
17 |   pub struct () {
   |              ~~

I'm starting to play around with cross-compiling to watchOS, tvOS, and eventually visionOS, but am finding these targets only seem to build on nightly.

The text was updated successfully, but these errors were encountered:

chinedufn · 2024-04-28T14:16:54Z

Thanks for reporting this.

Looks like a similar to issue was opened for rustc 3 weeks ago rust-lang/rust#123418

Mind sharing a simple bridge module that reproduces this warning?

I cannot reproduce this warning using the provided example.

Bright-Shard · 2024-04-28T14:17:58Z

Also on nightly some of the ui tests fail to run. It looks like very minor issues - unexpected warnings (maybe related to that issue?) and some of the spans seem to have changed.

Output

test tests/ui/args-into-argument-not-found.rs ... mismatch

EXPECTED:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
error: Argument "arg_typo" was not found in "fn some_function(..)"
 --> tests/ui/args-into-argument-not-found.rs:7:42
  |
7 |         #[swift_bridge(args_into = (arg, arg_typo))]
  |                                          ^^^^^^^^

error: Argument "bar" was not found in "fn some_method(..)"
  --> tests/ui/args-into-argument-not-found.rs:13:42
   |
13 |         #[swift_bridge(args_into = (foo, bar))]
   |                                          ^^^
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈

ACTUAL OUTPUT:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
error: Argument "arg_typo" was not found in "fn some_function(..)"
 --> tests/ui/args-into-argument-not-found.rs:7:42
  |
7 |         #[swift_bridge(args_into = (arg, arg_typo))]
  |                                          ^^^^^^^^

error: Argument "bar" was not found in "fn some_method(..)"
  --> tests/ui/args-into-argument-not-found.rs:13:42
   |
13 |         #[swift_bridge(args_into = (foo, bar))]
   |                                          ^^^

warning: unused variable: `arg`
  --> tests/ui/args-into-argument-not-found.rs:18:18
   |
18 | fn some_function(arg: u8) {}
   |                  ^^^ help: if this is intentional, prefix it with an underscore: `_arg`
   |
   = note: `#[warn(unused_variables)]` on by default

warning: unused variable: `foo`
  --> tests/ui/args-into-argument-not-found.rs:23:27
   |
23 |     fn some_method(&self, foo: u8) {}
   |                           ^^^ help: if this is intentional, prefix it with an underscore: `_foo`
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
note: If the actual output is the correct output you can bless it by rerunning
      your test with the environment variable TRYBUILD=overwrite

test tests/ui/incorrect-argument-type.rs ... mismatch

EXPECTED:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
error[E0308]: mismatched types
  --> tests/ui/incorrect-argument-type.rs:15:16
   |
10 | #[swift_bridge::bridge]
   | ----------------------- arguments to this function are incorrect
...
15 |         fn fn1(arg: &str);
   |                ^^^^^^ expected `u16`, found `&str`
   |
note: function defined here
  --> tests/ui/incorrect-argument-type.rs:24:4
   |
24 | fn some_function(_arg: u16) {}
   |    ^^^^^^^^^^^^^ ---------
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈

ACTUAL OUTPUT:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
error[E0308]: mismatched types
  --> tests/ui/incorrect-argument-type.rs:15:16
   |
10 | #[swift_bridge::bridge]
   | ----------------------- arguments to this function are incorrect
...
15 |         fn fn1(arg: &str);
   |                ^^^^^^^^^ expected `u16`, found `&str`
   |
note: function defined here
  --> tests/ui/incorrect-argument-type.rs:24:4
   |
24 | fn some_function(_arg: u16) {}
   |    ^^^^^^^^^^^^^ ---------
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
note: If the actual output is the correct output you can bless it by rerunning
      your test with the environment variable TRYBUILD=overwrite

test tests/ui/incorrect-return-type.rs ... mismatch

EXPECTED:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
error[E0308]: mismatched types
  --> tests/ui/incorrect-return-type.rs:6:1
   |
6  |   #[swift_bridge::bridge]
   |   ^^^^^^^^^^^^^^^^^^^^^^^ expected `SomeType`, found `&SomeType`
...
9  |           type SomeType;
   |  ______________-
10 | |
11 | |         #[swift_bridge(rust_name = "some_function")]
   | |_________- expected due to this
   |
   = note: this error originates in the attribute macro `swift_bridge::bridge` (in Nightly builds, run with -Z macro-backtrace for more info)

error[E0308]: mismatched types
  --> tests/ui/incorrect-return-type.rs:6:1
   |
6  |   #[swift_bridge::bridge]
   |   ^^^^^^^^^^^^^^^^^^^^^^^ expected `SomeType`, found `Option<SomeType>`
...
9  |           type SomeType;
   |  ______________-
10 | |
11 | |         #[swift_bridge(rust_name = "some_function")]
12 | |         fn fn1() -> SomeType;
13 | |         #[swift_bridge(rust_name = "another_function")]
   | |_________- expected due to this
   |
   = note: expected struct `SomeType`
                found enum `Option<SomeType>`
   = note: this error originates in the attribute macro `swift_bridge::bridge` (in Nightly builds, run with -Z macro-backtrace for more info)
help: consider using `Option::expect` to unwrap the `Option<SomeType>` value, panicking if the value is an `Option::None`
   |
6  | #[swift_bridge::bridge].expect("REASON")
   |                        +++++++++++++++++
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈

ACTUAL OUTPUT:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
error[E0308]: mismatched types
  --> tests/ui/incorrect-return-type.rs:6:1
   |
6  |   #[swift_bridge::bridge]
   |   ^^^^^^^^^^^^^^^^^^^^^^^ expected `SomeType`, found `&SomeType`
...
9  |           type SomeType;
   |  ______________-
10 | |
11 | |         #[swift_bridge(rust_name = "some_function")]
12 | |         fn fn1() -> SomeType;
   | |_____________________________- expected due to this
   |
   = note: this error originates in the attribute macro `swift_bridge::bridge` (in Nightly builds, run with -Z macro-backtrace for more info)

error[E0308]: mismatched types
  --> tests/ui/incorrect-return-type.rs:6:1
   |
6  |   #[swift_bridge::bridge]
   |   ^^^^^^^^^^^^^^^^^^^^^^^ expected `SomeType`, found `Option<SomeType>`
...
9  |           type SomeType;
   |  ______________-
10 | |
11 | |         #[swift_bridge(rust_name = "some_function")]
12 | |         fn fn1() -> SomeType;
13 | |         #[swift_bridge(rust_name = "another_function")]
14 | |         fn fn2() -> SomeType;
   | |_____________________________- expected due to this
   |
   = note: expected struct `SomeType`
                found enum `Option<SomeType>`
   = note: this error originates in the attribute macro `swift_bridge::bridge` (in Nightly builds, run with -Z macro-backtrace for more info)
help: consider using `Option::expect` to unwrap the `Option<SomeType>` value, panicking if the value is an `Option::None`
   |
6  | #[swift_bridge::bridge].expect("REASON")
   |                        +++++++++++++++++
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
note: If the actual output is the correct output you can bless it by rerunning
      your test with the environment variable TRYBUILD=overwrite

test tests/ui/invalid-associated-to-attribute.rs ... ok
test tests/ui/invalid-copy-attribute.rs ... ok
test tests/ui/invalid-module-item.rs ... ok
test tests/ui/opaque-copy-type-mut-ref.rs ... ok
test tests/ui/unrecognized-argument-attribute.rs ... ok
test tests/ui/unrecognized-enum-attribute.rs ... ok
test tests/ui/unrecognized-function-attribute.rs ... ok
test tests/ui/unrecognized-opaque-type-attribute.rs ... ok


test ui_tests::ui ... FAILED

sax · 2024-05-02T01:12:45Z

@chinedufn Took a bit longer than expected, but here's a gist you can clone and see the issue:
https://gist.github.com/sax/9c4b3b841d15ed62fde135da28b783fa

sax · 2024-06-14T19:19:16Z

This error occurs on Rust 1.79.0, which was just released yesterday.

sax · 2024-06-17T18:42:38Z

Ok, I just verified that this problem happens when bridging a function or method returning a Result where the error variant is an ffi struct. When I instead change all of my code to return a Result<whatever, String>, then the compiler warning disappears.

I take advantage of the ? operator quite a bit, so am going to hold off on making the change. I've been going through the swift-bridge code as well as the expansion of my crate (using cargo-expand), but I really can't see where the problem is coming from.

…pe>` (#278) In Rust 1.79.0, dead code warnings are emitted from enums where the compiler does not infer that wrapped data is used, even when the enum has been annotated with `#[repr(C]`. This warning triggers in `swift-bridge` when defining transparent structs or enums that will be returned in error results. This appears to be a regression in the `dead_code` rustc lint. Pending upstream fixes to rustc, this change to `swift-bridge` annotates generated result enums with `#[allow(unused)]`. Fixes #270 ``` error: field `0` is never read | 1 | #[swift_bridge::bridge] | ----------------------- field in this variant ... 3 | enum ResultTransparentEnum { | ^^^^^^^^^^^^^^^^^^^^^ | help: consider changing the field to be of unit type to suppress this warning while preserving the field numbering, or remove the field | 3 | enum () { | ~~ ``` ## Example bridge The following bridge code is the minimal reproduction case: ``` rust #[swift_bridge::bridge] mod ffi { #[swift_bridge(swift_repr = "struct")] struct TransparentErrorStruct(pub String); extern "Rust" { fn rust_func_returns_result_transparent_struct( succeed: bool ) -> Result<(), TransparentErrorStruct>; } } fn rust_func_returns_result_transparent_struct( succeed: bool ) -> Result<(), ffi::ResultTestTransparentStruct> { if succeed { Ok(()) } else { Err(ffi::ResultTestTransparentStruct("failed".to_string())) } } impl std::error::Error for ffi:: TransparentErrorStruct {} impl std::fmt::Debug for ffi:: TransparentErrorStruct { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { write!(f, "{}", self) } } impl std::fmt::Display for ffi:: TransparentErrorStruct { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { write!(f, "{}", self.0) } } ```

Bright-Shard mentioned this issue Apr 28, 2024

Support Option<OpaqueSwiftType> #272

Merged

sax changed the title ~~ffi struct produces warnings on +nightly~~ ffi struct produces warnings on Rust 1.79.0 or +nightly Jun 16, 2024

sax mentioned this issue Jun 17, 2024

Fix dead_code warning when returning Result with bridged type in error case #278

Merged

chinedufn closed this as completed in #278 Jun 21, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ffi struct produces warnings on Rust 1.79.0 or +nightly #270

ffi struct produces warnings on Rust 1.79.0 or +nightly #270

sax commented Apr 24, 2024

chinedufn commented Apr 28, 2024 •

edited

Loading

Bright-Shard commented Apr 28, 2024 •

edited

Loading

sax commented May 2, 2024

sax commented Jun 14, 2024

sax commented Jun 17, 2024

ffi struct produces warnings on Rust 1.79.0 or +nightly #270

ffi struct produces warnings on Rust 1.79.0 or +nightly #270

Comments

sax commented Apr 24, 2024

chinedufn commented Apr 28, 2024 • edited Loading

Bright-Shard commented Apr 28, 2024 • edited Loading

sax commented May 2, 2024

sax commented Jun 14, 2024

sax commented Jun 17, 2024

chinedufn commented Apr 28, 2024 •

edited

Loading

Bright-Shard commented Apr 28, 2024 •

edited

Loading