Conflicting constified module enum/rustified enum regex leads to compile error when C enum has non-unique discriminant values

[tmfink]

Input C/C++ Header

typedef enum MyDupeEnum {
	A = 0,
	A_alias = 0,
	B,
} MyDupeEnum;

Bindgen Invocation

$ bindgen input.h test-constify-mod-enum.h --rustified-enum '.*' "--constified-enum-module" ".*"

Actual Results

/* automatically generated by rust-bindgen */

pub const MyDupeEnum_A_alias: MyDupeEnum::Type = MyDupeEnum::A;
#[repr(u32)]
#[repr(C)]
#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)]
pub enum MyDupeEnum {
    A = 0,
    B = 1,
}

and/or

error[E0223]: ambiguous associated type
 --> /tmp/a.rs:3:31
  |
3 | pub const MyDupeEnum_A_alias: MyDupeEnum::Type = MyDupeEnum::A;
  |                               ^^^^^^^^^^^^^^^^ ambiguous associated type
  |
  = note: specify the type using the syntax `<MyDupeEnum as Trait>::Type`

error: aborting due to previous error

Expected Results

There are several possibilities for correct behavior:

1. Error out immediately because regex passed to --rustified-enum/--constified-enum-module overlap.
2. Define the precedence of enum strategies (rustified, constified module, and constified).
3. Support a more complete way for users to convey what enum strategy is used for a given enum. For example, bindgen could take a callback or struct that implements a ConveysEnumStrategy trait).
4. Whichever regex is passed first has precedence (the order of arguments/builder pattern calls matters). This would complicate the internal bindgen logic.

Example of behavior 2:

/* automatically generated by rust-bindgen */

#[repr(u32)]
#[repr(C)]
#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)]
pub enum MyDupeEnum {
    A = 0,
    A_alias = 0,
    B = 1,
}

[fitzgen]

Thanks for the bug report @tmfink !

I think we should define a precedence and go with that.

[tmfink]

When can we expect a new bindgen release? I want to release a new version of capstone-sys that takes advantage of these bug fixes.

[emilio]

I just bumped the version, will publish it in a sec.

[tmfink]

Thanks!

…

On Sun, Mar 25, 2018 at 2:23 PM, Emilio Cobos Álvarez < ***@***.***> wrote: I just bumped the version, will publish it in a sec. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#1198 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFBYuBmJcQOJh1J630Be-e3Lr0hKAfMzks5th-CTgaJpZM4RMezT> .