Fixes #28507: Add ADR for Handling IO encoding in Rust modules

[m4rtinh4rt]

https://issues.rudder.io/issues/28507

[m4rtinh4rt]

PR updated with a new commit

[m4rtinh4rt]

PR updated with a new commit

[m4rtinh4rt]

PR updated with a new commit

[amousset]

only on Windows or everywhere?

[m4rtinh4rt]

I think we want to use those functions everywhere to abstract the OS-specific logic (BOM vs. non-BOM) so we don’t have to repeat it in every module.

[amousset]

why only on Windows? What would be the behavior on Linux?

[m4rtinh4rt]

The current code in rudderc adds an UTF-8 BOM only for the Windows backend.

impl Backend for Windows {
    fn generate(
        &self,
        technique: Technique,
        resources: &Path,
        _standalone: bool,
    ) -> Result<String> {
        // Powershell requires a BOM added at the beginning of all files when using UTF8 encoding
        // See https://docs.microsoft.com/en-us/windows/desktop/intl/using-byte-order-marks
        // Bom for UTF-8 content, three bytes: EF BB BF https://en.wikipedia.org/wiki/Byte_order_mark
        const UTF8_BOM: &[u8; 3] = &[0xef, 0xbb, 0xbf];
        let mut with_bom = String::from_utf8(UTF8_BOM.to_vec()).unwrap();
        with_bom.push_str(&Self::technique(technique, resources)?);
        Ok(with_bom)
    }
}

It is not needed on Linux and could potentially break things.

[m4rtinh4rt]

PR updated with a new commit

[m4rtinh4rt]

PR updated with a new commit

[m4rtinh4rt]

PR updated with a new commit