`warning: failed to parse serde attribute #[serde(rename_all = "camelCase")]` on enum struct variant

[ctison]

`warning: failed to parse serde attribute #[serde(rename_all = "camelCase")]` on enum struct variant

[escritorio-gustavo]

The rename_all fix was introduced in #122

[linulas]

It seems to be an issue with most serde field attributes, I get the same with #[serde(skip_serializing_if = "Option::is_none")], and there's a pull request for fixing the same issue for [serde(with = "...")]: #146

[escritorio-gustavo]

This has already been fixed by #185, sorry I failed to include this issue in the PR before it was merged, I tried to add all issues about this, but I seem to have missed this one.
@ctison, if you see this, please close the issue as completed

[NyxCode]

As @escritorio-gustavo mentioned, the no-serde-warnings feature can disable these warnings.
I think I'd still like to keep this issue open, in case we want to support serde's rename_all in the future.

[NyxCode]

right now, it's sadly neccessary to rename each field individually with #[serde(rename = "..")] or #[ts(rename = "..")]

[escritorio-gustavo]

My mistake, due to #165 I thought #[serde(rename_all = "..")] was generally supported and this issue was just older than support for it, I didn't notice it didn't apply to enums

[escritorio-gustavo]

right now, it's sadly neccessary to rename each field individually with #[serde(rename = "..")] or #[ts(rename = "..")]

Hey @NyxCode... I have forked the repo to start working on a fix.
I wrote the following test:

use serde::{Serialize, Deserialize};
use ts_rs::TS;

/// Used enum from issue to test
#[derive(Serialize, Deserialize, TS, Clone)]
#[ts(export)]
#[serde(rename_all = "camelCase")]
pub enum TaskStatus {
  #[serde(rename_all = "camelCase")]
  Running { started_time: String }, // Changed to string because the `chrono` dependency doesn't include the `serde` flag
  #[serde(rename_all = "camelCase")]
  Terminated {
    status: i32,
    stdout: String,
    stderr: String,
  },
}

#[test]
pub fn enum_struct_variant() {
    assert_eq!(TaskStatus::inline(), r#"{ "running": { startedTime: string, } } | { "terminated": { status: number, stdout: string, stderr: string, } }"#)
}

And the test is passing... Has the rename_all been fixed in the time since this issue was opened in a forgotten commit or am I doing something wrong?

[escritorio-gustavo]

The format_field function used to format enums containng structs does receive a rename_all parameter

fn format_field(
    formatted_fields: &mut Vec<TokenStream>,
    dependencies: &mut Dependencies,
    field: &Field,
    rename_all: &Option<Inflection>,
    generics: &Generics,
) -> Result<()> 

[NyxCode]

@escritorio-gustavo Hey!
Thanks for testing that, it seems you're right!
Maybe I got tripped up by an inconsistency our camelCase implementation has with serde (I'll add an other issue for that), idk.

Since you've already got the test, could you make a PR for that?