syn::Field.attrs doesn't contain inert attributes - rust

Question

Given the following struct definition:
#[derive(Builder)]
pub struct Command {
executable: String,
#[builder(each = "arg")]
args: Vec<String>,
#[builder(each = "env")]
env: Vec<String>,
current_dir: Option<String>,
}
(where the #[builder(...)] attributes are inert attributes defined by my Builder derive macro implementation),
the Builder macro cannot "see" the #[builder(...)] field attributes. The following code
let name = &f.ident;
let attrs = &f.attrs;
println!("Attributes: {:?}: {}", name, attrs.len());
(where f is a syn::Field) gives
Attributes: Some(Ident { ident: "executable", span: #0 bytes(2830..2840) }): 0
Attributes: Some(Ident { ident: "args", span: #0 bytes(2854..2858) }): 0
Attributes: Some(Ident { ident: "env", span: #0 bytes(2877..2880) }): 0
Attributes: Some(Ident { ident: "current_dir", span: #0 bytes(2899..2910) }): 0
(Full code: https://github.com/pdkovacs/forked-proc-macro-workshop/blob/a471007968f974ea3c1c684cc47a77fbd20b91dc/builder/src/lib.rs)
Can anybody help me find out what I am missing here, please?

Answer 1

I think you might just be interpreting the terminal output incorrectly since when I cloned your fork and removed the test code that failed to compile this is the output I got for the test with the attributes:
test tests/07-repeated-field.rs ... ok
WARNINGS:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
warning: unused import: `builder_trait:: Builder`
--> tests/07-repeated-field.rs:32:5
|
32 | use builder_trait:: Builder;
| ^^^^^^^^^^^^^^^^^^^^^^^
|
= note: `#[warn(unused_imports)]` on by default
warning: 1 warning emitted
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
STDOUT:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
Attributes: Some(Ident { ident: "executable", span: #0 bytes(1471..1481) }): 0
Attributes: Some(Ident { ident: "args", span: #0 bytes(1524..1528) }): 1
Attributes: Some(Ident { ident: "env", span: #0 bytes(1576..1579) }): 1
Attributes: Some(Ident { ident: "current_dir", span: #0 bytes(1598..1609) }): 0
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
STDERR:
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
warning: unused import: `builder_trait:: Builder`
--> /home/cassy/projects/forked-proc-macro-workshop/builder/tests/07-repeated-field.rs:32:5
|
32 | use builder_trait:: Builder;
| ^^^^^^^^^^^^^^^^^^^^^^^
|
= note: `#[warn(unused_imports)]` on by default
┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈
To be clear I completely commented out the body of main in 07-repeated-field.rs since it was failing to compile, and doing so allowed me to see this output.
I haven't used trybuild before but it seems like if a test fails to build then it doesn't give you the output from the proc macro invocation. Moreover it displays the STDOUT/STDERR of the test compilation process after it displays the test name, which likely was the source of the confusion.