I ran into a minor problem with code like this:

#[dir_test(
    dir: "$CARGO_MANIFEST_DIR/test_files/syntax/func",
    glob: "*.sntn"
)]
fn test_func(fixture: Fixture<&str>) {
    test_rule(Rule::functions, fixture)
}

#[dir_test(
    dir: "$CARGO_MANIFEST_DIR/test_files/syntax/module",
    glob: "*.sntn"
)]
fn test_module(fixture: Fixture<&str>) {
    test_rule(Rule::module, fixture)
}

where both of the directories contain a file called "simple.sntn", and thus two test functions called "simple" are generated. I could have manually specified a suffix to the dir_test attribute, but I think it's better for this situation to work by default. I solved this by prefixing the generated test names with the name of the function that the dir_test attribute is attached to; in this case the generated tests are test_func__simple and test_module__simple.

This change is also makes the cargo test output more clear when there are many test functions specified, and files with generic name. Eg

test test_item_list__use ... ok
test test_pat__record ... ok
test test_struct__empty ... ok
test test_stmt__assign ... ok
test test_struct__attr ... ok
test test_stmt__while ... ok
test test_stmt__for ... ok
test test_pat__path_tuple ... ok