-Zunpretty=identified runs too early

[huonw]

fn main() { println!("hello world"); }

rustc -Zunpretty=identified:

fn main() { println!("hello world"); } /* block 4294967040 */ /* 4294967040 */

Clearly it's useless if the NodeIds are all NodeId::MAX_AS_U32 (4294967040 aka -256_i32 as u32) (I think it may just have to be removed in favour of -Zunpretty=expanded,identified since the NodeIds are only assigned after macro expansion, and only correspond to elements of that AST, not the pre-expansion one).

[steveklabnik]

Running this on our stock Hello world gives

fn main() { println!("Hello, world"); } /* block 4294967295 */ /* 4294967295
*/

I would think that this would be okay to close, then. What do you think, @huonw ? I mean, the future of --pretty is in question too, isn't it?

[huonw]

It's still useless, that number is just (-1) as u32; these sorts of compiler-internals flags like --pretty are likely to stick around (at least as "unstable" options) for compiler hacking/debugging.

[frewsxcv]

Visiting for triage: the output has not changed

coreyf@frewbook-air /tmp> cat hello-world.rs
fn main() {
    println!("Hello world!");
}

coreyf@frewbook-air /tmp> rustc -Z unstable-options --pretty identified hello-world.rs
fn main() { println!("Hello world!"); } /* block 4294967295 */ /* 4294967295
*/

[steveklabnik]

Triage; still no change

[steveklabnik]

Triage: no changes

[jyn514]

Triage: this now prints

fn main() { println!("Hello world!"); } /* block 4294967040 */ /* 4294967040
*/

which is -256 as u32 and still pretty unhelpful.

[jieyouxu]

Triage: no changes