stage2: `@typeName` of errors and error unions is less useful

[leecannon]

Zig Version

0.10.0-dev.3660+4a98385b0

Steps to Reproduce

const std = @import("std");

pub fn main() void {
    std.debug.print("{s}\n", .{@typeName(Thing)});
    std.debug.print("{s}\n", .{@typeName(Error)});
}

const Error = error{ a, b };
const Thing = Error!usize;

Expected Behavior

Behaviour with stage 1:

$ zig build run -fstage1
Error!usize
Error

Actual Behavior

Behaviour with self-hosted:

$ zig build run
error{a,b}!usize
error{a,b}

[nektro]

• Thing behavior seems correct on stage1
• Error behavior seems correct on stage2

[leecannon]

I was prepared for this to be closed with the new behaviour being prefered, but I was not expecting someone to say they want error sets to be printed differently based on whether they are in an error union or not 😕

[nektro]

to be fair @typeName is rather inconsistent as it is, particularly when it comes to aliasing aggregate types to names and when it respects them. so i'd happily accept either behavior