Given the "unsafe guts" most of Rust stdlib contains, there's going to be some baseline of unsafe code that every project has. I could thus see this confusing new authors and users - you could have a project with zero unsafe lines that will still report a relatively large unsafe surface.
Besides, perhaps this would reinforce the notion that unsafe is - by its very nature - a bad thing. To quote Klabnik:
> it turns out actix-web is using a lot of unsafe code. That’s not inherently a problem. The problem is, a lot of that unsafe code is not actually needed.
If the code were actually needed, then there's no benefit in printing this metric - it just makes a project that is using the feature perfectly correctly look bad.
Besides, perhaps this would reinforce the notion that unsafe is - by its very nature - a bad thing. To quote Klabnik:
> it turns out actix-web is using a lot of unsafe code. That’s not inherently a problem. The problem is, a lot of that unsafe code is not actually needed.
If the code were actually needed, then there's no benefit in printing this metric - it just makes a project that is using the feature perfectly correctly look bad.