show(get_char_or_int(false)); } $ cargo run --quiet C 64 Whoa. Whoa whoa whoa, that could be its own article! Yes. And yet here we are. And there’s unsafe code in there, how do you know it’s okay? Well, miri is happy about it, so that’s a good start: $ cargo +nightly miri run --quiet C 64 And do I really need to write code like that? No you

quiet [src/main.rs:3] s.next() = Some( "1470", ) [src/main.rs:4] s.next() = Some( "1577", ) [src/main.rs:5] s.next() = Some( "1054", ) Hold on a second, I’m getting a warning: Whoa, you have inline error messages? Yeah, it’s the Error Lens vscode extension, it’s pretty neat! Okay so, about that error - it’s not really an error

refresh credentials pub trait CredentialsRefresher { fn refresh<'b>( &'b self, creds: &'b FutileCredentials, ) -> Pin<Box<dyn Future<Output = eyre::Result<FutileCredentials>> + Send + 'b>>; } What’s interesting here is that the returned future is Send, and Unpin (since it’s on the heap), a- Whoa whoa whoa you’re not

results = include_str!("input.txt") .split("\n\n") .map(|input| PassportBuilder::parse(input).and_then(|b| b.build())); let num_valid = results.filter(Result::is_ok).count(); println!("{} passport records were valid", num_valid); } $ cargo run --quiet 194 passport records were valid Whoa, we actually got it right the first

lib64/ld-linux-x86-64.so.2 Whoa. WHOA! It’s a function provided by ld-linux-x86-64.so.2! Yes! glibc’s dynamic linker slash loader! Yes!! Well yeah! It’s freaking dladdr, what did you expect? ld-linux.so, the loader, loads the binary, ls, which is itself linked against libc.so, which ends up calling back into ld.so, which is really what ld

just derive them: #[derive(PartialEq, Debug)] struct PasswordPolicy { byte: u8, range: RangeInclusive<usize>, } Whoa! Does that work because both u8 and RangeInclusive<usize> implement PartialEq and Debug? Yup! And they’re not implemented by default for all new structs because, well, if you don’t need them that’s just extra code in

let IcmpSendEcho: IcmpSendEcho = unsafe { iphlp.get_proc("IcmpSendEcho").unwrap() }; IcmpSendEcho( handle, dest, request_data, request_size, request_options, reply_buffer, reply_size, timeout, ) } Whoa. Okay, yeah, we’re definitely going to need to come back to that. But for

pub fn eval(self, old: u64) -> u64 { match self { Term::Old => old, Term::Constant(c) => c, } } } And then, whoa here we go: use nom::{ branch::alt, bytes::complete::tag, character::complete as cc, character::complete::{one_of, space1}, combinator::{map, value}, error::ParseError

adding cargo-features = ["edition2021"] did the trick). Here are the timings: Cold 1.54.0 build: 1m12s (vs 2m04s for 1.57.0) Hot 1.54.0 build: 18.07s (vs 1m11s for 1.57.0) Whoa. WHOA. Yeah. It’s a pretty bad regression. And it is known. Apparently Rust 1.58 should improve the situation. Does that mean… nightly should do better? $ rustup

O: $ cargo new surviving Created binary (application) `surviving` package $ cargo add argh sha3 color-eyre Updating 'https://github.com/rust-lang/crates.io-index' index Adding argh v0.1.3 to dependencies Adding sha3 v0.9.1 to dependencies Adding color-eyre v0.5.1 to dependencies Whoa, amos, cool off. Why all the