If your IDE defaults to using CRLF instead of LF, which generally happens in Windows, the input str has an extra character at the end on all lines except the last one.

Like in the image, line 1 shows a length of 22 while it's 21, and line 2 shows the correct length as it is the last. It would be nice to see this template handle that issue as it causes regex expressions using \n to not work properly.

Thank you for creating this template!

I have a very small request for a nice-to-have feature. After setting up my local copy of the template, l find myself running cargo scaffold <day> and, right after, I am running cargo download <day>.. I think a new flag to cargo scaffold like --download or even a separate command like cargo generate <day> that combines both steps would be nice.

What do you think?

aoc-cli supports submitting inputs with a command. It would valuable to support automatic submissions for solve behind a flag, e.g. cargo solve 04 --submit.

https://www.theserverside.com/feature/Why-GitHub-renamed-its-master-branch-to-main

https://github.com/scarvalhojr/aoc-cli/tree/main/aoc-client

This would reduce the amount of setup work required and also make this interaction a bit safer. On the downside, we would now have to compile the aoc_client crate before running stuff, which is fairly slow.

Implementation

Currently experimenting with a higher sample count for benching on my solution repo, might add to template after this year concludes. Unfortunately, using criterion with binary crates is not possible yet, would make this a bit simpler.

Advantages:

• more accurate timings

Disadvantages:

• Solutions take (at least) 2 seconds to execute (min. 1 sec or 5 iterations per part)
• Input to benched functions required to be Copy (not an issue so far).

Example output:

🎄 Parser 🎄
✓ (avg. time: 34.30µs / 34507 samples)
🎄 Part 1 🎄
71506 (avg. time: 5.00ns / 3430531 samples)
🎄 Part 2 🎄
209603 (avg. time: 5.22µs / 125659 samples)

The download script run aoc

But because the package's name is also aoc, when it is built it generates a aoc.exe that shadows aoc-cli.

Tried cargo day 01 having created day 01 with cargo scaffold 1

Also tried adding the missing alias

[aliase]
day = "run --bin --"

But no joy I get

cargo day 01
error: "--bin" takes one argument.
Available binaries:
    01
    aoc
    download
    scaffold

First of all, thank you for an amazing starter template. This repo makes it so easy to try out Rust for advent of code!

In AOC 2023 the first day puzzle has two example inputs, one for each part. The default read_file function only accepts a day which is strongly typed so you can only ever read 01.txt, etc.

In my project I just added a new function that takes a part number too:

pub fn read_file_part(folder: &str, day: Day, part: u8) -> String {
    let cwd = env::current_dir().unwrap();
    let filepath = cwd.join("data").join(folder).join(format!("{day}-{part}.txt"));
    let f = fs::read_to_string(filepath);
    f.expect("could not open input file")
}

This template uses bash scripts for automation which does not work on windows. Add equivalent powershell or implement scripts in rust to work around.

@NangiDev mentioned:

PS: I saw this Youtuber fetching the example input somehow 🤔 so that should be possible

Edit: I did quick test on one way of fetching example input. Gist
It's not perfect because it highly depends on the text in puzzle description being typed in a particular way. It works for all (1-7) puzzles release so far this year (2022)

If it turns out that this is consistent, it would be neat to support this. We already download the puzzle by default now, maybe we can parse it from the markdown.

I decided personally it was a good idea to cache the crates.io index.

Do you think it's a good idea to do this for the template?

• A disadvantage is that the index (which is about 300MB) is now stored on Github's servers.
• An advantage is now the action doesn't have to fetch the entire crates.io index every time, which takes too long for my impatient self.

Example:

            - uses: actions/cache@v3
              with:
                path: |
                    ~/.cargo/registry/index/
                key: ${{ runner.os }}-cargo-${{ hashFiles('**/Cargo.lock')

I had this idea of trying to store benchmarks in a file when running them with a specific command.

example on command: cargo bm 02
My cargo .config

That would call on a rust binary that runs the benchmarking command with --release flag and stores it into a txt-file.

I started on the feature here
My attempt

But could not really finish it in a smart and reliable way.

The idea is that inside /src/benchmarks/table.txt should contain a line per day.
Containing Day, latest and best score for both parts.

Ex: 01 L1:85.40µs B185.40µs L2:74.10µs B2:74.10µs
first being the day (01). L1 being latest part 1 and B1 being best part 1 and so on.
My table so far

I'm pretty new at Rust so I think I sort of tangled myself into a weird mess.
My biggest problem is how to replace and update the correct line of the file.

What do you think? Anyone have a better and simpler way of doing this? Is it even valuable?

Implementation

Currently experimenting with a separately timed parser util on my solution repo, might add to template after this year concludes.

Advantages:

• parser code is not timed twice.
• parsing set up in template.
• can use macro to generate main().

Disadvantages:

• input has to be Clone.
• replacing the macro (if needed for some reason) might not be clear for rust beginners (which this is aimed at).

type Input<'a> = Vec<&'a str>;

fn parse(input: &str) -> Input {
    input.lines().collect()
}

pub fn part_one(input: Input) -> Option<u32> {
    None
}

pub fn part_two(input: Input) -> Option<u32> {
    None
}

advent_of_code::main!(4);

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_part_one() {
        let input = advent_of_code::read_file("examples", 4);
        assert_eq!(part_one(parse(&input)), None);
    }

    #[test]
    fn test_part_two() {
        let input = advent_of_code::read_file("examples", 4);
        assert_eq!(part_two(parse(&input)), None);
    }
}

Example output:

🎄 Parser 🎄
(elapsed: 24.96µs)
🎄 Part 1 🎄
13005 (elapsed: 6.00ns)
🎄 Part 2 🎄
11373 (elapsed: 831.00ns)

Hello! I was trying to solve the first exercise with your template, but if I run cargo build or cargo scaffold 01, the output is always the same:

error: failed to parse manifest at `/home/herz/Cargo.toml`

Caused by:
  no targets specified in the manifest
  either src/lib.rs, src/main.rs, a [lib] section, or [[bin]] section must be present

Thank you :)

It would be nice to have a flag for reading only part 2 of a puzzle.

If you accidentally run the cargo scaffold command without the --download argument or want to redo an entire problem, a manual deletion of the file is required before running the cargo scaffold <day> --download command again. It would be nice to have sort of a --overwrite argument that can overwrite the existing solution file

I see that year can be specified at the command line, but it doesn't seem like the code/inputs themselves are organized in any particular way to support different years.

Is there a way to accomplish catching up over a few years of missed AoC?

Hi, I am a beginner in rust and I have problems extracting common helper functions to a file.

For example let's say that I wanted to factor out a common transpose function that I repeatedly used in days 7, 11 and 13. With the template it seems I can't do that in the bin folder, right? 😕

If possible, it would be really nice if there was an empty file in the template already for this purpose.

Thanks for the great template! It helps a lot to get started quickly with using Rust for AoC.

I was wondering how to go about introducing modules to split out some of the code for a specific puzzle. E.g. for a puzzle where there is a lot of code for parsing the input into structs I would like to have a dedicated file for that day that handles this.

I tried to add a 03 folder with a input.rs but I wasn't able to include anything from it in my 03.rs. I guess it might be that modules may not have a numeric name. But it feels like a more general issue with the structure of the template.

Another case I could imagine is if I wanted to add some more generic functionality that could be reused by multiple puzzles. That I would like to put into a separate module as well. How would that work?

The docs say setting the --release flag should run the real puzzle inputs, and omitting the flag should run the examples.

The solve command runs your solution. If you set the --release flag, real puzzle inputs will be passed to your solution, otherwise the example inputs will be used.

As far as I can tell this is not the case. Running tests uses the example file, but any configuration of solve, with or without the release flag, will run the real puzzle inputs. I think one of these things would need to be updated, but first I would want to check which behavior is correct.

Is the expectation that solve should run the example or puzzle based on its input? Or does the documentation need to be updated that "solve" will always run puzzle input and "test" should be used for running examples?

I added dhat to my repo (which was generated from the template), so that I can run and get output like the following:

$ cargo solve 5 --dhat
    Finished dhat [optimized + debuginfo] target(s) in 0.01s
     Running `target/dhat/05`
dhat: Total:     23,920 bytes in 244 blocks
dhat: At t-gmax: 8,176 bytes in 12 blocks
dhat: At t-end:  0 bytes in 0 blocks
dhat: The data has been saved to dhat-heap.json, and is viewable with dhat/dh_view.html
Part 1: 382895070 (6.2ms)
dhat: Total:     32,528 bytes in 307 blocks
dhat: At t-gmax: 8,384 bytes in 11 blocks
dhat: At t-end:  0 bytes in 0 blocks
dhat: The data has been saved to dhat-heap.json, and is viewable with dhat/dh_view.html
Part 2: 17729182 (4.2ms)

Do you have any interest in getting this upstream? I'm not sure how many people would care to use it. I was mainly curious about optimizing allocations in various computations (e.g. using custom iterators instead of allocating vectors for temporary results) as a learning experience, so I found it useful to be able to see what's going on in the memory allocator.

In the light of solutions that may take a long time to run it would be great to have the option to time only one puzzle.
I.e.: allow cargo time 5 to only time day 5.

I guess the hardest bit would be to include the result in the benchmark table. Though it should be manageable.

Hey, thanks for this amazing template. It makes AoC even more fun!

As some kind of enhancement I would suggest to allow to have 2 separate files with example input for each part of the day’s problem. I encountered this problem in last year’s AoC a few times and in the very first puzzle of this year’s. Currently there can be only one set of example input data, which doesn’t necessarily work with both solutions and therefore Github CI throws error because not all tests are green.

I suggest to add publish = false in Cargo.toml to avoid accidental publishing

aoc-cli release v0.5.0 4 days ago and rename the --file command flag by --input-file.

It can be solved by either :

• install aoc-cli v0.4 with cargo install aoc-cli --version 0.4
• patch src/bin/download.rs with
  

  advent-of-code-rust/src/bin/download.rs

  Lines 66 to 72 in a6405f2

  	cmd_args.append(&mut vec![
  	"--file".into(),
  	tmp_file_path.to_string_lossy().to_string(),
  	"--day".into(),
  	args.day.to_string(),
  	"download".into(),
  	]);

    cmd_args.append(&mut vec![
-      "--file".into(),
+      "--input-file".into(),
        tmp_file_path.to_string_lossy().to_string(),
        "--day".into(),
        args.day.to_string(),
        "download".into(),
    ]);