feat: Discover current repository

[spencerwilson]

• Use git2's Repository.discover to locate the current git repo root, if one exists

git2 is a crate of Rust bindings to libgit2. The git book suggests it's the dominant way to "embed" git-like functionality in apps w/o shelling out to the git binary.

Cargo.lock changes are the result of running cargo build on this branch, w/ cargo version 1.39.0.

Manual tests:

# State of this branch:
~/code/git-codeowners/src
$ ../target/debug/git-codeowners src/main.rs
@softprops

# When temporarily deleting CODEOWNERS
~/code/git-codeowners/src
$ ../target/debug/git-codeowners src/main.rs
No CODEOWNERS file found in this repo.
Ensure one exists at any of the locations documented here:
https://help.github.com/en/github/creating-cloning-and-archiving-repositories/about-code-owners#codeowners-file-location

# When outside of a git repo
~/code
$ ./git-codeowners/target/debug/git-codeowners src/main.rs
Repo discovery failed. Is /Users/ssw/code within a git repository? Error: could not find repository from '/Users/spencer/code'; class=Repository (6); code=NotFound (-3)

Resolves #3.

[softprops]

If possible could we avoid unwrapping?

[spencerwilson]

Sure thing. Docs say that current_dir() could be Err if (perhaps not comprehensive?)

• Current directory does not exist.
• There are insufficient permissions to access the current directory.

Perhaps desired behavior could be printing something like "Could not confirm working directory. Does it still exist? Is it readable?", then exit 2/some not-yet-taken number. What do you think?

[softprops]

I wonder if we should just use std::process::Command with "git rev-parse --show-toplevel" since there's not much heavy git lifting needed here.

How much bigger did adding a crate to perform this a action add?

[spencerwilson]

Here are some numbers I get:

master branch

# Compilation time
$ rm -rf target/ && time cargo build --release
...
    Finished release [optimized] target(s) in 1m 12s

real	1m12.546s
user	3m13.753s
sys	0m3.983s

# Binary size
$ du -h target/release/git-codeowners
1.7M	target/release/git-codeowners

PR branch

$ rm -rf target/ && time cargo build --release
...
    Finished release [optimized] target(s) in 1m 58s

real	1m49.198s
user	6m23.143s
sys	0m19.478s

$ du -h target/release/git-codeowners
2.0M	target/release/git-codeowners

Thanks for encouraging me to measure this. I expect that whatever performance benefit to be had from leveraging git2 is likely not worth doubling the CPU time required to compile. But I'll try using std::process::Command as you described and see what perf difference I can measure.