I’ve been neglecting the updates, but here we go! Since the last update, I got a prototype out, but it was something that wasn’t on my list… I need to figure out if it counts as one of the 10 or not :-)
rust-bisect is a little tool to help track down changes in Rust nightlies. It
works a lot like
git bisect run. You give it a command, a nightly build where
it succeeds, and a nightly build where it fails, and it does binary search to
find the first nightly where it fails.
I’ve decided to give a name to the project formerly described as ‘rustfmt line
range & diff reading (format a patch)’:
cargo fmt-diff. It’ll be a command
you can run to format the lines you’ve changed. Ideally, you’ll be able to put
it in a VCS commit hook, and never introduce bad formatting. Think of it as a
style-oriented analogue of not rocket science:
The Not Rocket Science Rule Of Software Engineering:
automatically maintain a repository of code that always passes all the tests
A (very!) rough plan of how I intend to get this done:
- add command line flag that takes a set of line ranges to format (progress: rustfmt PR #844, ‘rustfmt: Add an ineffectual –file-lines flag behind a feature’)
- parse a unified diff to get sets of line ranges to pass to the above flag
If you’re especially interested in this feature, the rustfmt issue to track is #434.
containy-thing (name TBC)
I’m promoting a project I was working on over the holidays onto my list of 10: containy-thing. It was formerly described as ‘container thing for running commands in container context without privileges’, and I mentioned it in Eat your greens and read your man pages a month ago. The goal is to allow running commands in a container environment without root privileges. I just pushed up a basic repository for it.
Rust and The Update Framework
I’m also promoting a fairly ambitious project onto the list of ten. In the list of 100, it was cryptically described as ‘tuf + rust + crates’. The goal of this project is to secure the Rust crate ecosystem using The Update Framework (TUF). I’ve been interested in this project since well before I wrote any Rust. Verifiable provenance of software artifacts is extremely important, and Rust absolutely should have a good approach to it.
The Update Framework is a really well-designed and well-thought out approach to securing software updates and distribution. It’s being implemented for Python, Docker, and other ecosystems. While I was in New York attending the Recurse Center, I met up with the researchers behind TUF (minutes) to talk about implementing it for Rust, but I hadn’t followed up until recently. The issue to watch for this one is #75 on the Crates.io repository.
This project is really exciting, and more than a little bit scary because of how important it is to get right. But I won’t be working on it alone, so that’s great! The first steps are to refamiliarize myself with TUF, and then put together a Rust RFC. The TUF folks have recently had a paper accepted that describes how they have improved TUF even further, and I’m starting with reading that.
Summary of the list-of-ten:
My list of ten now contains four projects:
- cargo fmt-diff
- silly memcached or redis-speaking key value store whose nature of silliness is as yet undisclosed
- TUF and Crates.io