wg VS rust-semverver

Inspired by James Munns's call, and as 2023 is coming to an end, I figure it's a good opportunity to reflect and look forward to 2024. It's been a bit over 1.5 years since I embarked on my embedded Rust journey and it's been nothing less than exciting since. So here it goes.

Because you are an embedded guy. There is the https://github.com/rust-embedded/wg working-group. Rust on embedded is really on a got track forward. There are many chips/vendors that are supported both in no std / std rust world, but still there is a lot of niche things where you can actively help to be the first to get it run in Rust.

To specifically answer your question, here:

* <http://www.areweembeddedyet.com/>

It currently redirects to:

* <https://rust-embedded.org>

Which doesn't really contain anything other than a link to <https://github.com/rust-embedded>.

(via <https://github.com/rust-embedded/wg/issues/15>)

Then the Rust Embedded workgroup provides: - Direction on how to using generics and zero-sized types to achieve functional safety - svd2rust, which provides safe abstractions to peripheral access from SVD files and achieves this functional safety - The embedded HAL spec, which makes porting to different vendors/hardware easy - Peripheral access controllers and HALs for various vendors & hardware

The Rust Embedded Devices Working Group curates a list of useful embedded Rust resources, including Peripheral Access Crates (autogenerated from SVD files), embedded-hal Implementation Crates (hand-written libraries implementing the traits (interfaces) specified by the embedded-hal), and Board Support Crates.

You should also look at https://github.com/rust-embedded/wg/issues/41 and https://github.com/rust-lang/rust/pull/55011#issuecomment-429336055.

Here's an interesting discussion, consolidated here. My view is you should use a restricted scope atomic (as best as can be supported) and interact with that through a handler struct. I.e. no global state.

Most of the issues are explained in EWG RFC 419. The TL;DR is that some resources need to implement Send to be usable from interrupts, but they must not be sent across cores.

1) No. I think semver is just fine for its intended purpose. I mean, I'm sure its spec could be improved in various ways, but its fundamental idea seems fine to me. I think it's just important to remember that semver is a means to an end, and not an end itself. It is a tool of communication most useful in a decentralized context.

2) No.

3) See: https://github.com/rust-lang/rust-semverver --- But also, this is only ever going to be a "best effort" sort of thing. Semver isn't just about method additions or deletions, but also behavior.

How does this compare to cargo-semverver?

Found this: https://github.com/rust-lang/rust-semverver but it doesn't seem to act on git log/diffs... just FYI.

Enter rust-semverver!

Rust type system on the other hand does not allow this. Traits are monotonic logic: adding trait-impls / most qualifiers does't influence already existing and compiling code (note: for code that doesn't rely on disambiguation to compile). There's rules that clarify this disambiguations and breaking/non-breaking changes according to the type system. There's SemVerVer to automatically verify those guidelines.

In my case it wasn't so bad and an easy fix was fast to write, but it got me thinking of how much of a problem this is for the wider ecosystem. Some searching showed me, that of course there is a tool rust-semverver to do exactly that. Sadly it errors on my system (or maybe im just using it wrong). Would've been interesting to see how often this actually happens on crates.io and how much of a problem this really is.

There's been an attempt at this for the Rust ecosystem: https://stackoverflow.com/questions/41185023/what-exactly-is...

There's also a library that attempts to automatically check sermver adherence of a crate: https://github.com/rust-lang/rust-semverver

And there has been quite a bit of effort into preventing semver requirements from fracturing the ecosystem. This revolves around the compiler working with multiple major versions of a single library: https://github.com/dtolnay/semver-trick

Turns out it already exists: https://github.com/rust-lang/rust-semverver