The categories listed on crates.io are a good means of classifying crates – if you know what you are looking for – but I got the idea of a Community Wiki as a guide to the landscape of everyday crates in the Rust ecosystem.
Do you think the need or benefit is there, and the liveliness of the community to maintain such a wiki?
The Community Wiki could be a place for sharing tips, code and cross-crate commentaries. Perhaps starting with the various Working Groups, crates central to their domains could be introduced and presented. For some domains, perhaps a process-oriented (what-do-you-want-to-do) exposition could be used. Crates-io today has a list of categories but what I would like to see, I guess, is a more human touch to it. Other things that could be of interest on this wiki:
Directory of crates suitable for no-std/embedded
Security best-practices and recipes
FFI handbook
“Super groups”, or topics of crate categories.
Rust-by-example-style tips for the crates of the community, open for anyone to contribute to.
In part, but a wiki doesn’t require approved PRs for additions or changes to be published. I want a more inclusive form.
Perhaps it’s also still too much of a generated “list” and not enough of a living document with links and sections?
It would also be useful to integrate this information with crates.io as this is where crates are found. I think crates-io is the natural centrepoint of the community, where everybody goes, and the place where the biggest impact is possible. I might be wrong about that.
I think curation outside of crates-io is a good idea.
On crates.rs I’ve already created three new categories that don’t exist on crates-io: Math, Machine Learning, and Cryptocurrencies. IMHO categories work well when they have 50-100 crates, so I’m happy to add more and especially split the existing large ones.
I am considering adding even smaller categories for specific tasks, e.g. error handling.
There are a few sites “Are We * Yet” for web frameworks, machine learning. These are pretty cool IMHO.
The difficulties I see is in keeping them up to date long term. When I look at Awesome Rust list, I don’t know if there are still awesome, or something else has come since. These list tend to add stuff easily, but remove only when the crates obviously rot.
Crates.rs looks very good! More categories and/or filters could reduce the list lengths to something manageable.
What to do about crate rot? Perhaps a default sorting algorithm could be useful within each category, taking various properties into account: recent downloads, does the crate build?, last update...
Another, somewhat related idea, I had was linking/adding a public comment section to crate pages, a bit like the User Contributed Notes on the PHP website. But perhaps that would be more of a docs.rs thing. I guess my main problem is that I feel that the information is too dispersed: crates-io/rs, docs.rs, cookbook, awesome list, wg pages... Maybe that's just me feeling a bit lost.
I think this is a very good idea. I made a similar suggestion in the Security WG proposal, regarding crates under the cryptography category. Though this more meant as a officially endorsed list, where listed crates had been reviewed, I think a community effort like this one could work in conjunction with the Security WG if they decide to implement my suggestion.
Or a community-wiki could be a starting point of something the Security WG could base their list off of, taking crates listed there and review further.
