Let's try this: Dear #lazyweb / #lazyfedi, what is the current go-to if I have to/want to write application #documentation for #endusers?
It is a #rust application, that does contain a lot of #rustdoc already, but that is very specific to the code, like api/developers doc. It is not end user stuff.

So, what to do? Own branch and something in it, that then gets rendered on a Github-or-similar page? A .md file used with #mkdocs? Using #readthedocs and their #sphinx based setup (also their examples are pretty python related).

I would like to stay as near to the code as I can. *Ideally* it would be rustdoc comments, but that can't be, as that is used for the developers doc already. "Just" a file besides, maybe in doc/ sounds better than whole new branch?

Any #suggestions?

@shapr That is nice! Couldn't we implement this to search within a crate in #rustdoc? @imperio

We would need to ignore lifetimes for simplification.

Oh I have been so blind. Really love #Rust's #rustdoc documentation generation, and even often can tell how things are supposed to work based on the type defs in there.

But this one.. oh so blind. Was working on replacing some internal, rather complex, buffer in a project with the `bytes::BytesMut` type, and spent hours trying to figure out how to peek the first byte without consuming it.

And finally realized it derefs to a slice so you can just treat it as regular slice as well 🤦‍♂️

@JoshStrobl

Have you looked at some of the language-specific documentation tools for inspiration? I've heard good things about the documentation tooling in #rust and #golang, for example. (I know that #rustdoc supports #markdown and inline testing of code within docs, which seems like it'd be very helpful for API documentation.)

@QuietMisdreavus

>rustdoc is an awkward static site generator

Correct. As evidence, see this awkward static site generated with #rustdoc: https://newrustacean.com/show_notes/

(But the podcast attached to the awkward site is great)

Some of the most interesting changes in #rustlang at the moment are in #rustdoc. How did I not hear of compile_fail in doctests until it was released in stable?