Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> Some areas of the standard library still aren't great; I'm working on it.

Oh, I don't doubt it. I'm not even saying that the Rust's documentation is bad. I only object to the claim that it's one of the best (because I've used better ones). It simply needs more work to reach that state, as you seem to be perfectly aware.

> [...] So boom, wrap it in a BufReader, call lines, then call next. Done.

From my perspective, this is a non-obvious process. I couldn't find the solution on my own, and I'm not a newbie in reading documentation, as I have a habit of reading reference docs, which historically included learning most of Perl (which is not known for being simple) and, more recently, much of Erlang (which has documentation that needs proficiency in reading it).

To give some constructive feedback, put some thought to how can one discover from the documentation all the traits applicable to the situation (possibly avoiding using the search box, which I usually don't expect to provide precise results). If there was some connection from TcpStream to BufReader, I wouldn't have any problem with finding the latter.



Totally! std::net is an area I haven't touched at all. Whatever's there is the accumulation of whoever decided to write some things, no systematic work was really done to document it yet.


I actually sort of expected the documentation to be generated, not completely hand-written (mainly because I'm a fan of JavaDoc descendants, if they are used correctly). If this was the case, then the doc build tool would need to extract and present some additional information about the applicable traits.


The type signatures are generated, but the text and examples come from extracting documentation comments.


So I was right, it's a generated thing (obviously, a tool cannot provide the description nor usage examples).




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: