Most annoying thing about #rustdoc: when you look up the impl of a standard #RustLang trait on a particular concrete type, and it just repeats the documentation from the trait itself.
1. I wanted to know how the trait's functionality is interpreted for _this type_. E.g. don't just tell me `Add` implements the addition operator – tell me what addition _means_ for this particular type. Or if the trait docs say "you can do X or Y", I want to know which option this type chose, and maybe why.
2. Trait docs often include advice to the trait implementor, and in this context, that advice is going to the wrong person. E.g. the docs for PartialEq::ne() say it's a mistake to override the default implementation without a really good reason. This is useful advice if I'm implementing PartialEq on my own type. But when I look up the impl of PartialEq on someone else's type, _I_ don't need to be told what the crate author ought not to have done!
In fact rustdoc doesn't even tell me _whether_ the crate author _did_ override ne() (without providing a replacement doc comment), or whether they left the default in place. So it's not just nagging the wrong person – it's nagging the wrong person about a thing that very likely didn't even happen!
In principle, crate authors could fix #1 by being less lazy, and writing custom docs for their impls. But fixing #2 is really hard. If I want to delete that nagging comment about PartialEq::ne(), I have to do it by overriding the default implementation – exactly what the comment tells me not to do!