Where should we put macros?

[ghost]

Consider the following macros:

• async_std::task_local
• async_std::println
• async_std::task::ready

They are not organized in a very sensible way. Why would task_local be in the root module and not in the task module? And why is println not in the io module if standard I/O functionality lives there?

In the Rust standard library, all macros live in the root module. This is perhaps a historical artifact because macros couldn't be placed anywhere else until Rust 2018 (remember the age of #[macro_use]?). If the standard library was designed today, perhaps we'd have std::io::println instead of std::println.

So we're facing a dilemma: do we organize our macros in async_std more logically (unlike std) or just dump them all in the root module (like std)? Or, do we keep the current situation and have equivalents of macros from std in the root module and place other macros elsewhere?

Macros in the root module are also re-exported in the prelude (just like in std). However, other macros aren't.

Opinions?

[yoshuawuyts]

@stjepang I'm thinking we re-export the std counterparts as part of the root of the crate. But then we also export the individual macros from their respective submodules.

Probably the odd macro odd out here is write / writeln which makes sense to have both part of io and fmt. We don't have a fmt crate currently so it's not a problem, but yeah still.

Related PR: #328