RFC: Resolve ::foo::... when compiling the crate named foo

[kmcallister]

No description provided.

[alexcrichton]

This seems like it may be the wrong solution for a very real problem. You'll notice that all of the facade crates have an inner mod std module for the same reason the standard library does, but this solution would only affect the standard library, not any of the other crates.

This is exactly what macro hygiene is supposed to solve, and pasting another resolve solution on top may not be the best approach for tackling it in the interim.

[kmcallister]

Hm, good point. The facade pattern seems pretty unusual, though. Do you expect that users of Rust, as opposed to Rust itself, will encounter similar situations?

[alexcrichton]

I would imagine that requiring hygiene will show up quite frequently for all users of rust using macros in one form or another, but the specific facade use case is quite rare because the compiler is the one with the deriving implementations, not a library.

That said, all users of libcore and friends without using libstd suffer from this problem as well, and likely have their own workarounds to fix it.

[wycats]

This problem occurs if you define a macro for use both inside a crate and to import into other crates that needs to use a define defined in the crate.

[alexcrichton]

This was discussed at the triage meeting today. While the motivation is definitely valid, we feel that this RFC specifically may not be the best solution to the problem. There are a few surprises such as not all crate names are valid rust identifiers, and it may be desired to extend this to modules at some point in the future as well.

For now, I'm going to close this, and not mark it as postponed. This is a very real problem, however, and it would be very nice to have a hygiene-based solution!