use globs as RFC

[seanmonstar]

Allow aliasing imports when used in a glob import.

use std::io::{
    Error as IoError,
    Read,
    Write
}

Rendered

[sfackler]

👍

Nit: "glob" refers to use a::*, not use a::{b, c}.

[seanmonstar]

@sfackler I thought so, but the reference calls use a::{b, c} a path_glob.

[ftxqxd]

@seanmonstar Then I guess the reference is wrong, or at least slightly inaccurate with regard to naming. * imports were feature gated with the name glob, the term ‘glob’ usually refers to * (and similar) characters that get expanded in shells, and I’ve just never heard the term ‘glob’ used to describe the brace syntax in Rust.

I think the reference might be using the term ‘glob’ in a slightly more general sense, referring to any sort of expansion done by shells (both foo* and foo{a,b} are syntaxes used in shells like bash).

[petrochenkov]

+1
When I tried it I was surprised it didn't just worked.

[retep998]

👍
Are there any possible reasons for why we shouldn't allow this?

[jfager]

+1, been bumping into this a lot recently.

[pnkfelix]

So, to be clear, the RFC proposed is not allowing use foo::{self as quux, bar};

Is that intentional? (It seems reasonable to allow it, but also reasonable to delay implementing such an extension, if for some reason its more difficult in the resolve infrastructure.)

[seanmonstar]

@pnkfelix that's actually in the Unresolved Questions section. I can see it being useful, but wasn't sure the consensus.

[pnkfelix]

@seanmonstar ah, right you are, sorry for the noise

[reem]

+1, I have wanted this.

[pnkfelix]

Thank you very much for your contribution!

Clearly lots of people really like this idea. However, it is something we can add backwards compatibly after 1.0 ships (and it seems like we can continue getting by without it in the meantime).

So, we are going to postpone this RFC (see issue #933), but hopefully will address it sometime after 1.0 ships.

Thanks again

[seanmonstar]

@pnkfelix being postponed, would this mean accepted essentially? If I or someone else were to create a PR for 1.1, would that be accepted?

[nrc]

@seanmonstar no, it means that a decision has been postponed. Post-1.0, you could open a new RFC PR or re-open this one.

[seanmonstar]

@nrc You said i could re-open this one? I don't believe I have permission to do so (I don't see the button), but if I were to open an new PR, it would be the same commit exactly...

[nrc]

"re-open" means ask someone with the right privs to re-open for you, I guess (due to the GH security model, not Rust policy), but creating a new PR (like you've done) is totally fine.

[steveklabnik]

[nixpulvis]

Related #1400