Next round of removing feature flags from libsyntax #24715
Conversation
|
r? @nrc (rust_highfive has picked a reviewer for you, use r? to override) |
| @@ -86,7 +86,31 @@ impl<'a> Iterator for LinkedPath<'a> { | |||
| } | |||
|
|
|||
| /// The type of the iterator used by with_path. | |||
| pub type PathElems<'a, 'b> = iter::Chain<iter::Cloned<slice::Iter<'a, PathElem>>, LinkedPath<'b>>; | |||
| #[derive(Clone)] | |||
| pub struct PathElems<'a, 'b> { | |||
There was a problem hiding this comment.
Oh dear the instability of Cloned is actually a mistake and it's intended to be #[stable] (in which case this change may not be required)
There was a problem hiding this comment.
I see it's now stable in master, so I'll remove this patch :)
…drop While this is slower than before, it's not necessarily forward compatible with planned changes to the drop flag. Benchmarking suggests that this does not dramatically impact the compile times of rust.
|
@alexcrichton: Updated to remove the patches. I think I could even remove the |
|
@kwantam: Moving the comments off the outdated commit. Yes, I was hoping that we could do what we did for On making rustc dependent on crates.io, I haven't heard of any plans going in that direction. I do think though that it's a little dangerous that we have these internal and external crates diverging from one another. We should probably shift over to using git submodules to pull those packages in to rust-proper. |
|
@erickt Regarding crates.io: I agree, it seems to make the bootstrapping problem a little gross.
|
For what it’s worth, the relevant part of the standard is Unicode Standard Annex #31: |
|
@erickt By the way, I saw a couple times a ping from you on IRC in what’s the middle of the night in my time zone, and by the time I saw it you were not on IRC anymore. Please remember http://www.nohello.com/ :) |
|
|
||
| // FIXME: This was copied from core/str/mod.rs because it is currently unstable. | ||
| #[inline] | ||
| pub fn slice_shift_char(s: &str) -> Option<(char, &str)> { |
There was a problem hiding this comment.
It looks like this function is actually pretty rarely used (only used twice here), could this just be removed entirely and each call site just have a custom implementation? They don't really need slice_shift_char in its full glory but rather just some smaller specialized versions.
|
Thanks @erickt! I'm fine with transitioning away from |
|
@SimonSapin: @kwantam helped me out. I wasn't sure which of you two were in charge of unicode-rs, so I pinged you both :) I'm good now though, but thanks for checking in! @alexcrichton: I'll look into simplifying |
|
☔ The latest upstream changes (presumably #24967) made this pull request unmergeable. Please resolve the merge conflicts. |
|
ping @erickt, is this still necessary? |
|
@alexcrichton: nope! |
This is part two of #24518, which is trying to get libsyntax to compile without feature flags. After this, all that remains is
#[feature(collections, libc, rustc_private, staged_api)]. The one real unfortunate thing is that I had to copy some tables fromlibrustc_unicode, but I couldn't figure out a great way to share these tables between the two modules.