Implement RFC 2052 (Epochs) #48014

Manishearth · 2018-02-05T14:25:23Z

This adds -Zepochs and uses it for tyvar_behind_raw_pointer (#46906)

When we move this to --epoch=XXX, we'll need to gate the 2018 epoch on nightly, but not the 2015 one. I can make these changes here itself though it's kinda pointless given that the entire flag is nightly-only.

r? @nikomatsakis @aturon

cc #44581 (epoch tracking)
cc #46906 (tyvar_behind_raw_pointer)

kennytm · 2018-02-05T14:29:09Z

src/librustc/session/config.rs

@@ -1297,6 +1325,13 @@ options! {DebuggingOptions, DebuggingSetter, basic_debugging_options,
        `everybody_loops` (all function bodies replaced with `loop {}`),
        `hir` (the HIR), `hir,identified`, or
        `hir,typed` (HIR with types for each node)."),
+    epoch: Epoch = (Epoch::Epoch2015, parse_epoch, [TRACKED],
+        "Present the input source, unstable (and less-pretty) variants;
+        valid types are any of the types for `--pretty`, as well as:


This comment is wrong?

kennytm · 2018-02-05T14:30:54Z

src/librustc/session/config.rs

+#[derive(Clone, Copy, Hash, PartialOrd, Ord, Eq, PartialEq)]
+/// The epoch of the compiler (RFC 2052)
+///
+/// Pronounced e-poach


😜

(i can remove it if you want)

😄

(it's fine)

kennytm · 2018-02-05T14:33:38Z

src/librustc/session/config.rs

@@ -112,6 +112,23 @@ pub enum OutputType {
    DepInfo,
 }

+#[derive(Clone, Copy, Hash, PartialOrd, Ord, Eq, PartialEq)]


Nit-pick: We usually write the doc comments before the attributes, right? i.e.

/// The epoch of the compiler (RFC 2052) #[derive(Stuff)] pub enum Epoch {}

kennytm · 2018-02-05T14:35:20Z

src/librustc_typeck/diagnostics.rs

+
+For example, you may have done something like:
+
+```


This makes me wonder if it is possible to tell the doc comment which epoch it should use.

not really. there were plans for making the extended messages dynamic, but they didn't really take off.

@Manishearth I mean something like

```rust,epoch2018 let foo: Box<dyn Debug> = box 5; ``` ```rust,epoch2015 let foo: Box<Debug> = Box::new(5); ```

Hmmm.. Would be nice to have. Out of scope for this PR though 😄

Anyway, this specific warning only occurs on Rust 2018+, and the extended error stays the same.

Manishearth · 2018-02-05T14:54:27Z

addressed comments

steveklabnik · 2018-02-05T15:30:21Z

src/librustc/session/config.rs

+    Epoch2015 = 2015,
+    /// The 2018 epoch
+    Epoch2018 = 2018,
+


should this enum be non-exhasitve?

that's a little, uh, existential of a question I guess...

Yeah, this makes sense.

steveklabnik · 2018-02-05T15:31:28Z

src/librustc/session/config.rs

+#[derive(Clone, Copy, Hash, PartialOrd, Ord, Eq, PartialEq)]
+pub enum Epoch {
+    /// The 2015 epoch
+    Epoch2015 = 2015,


it's kind of a shame that we end up with stutter here.... hm.

Manishearth · 2018-02-05T16:32:25Z

cargo PR at rust-lang/cargo#5011

still need to do tests for both

mgattozzi · 2018-02-05T17:26:17Z

src/librustc/session/config.rs

+    epoch: Epoch = (Epoch::Epoch2015, parse_epoch, [TRACKED],
+        "The epoch to build Rust with. Newer epochs may include features
+         that require breaking changes. The default epoch is 2015 (the first
+         epoch). Crates compiled with different epoches can be linked together."),


epoches -> epochs

nikomatsakis · 2018-02-05T17:28:38Z

src/librustc/session/config.rs

+    // - add a `rust_####()` function to the session
+
+    #[doc(hidden)]
+    __InexhaustiveMatch,


can't we use the #[non_exhaustive] attribute now?

nikomatsakis

Thanks @Manishearth! Left a few comments.

nikomatsakis · 2018-02-05T17:59:30Z

src/librustc/session/config.rs

+#[derive(Clone, Copy, Hash, PartialOrd, Ord, Eq, PartialEq)]
+pub enum Epoch {
+    /// The 2015 epoch
+    Epoch2015 = 2015,


I would rather we did not specify the discriminants. For one thing, we may still want epochs with names like 2015-pre.

nikomatsakis · 2018-02-05T18:00:36Z

src/librustc/session/config.rs

@@ -112,6 +112,26 @@ pub enum OutputType {
    DepInfo,
 }

+/// The epoch of the compiler (RFC 2052)
+///
+/// Pronounced e-poach


I would rather we remove "E-poach" -- I like a funny comment as much as the next person, but I worry that people will actually believe us! (And this is part of rustc's public API)

More importantly, we should comment that the ordering of this enum is significant, and that epochs must go from "early to late" (since there is code that tests e.g., >= Epoch2018).

nikomatsakis · 2018-02-05T18:10:59Z

src/librustc/session/mod.rs

@@ -864,6 +864,11 @@ impl Session {
    pub fn teach(&self, code: &DiagnosticId) -> bool {
        self.opts.debugging_opts.teach && !self.parse_sess.span_diagnostic.code_emitted(code)
    }
+
+    /// Are we allowed to use features from the Rust 2018 epoch?
+    pub fn rust_2018(&self) -> bool {


I think we should prepare for macro compatibility here, and make this some kind of test that takes a span. I'm not sure where the method should go -- maybe put it on the tcx for now? That seems like it would definitely have access to whatever data it needs to make the determination.

I don't think the tcx makes sense; the idea behind epoches was that most of the changes would be to stuff like parsing and all, the "language frontend". We don't have a tcx when parsing. I think we should keep this unresolved for now and when someone knowledgeable about macro stuff comes along they can look into this.

I think the Session already has a lot of the Span info that can be used here; at least "originates from macro from X crate" can be done from the Session, though the interaction of many macros may be confusing.

I don't think the tcx makes sense

In the future, we plan to move so that the tcx is the first thing that is created, so that everything can be query driven. This is why I suggested tcx. However, session is also fine.

I think we should keep this unresolved for now and when someone knowledgeable about macro stuff comes along they can look into this.

Mm, I'm not sure. I think it'll be useful to require a span so that, for any change, we are at least thinking about how to deal with it.

--

Eh. Having thought about it some more, while I'd rather lay out a "forwards compatible" API with what we will likely eventually need, I don't care enough to block on it. And there's always an argument for doing the simplest possible thing to start.

kennytm · 2018-02-05T18:18:22Z

src/librustc_typeck/diagnostics.rs

+For example, you may have done something like:
+
+```
+let bar = foo as *const _;


The examples needs to be runnable.

[01:24:14] failures: [01:24:14] [01:24:14] ---- /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md - Rust_Compiler_Error_Index::E0908 (line 11384) stdout ---- [01:24:14] error[E0425]: cannot find value `foo` in this scope [01:24:14] --> /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md:11385:11 [01:24:14] | [01:24:14] 3 | let bar = foo as *const _; [01:24:14] | ^^^ not found in this scope [01:24:14] [01:24:14] thread 'rustc' panicked at 'couldn't compile the test', librustdoc/test.rs:298:13 [01:24:14] note: Run with `RUST_BACKTRACE=1` for a backtrace. [01:24:14] [01:24:14] ---- /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md - Rust_Compiler_Error_Index::E0908 (line 11395) stdout ---- [01:24:14] error[E0425]: cannot find value `foo` in this scope [01:24:14] --> /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md:11396:11 [01:24:14] | [01:24:14] 3 | let bar = foo as *const u8; [01:24:14] | ^^^ not found in this scope [01:24:14] [01:24:14] thread 'rustc' panicked at 'couldn't compile the test', librustdoc/test.rs:298:13 [01:24:14] [01:24:14] ---- /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md - Rust_Compiler_Error_Index::E0908 (line 11408) stdout ---- [01:24:14] error[E0412]: cannot find type `Foo` in this scope [01:24:14] --> /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md:11409:6 [01:24:14] | [01:24:14] 3 | impl Foo { [01:24:14] | ^^^ not found in this scope [01:24:14] [01:24:14] error[E0308]: mismatched types [01:24:14] --> /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md:11410:43 [01:24:14] | [01:24:14] 4 | fn is_null(self: *const Self) -> bool { [01:24:14] | ___________________________________________^ [01:24:14] 5 | | // do something else [01:24:14] 6 | | } [01:24:14] | |_____^ expected bool, found () [01:24:14] | [01:24:14] = note: expected type `bool` [01:24:14] found type `()` [01:24:14] [01:24:14] thread 'rustc' panicked at 'couldn't compile the test', librustdoc/test.rs:298:13 [01:24:14] [01:24:14] [01:24:14] failures: [01:24:14] /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md - Rust_Compiler_Error_Index::E0908 (line 11384) [01:24:14] /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md - Rust_Compiler_Error_Index::E0908 (line 11395) [01:24:14] /checkout/obj/build/x86_64-unknown-linux-gnu/test/error-index.md - Rust_Compiler_Error_Index::E0908 (line 11408) [01:24:14] [01:24:14] test result: FAILED. 666 passed; 3 failed; 21 ignored; 0 measured; 0 filtered out

Manishearth · 2018-02-06T03:06:37Z

Addressed

@acrichto

… r=alexcrichton Implement RFC 2052: Epoches Todo: - Make epoches affect the fingerprint - Tests cc rust-lang/rust#44581 Rust PR: rust-lang/rust#48014 r? @acrichto

nikomatsakis · 2018-02-06T18:40:15Z

Travis failure: https://travis-ci.org/rust-lang/rust/builds/337848627#L3684

nikomatsakis · 2018-02-06T18:40:58Z

r=me when travis is happy

Manishearth · 2018-02-06T22:07:49Z

@bors r=nikomatsakis

bors · 2018-02-06T22:07:50Z

📌 Commit b8aa8ca has been approved by nikomatsakis

@nikomatsakis

…e-internet, r=nikomatsakis Implement RFC 2052 (Epochs) This adds -Zepochs and uses it for tyvar_behind_raw_pointer (rust-lang#46906) When we move this to --epoch=XXX, we'll need to gate the 2018 epoch on nightly, but not the 2015 one. I can make these changes here itself though it's kinda pointless given that the entire flag is nightly-only. r? @nikomatsakis @aturon cc rust-lang#44581 (epoch tracking) cc rust-lang#46906 (tyvar_behind_raw_pointer)

Rollup of 10 pull requests - Successful merges: #47613, #47631, #47810, #47883, #47922, #47944, #48014, #48018, #48020, #48028 - Failed merges:

rust-highfive assigned nikomatsakis Feb 5, 2018

kennytm added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Feb 5, 2018

kennytm reviewed Feb 5, 2018

View reviewed changes

Manishearth force-pushed the stealing-chickens-on-the-internet branch from 8ef92f7 to 5f20d45 Compare February 5, 2018 14:54

steveklabnik reviewed Feb 5, 2018

View reviewed changes

Manishearth force-pushed the stealing-chickens-on-the-internet branch from 5f20d45 to 23fb447 Compare February 5, 2018 15:34

Manishearth mentioned this pull request Feb 5, 2018

Implement RFC 2052: Epoches rust-lang/cargo#5011

Merged

mgattozzi reviewed Feb 5, 2018

View reviewed changes

nikomatsakis reviewed Feb 5, 2018

View reviewed changes

nikomatsakis requested changes Feb 5, 2018

View reviewed changes

kennytm added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 5, 2018

kennytm reviewed Feb 5, 2018

View reviewed changes

Manishearth force-pushed the stealing-chickens-on-the-internet branch 3 times, most recently from 11eb29a to aff1216 Compare February 6, 2018 03:03

Manishearth added 2 commits February 5, 2018 22:16

Add -Zepoch

2cff123

Convert tyvar_behind_raw_pointer to hard error for the 2018 epoch

d0ab8f0

Manishearth force-pushed the stealing-chickens-on-the-internet branch from aff1216 to c69b599 Compare February 6, 2018 03:16

nikomatsakis approved these changes Feb 6, 2018

View reviewed changes

Manishearth force-pushed the stealing-chickens-on-the-internet branch 2 times, most recently from ddd80e3 to 19a32bd Compare February 6, 2018 19:46

Fill in long diagnostic

83a9f49

Add tests for -Zepoch using tyvar_raw_pointer

b8aa8ca

Manishearth force-pushed the stealing-chickens-on-the-internet branch from 19a32bd to b8aa8ca Compare February 6, 2018 19:46

kennytm added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Feb 6, 2018

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 6, 2018

Manishearth mentioned this pull request Feb 7, 2018

Rollup of 10 pull requests #48053

Merged

bors added a commit that referenced this pull request Feb 7, 2018

Auto merge of #48053 - Manishearth:rollup, r=Manishearth

29c8276

Rollup of 10 pull requests - Successful merges: #47613, #47631, #47810, #47883, #47922, #47944, #48014, #48018, #48020, #48028 - Failed merges:

bors merged commit b8aa8ca into rust-lang:master Feb 7, 2018

aturon mentioned this pull request Feb 22, 2018

Tracking issue for RFC 2052: Epochs #44581

Closed

3 tasks

dtolnay mentioned this pull request Jul 29, 2018

Attribute to set edition: #![edition = "2018"] #52817

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Implement RFC 2052 (Epochs) #48014

Implement RFC 2052 (Epochs) #48014

Manishearth commented Feb 5, 2018

kennytm Feb 5, 2018

kennytm Feb 5, 2018

Manishearth Feb 5, 2018

kennytm Feb 5, 2018

kennytm Feb 5, 2018

kennytm Feb 5, 2018

Manishearth Feb 5, 2018

kennytm Feb 5, 2018

Manishearth Feb 5, 2018

Manishearth commented Feb 5, 2018

steveklabnik Feb 5, 2018

Manishearth Feb 5, 2018

steveklabnik Feb 5, 2018

Manishearth commented Feb 5, 2018

mgattozzi Feb 5, 2018

nikomatsakis Feb 5, 2018

nikomatsakis left a comment

nikomatsakis Feb 5, 2018

nikomatsakis Feb 5, 2018

nikomatsakis Feb 5, 2018

Manishearth Feb 6, 2018

nikomatsakis Feb 6, 2018

kennytm Feb 5, 2018

Manishearth commented Feb 6, 2018

nikomatsakis commented Feb 6, 2018

nikomatsakis commented Feb 6, 2018

Manishearth commented Feb 6, 2018

bors commented Feb 6, 2018

Implement RFC 2052 (Epochs) #48014

Implement RFC 2052 (Epochs) #48014

Conversation

Manishearth commented Feb 5, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Manishearth commented Feb 5, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Manishearth commented Feb 5, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nikomatsakis left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Manishearth commented Feb 6, 2018

nikomatsakis commented Feb 6, 2018

nikomatsakis commented Feb 6, 2018

Manishearth commented Feb 6, 2018

bors commented Feb 6, 2018