Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

where clause in generic associated types causes no output #47206

Closed
sunjay opened this issue Jan 5, 2018 · 5 comments · Fixed by #67160
Closed

where clause in generic associated types causes no output #47206

sunjay opened this issue Jan 5, 2018 · 5 comments · Fixed by #67160
Labels
A-associated-items Area: Associated items such as associated types and consts. C-bug Category: This is a bug. F-generic_associated_types `#![feature(generic_associated_types)]` a.k.a. GATs P-medium Medium priority requires-nightly This issue requires a nightly compiler in some way. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue.

Comments

@sunjay
Copy link
Member

sunjay commented Jan 5, 2018

I'm working on implementing generic associated types in the compiler with mentoring by @nikomatsakis. He recently discovered an interesting issue that we're haven't found the source of yet. It seems that having a where clause like the one below results in a compilation error but no output.

The feature isn't actually fully finished yet, so we expect a compilation error. It just isn't clear why the compiler isn't actually reporting any specific problems. The message says that it failed due to a previous error but doesn't list any errors at all.

Here's the code that reproduces the issue: (Playground link)

#![feature(generic_associated_types)]

trait Foo {
    type Assoc3<T>;
}

struct Bar;

impl Foo for Bar {
    type Assoc3<T> where T: Iterator = Vec<T>;
}

fn main() {}

Here's the output:

   Compiling playground v0.0.1 (file:///playground)
error: cannot continue compilation due to previous error

error: Could not compile `playground`.

To learn more, run the command again with --verbose.

This is on the nightly compiler because the feature is gated while we're still working on it. If you try to run this on stable or beta, you get a normal error message.

If you delete the where T: Iterator you get an ICE which I think is expected given what I still have left to implement for the generic associated types feature.

It might be that this gets fixed while we keep implementing generic associated types. Creating this issue will make sure that this doesn't get forgotten along the way.

@pietroalbini pietroalbini added A-associated-items Area: Associated items such as associated types and consts. C-bug Category: This is a bug. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Feb 6, 2018
@nikomatsakis
Copy link
Contributor

cc @rust-lang/compiler — this is the source of that recent regression. Anybody care to investigate?

kennytm added a commit to kennytm/rust that referenced this issue Aug 11, 2018
Feature gate where clauses on associated type impls

Fixes rust-lang#52913. This doesn't address the core problem, which is tracked by rust-lang#47206. However, it fixes the stable-to-stable regression: you now have to enable `#![feature(generic_associated_types)]` to trigger the weird behaviour.
kennytm added a commit to kennytm/rust that referenced this issue Aug 13, 2018
Feature gate where clauses on associated type impls

Fixes rust-lang#52913. This doesn't address the core problem, which is tracked by rust-lang#47206. However, it fixes the stable-to-stable regression: you now have to enable `#![feature(generic_associated_types)]` to trigger the weird behaviour.
Mark-Simulacrum added a commit to Mark-Simulacrum/rust that referenced this issue Aug 22, 2018
Feature gate where clauses on associated type impls

Fixes rust-lang#52913. This doesn't address the core problem, which is tracked by rust-lang#47206. However, it fixes the stable-to-stable regression: you now have to enable `#![feature(generic_associated_types)]` to trigger the weird behaviour.
bors added a commit that referenced this issue Aug 23, 2018
Feature gate where clauses on associated type impls

Fixes #52913. This doesn't address the core problem, which is tracked by #47206. However, it fixes the stable-to-stable regression: you now have to enable `#![feature(generic_associated_types)]` to trigger the weird behaviour.
@nikomatsakis
Copy link
Contributor

Huh I'm still not sure what the heck is going on here. I started doing a bit of digging. I found that we seem to be dying sometime in astconv for .. some reason I have not yet determined. =)

@nikomatsakis
Copy link
Contributor

OK, rust seems to panic. I don't know why this doesn't yield any output yet. I intercepted it in gdb:

Breakpoint 1, rust_panic () at libstd/panicking.rs:525
525             __rust_start_panic(obj as usize)
Missing separate debuginfos, use: dnf debuginfo-install libstdc++-8.1.1-1.fc28.x86_64
(gdb) up
#1  0x00007ffff763037c in std::panicking::update_count_then_panic () at libstd/panicking.rs:516
516         rust_panic(&mut RewrapBox(msg))
(gdb)
#2  0x00007ffff7630996 in std::panic::resume_unwind () at libstd/panic.rs:425
425         panicking::update_count_then_panic(payload)
(gdb)
#3  0x00007ffff0bf1032 in rustc_errors::FatalError::raise () at librustc_errors/lib.rs:238
238             panic::resume_unwind(Box::new(FatalErrorMarker))
(gdb)
#4  0x00007ffff5b8ad64 in <(dyn rustc_typeck::astconv::AstConv<'gcx, 'tcx> + 'o)>::trait_def_id () at librustc_typeck/astconv.rs:697
697                     FatalError.raise();
(gdb)
#5  <(dyn rustc_typeck::astconv::AstConv<'gcx, 'tcx> + 'o)>::instantiate_poly_trait_ref_inner () at librustc_typeck/astconv.rs:711
711             let trait_def_id = self.trait_def_id(trait_ref);
(gdb)
#6  0x00007ffff5b8adf8 in <(dyn rustc_typeck::astconv::AstConv<'gcx, 'tcx> + 'o)>::instantiate_poly_trait_ref ()
    at librustc_typeck/astconv.rs:744
744             self.instantiate_poly_trait_ref_inner(&poly_trait_ref.trait_ref, self_ty,
(gdb)
#7  0x00007ffff5bc4918 in rustc_typeck::collect::explicit_predicates_of () at librustc_typeck/collect.rs:1808
1808                                let trait_ref = AstConv::instantiate_poly_trait_ref(
(gdb)
#8  0x00007ffff2b1ef18 in rustc::ty::query::<impl rustc::ty::query::config::QueryAccessors<'tcx> for rustc::ty::query::queries::explicit_predicates_of<'tcx>>::compute::{{closure}} () at librustc/ty/query/plumbing.rs:822
822                         provider(tcx.global_tcx(), key)
(gdb)
#9  rustc::ty::query::__query_compute::explicit_predicates_of () at librustc/ty/query/plumbing.rs:790
790                     f()
(gdb)

#10 0x00007ffff2cc91a5 in rustc::ty::query::<impl rustc::ty::query::config::QueryAccessors<'tcx> for rustc::ty::query::queries::explicit_predicates_of<'tcx>>::compute () at librustc/ty/query/plumbing.rs:820
820                     __query_compute::$name(move || {
(gdb)
#11 0x00007ffff308af86 in rustc::dep_graph::graph::DepGraph::with_task_impl () at librustc/dep_graph/graph.rs:340
340                     (task(cx, arg), DepNodeIndex::INVALID)
(gdb) #12 0x00007ffff3046f1c in rustc::dep_graph::graph::DepGraph::with_task () at librustc/dep_graph/graph.rs:206
206             self.with_task_impl(key, cx, arg, false, task,

@nikomatsakis
Copy link
Contributor

nikomatsakis commented Aug 27, 2018

Found the immediate problem, which is that we invoked FatalError.raise() directly. Whoops. =)

Here:

Def::Err => {
FatalError.raise();
}

@pnkfelix
Copy link
Member

visited for triage. Clearing nomination tag as @nikomatsakis did the investigation that he asked for. PR #53748 should fix the most problematic issues (namely that this would affect non-feature-gated code). @nikomatsakis recommends P-medium for the remaining issue (which I infer is a remaining effect on some other feature-gated code)

@pnkfelix pnkfelix added P-medium Medium priority and removed I-nominated labels Aug 30, 2018
@Centril Centril added F-generic_associated_types `#![feature(generic_associated_types)]` a.k.a. GATs requires-nightly This issue requires a nightly compiler in some way. labels Aug 5, 2019
Centril added a commit to Centril/rust that referenced this issue Dec 21, 2019
…atsakis

Make GATs less ICE-prone.

After this PR simple lifetime-generic associated types can now be used in a compiling program. There are two big limitations:

* rust-lang#30472 has not been addressed in any way (see src/test/ui/generic-associated-types/iterable.rs)
* Using type- and const-generic associated types errors because bound types and constants aren't handled by trait solving.
    * The errors are technically non-fatal, but they happen in a [part of the compiler](https://github.com/rust-lang/rust/blob/4abb0ad2731e9ac6fd5d64d4cf15b7c82e4b5a81/src/librustc_typeck/lib.rs#L298) that fairly aggressively stops compiling on errors.

closes rust-lang#47206
closes rust-lang#49362
closes rust-lang#62521
closes rust-lang#63300
closes rust-lang#64755
closes rust-lang#67089
@bors bors closed this as completed in b50c3b7 Dec 22, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
A-associated-items Area: Associated items such as associated types and consts. C-bug Category: This is a bug. F-generic_associated_types `#![feature(generic_associated_types)]` a.k.a. GATs P-medium Medium priority requires-nightly This issue requires a nightly compiler in some way. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue.
Projects
None yet
5 participants