Don't create impl candidates when obligation contains errors by Aaron1011 · Pull Request #73005 · rust-lang/rust

Aaron1011 · 2020-06-04T23:39:04Z

In PR #72621, trait selection was modified to no longer bail out early
when an error type was encountered. This allowed us treat ty::Error as
Sized, causing us to avoid emitting a spurious "not sized" error after
a type error had already occured.

However, this means that we may now try to match an impl candidate
against the error type. Since the error type will unify with almost
anything, this can cause us to infinitely recurse (eventually triggering
an overflow) when trying to verify certain where clauses.

This commit causes us to skip generating any impl candidates when an
error type is involved.

Fixesrust-lang#72839 In PR rust-lang#72621, trait selection was modified to no longer bail out early when an error type was encountered. This allowed us treat `ty::Error` as `Sized`, causing us to avoid emitting a spurious "not sized" error after a type error had already occured. However, this means that we may now try to match an impl candidate against the error type. Since the error type will unify with almost anything, this can cause us to infinitely recurse (eventually triggering an overflow) when trying to verify certain `where` clauses. This commit causes us to skip generating any impl candidates when an error type is involved.

rust-highfive · 2020-06-04T23:39:07Z

r? @ecstatic-morse

(rust_highfive has picked a reviewer for you, use r? to override)

estebank · 2020-06-05T00:13:40Z

You'll need to rebless.

estebank · 2020-06-05T01:50:07Z

+ // emitting additional spurious errors, since we're guaranteed
+ // to have emitted at least one.
+ if stack.obligation.references_error() {
+ debug!("no results for error type, treating as ambiguous");


Add a delay_span_bug, just in case.

With #70551, seeing an error type guarantees that delay_span_bug has been called.

r=me when CI is green then :)

Did you want to wait for #70551, or can I r+ this now?

Co-authored-by: Randy Taylor <tehgecKozzz@gmail.com>

Aaron1011 · 2020-06-09T00:45:37Z

The MCP for #70551 was accepted, so I'm going to

@bors r=estebank

bors · 2020-06-09T00:45:39Z

📌 Commit ae42c91 has been approved by estebank

Aaron1011 · 2020-06-09T21:30:20Z

Nominating for beta backport: while #72839 is 'only' a diagnostics issue, it's a very annoying one. When working on rustc, I'm frequently getting thousands of lines of useless output when my change fails to compile.

@ghost

Rollup of 9 pull requests Successful merges: - rust-lang#72706 (Add windows group to triagebot) - rust-lang#72789 (resolve: Do not suggest imports from the same module in which we are resolving) - rust-lang#72890 (improper ctypes: normalize return types and transparent structs) - rust-lang#72897 (normalize adt fields during structural match checking) - rust-lang#73005 (Don't create impl candidates when obligation contains errors) - rust-lang#73023 (Remove noisy suggestion of hash_map ) - rust-lang#73070 (Add regression test for const generic ICE in rust-lang#72819) - rust-lang#73157 (Don't lose empty `where` clause when pretty-printing) - rust-lang#73184 (Reoder order in which MinGW libs are linked to fix recent breakage) Failed merges: r? @ghost

pnkfelix · 2020-06-17T17:31:34Z

unilaterally accepting for beta-backport.

Mark-Simulacrum · 2020-06-26T13:37:42Z

@Aaron1011 -- on beta, it looks like src/librustc_trait_selection/traits/select/candidate_assembly.rs doesn't exist. Can you prepare a commit that backports this cleanly? (Or alternatively we can just not backport this as it's solely a diagnostics fix I guess?)

Aaron1011 · 2020-06-26T21:14:07Z

@Mark-Simulacrum: I've created e461d2e

Mark-Simulacrum · 2020-06-26T21:24:28Z

Thanks, included.

@ghost

…ulacrum [beta] backports This PR backports the following: * Make novel structural match violations not a `bug` rust-lang#73446 * linker: Never pass `-no-pie` to non-gnu linkers rust-lang#73384 * Disable the `SimplifyArmIdentity` pass rust-lang#73262 * Allow inference regions when relating consts rust-lang#73225 * Fix link error with #[thread_local] introduced by rust-lang#71192 rust-lang#73065 * Ensure stack when building MIR for matches rust-lang#72941 * Don't create impl candidates when obligation contains errors rust-lang#73005 r? @ghost

…ompiler-errors Don't Create `ParamCandidate` When Obligation Contains Errors Fixesrust-lang#121941 I'm not sure if I understand this correctly but this bug was caused by an error type incorrectly matching against `ParamCandidate`. This was introduced by the changes made in rust-lang#72621 (figured using cargo-bisect-rustc). This PR fixes it by skipping `ParamCandidate` generation when an error type is involved. Also, this is similar to rust-lang#73005 but addresses `ParamCandidate` instead of `ImplCandidate`.

Rollup merge of rust-lang#122360 - veera-sivarajan:bugfix-121941, r=compiler-errors Don't Create `ParamCandidate` When Obligation Contains Errors Fixesrust-lang#121941 I'm not sure if I understand this correctly but this bug was caused by an error type incorrectly matching against `ParamCandidate`. This was introduced by the changes made in rust-lang#72621 (figured using cargo-bisect-rustc). This PR fixes it by skipping `ParamCandidate` generation when an error type is involved. Also, this is similar to rust-lang#73005 but addresses `ParamCandidate` instead of `ImplCandidate`.

rust-highfive assigned ecstatic-morse Jun 4, 2020

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jun 4, 2020

Treat selection error as ambiguous when error type is present
3295c26

estebank reviewed Jun 5, 2020
View reviewed changes

aclonegeek reviewed Jun 5, 2020
View reviewed changes

Comment threadsrc/librustc_trait_selection/traits/select/candidate_assembly.rs Outdated

Fix typo
ae42c91
Co-authored-by: Randy Taylor <tehgecKozzz@gmail.com>

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 Jun 9, 2020

Aaron1011 added the beta-nominated Nominated for backporting to the compiler in the beta channel. label Jun 9, 2020

Dylan-DPC-zz mentioned this pull request Jun 10, 2020
Rollup of 9 pull requests #73206
Merged

bors merged commit 024f025 into rust-lang:masterJun 11, 2020

spastorino added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Jun 17, 2020

pnkfelix mentioned this pull request Jun 17, 2020
Unreasonably long (and useless) error message doing modulo on non-existent variable: overflow evaluating the requirement `typenum::UInt<typenum::UInt… #72839
Closed

pnkfelix added the beta-accepted Accepted for backporting to the compiler in the beta channel. label Jun 17, 2020

Aaron1011 added a commit to Aaron1011/rust that referenced this pull request Jun 26, 2020
Beta backport of PRrust-lang#73005(obligation error impl candidate)
e461d2e

Mark-Simulacrum mentioned this pull request Jun 26, 2020
[beta] backports #73766
Merged

Mark-Simulacrum pushed a commit to Mark-Simulacrum/rust that referenced this pull request Jun 26, 2020
Beta backport of PRrust-lang#73005(obligation error impl candidate)
8e8d53f

Elinvynia removed the beta-nominated Nominated for backporting to the compiler in the beta channel. label Jul 6, 2020

veera-sivarajan mentioned this pull request Mar 12, 2024
Don't Create ParamCandidate When Obligation Contains Errors #122360
Merged

Uh oh!

Uh oh!

Conversation

Aaron1011 commented Jun 4, 2020

Uh oh!

rust-highfive commented Jun 4, 2020

Uh oh!

estebank commented Jun 5, 2020

Uh oh!

estebankJun 5, 2020

Choose a reason for hiding this comment

Uh oh!

Aaron1011Jun 5, 2020

Choose a reason for hiding this comment

Uh oh!

estebankJun 5, 2020

Choose a reason for hiding this comment

Uh oh!

Aaron1011Jun 7, 2020

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Aaron1011 commented Jun 9, 2020

Uh oh!

bors commented Jun 9, 2020

Uh oh!

Aaron1011 commented Jun 9, 2020

Uh oh!

pnkfelix commented Jun 17, 2020

Uh oh!

Mark-Simulacrum commented Jun 26, 2020

Uh oh!

Aaron1011 commented Jun 26, 2020

Uh oh!

Mark-Simulacrum commented Jun 26, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

10 participants