Commit 1fcf2ea

committed

Uniquify ReError on input mode in canonicalizer

1 parent 1447f9d commit 1fcf2eaCopy full SHA for 1fcf2ea

File tree

+50

-2

lines changed

+50

-2

lines changed

Lines changed: 1 addition & 2 deletions

Original file line number	Diff line number	Diff line change
`@@ -239,7 +239,7 @@ impl<Infcx: InferCtxtLike<Interner = I>, I: Interner> TypeFolder<I>`
`239`	`239`	`// FIXME: We should investigate the perf implications of not uniquifying`
`240`	`240`	// `ReErased`. We may be able to short-circuit registering region
`241`	`241`	// obligations if we encounter a `ReErased` on one side, for example.
`242`		`- ty::ReStatic \| ty::ReErased => match self.canonicalize_mode {`
	`242`	`+ ty::ReStatic \| ty::ReErased \| ty::ReError(_)=> match self.canonicalize_mode {`
`243`	`243`	`CanonicalizeMode::Input => CanonicalVarKind::Region(ty::UniverseIndex::ROOT),`
`244`	`244`	`CanonicalizeMode::Response { .. } => return r,`
`245`	`245`	`},`
`@@ -277,7 +277,6 @@ impl<Infcx: InferCtxtLike<Interner = I>, I: Interner> TypeFolder<I>`
`277`	`277`	`}`
`278`	`278`	`}`
`279`	`279`	`}`
`280`		`- ty::ReError(_) => return r,`
`281`	`280`	`};`
`282`	`281`
`283`	`282`	`let existing_bound_var = match self.canonicalize_mode {`

Lines changed: 28 additions & 0 deletions

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,28 @@`
	`1`	`+//@ compile-flags: -Znext-solver`
	`2`	`+`
	`3`	`+trait Tr<'a> {}`
	`4`	`+`
	`5`	`+// Fulfillment in the new solver relies on an invariant to hold: Either`
	`6`	+// `has_changed` is true, or computing a goal's certainty is idempotent.
	`7`	+// This isn't true for `ReError`, which we used to pass through in the
	`8`	`+// canonicalizer even on input mode, which can cause a goal to go from`
	`9`	+// ambig => pass, but we don't consider `has_changed` when the response
	`10`	`+// only contains region constraints (since we usually uniquify regions).`
	`11`	`+//`
	`12`	`+// In this test:`
	`13`	+// Implicit negative coherence tries to prove `W<?0>: Constrain<'?1>`,
	`14`	+// which will then match with the impl below. This constrains `'?1` to
	`15`	+// `ReError`, but still bails w/ ambiguity bc we can't prove `?0: Sized`.
	`16`	+// Then, when we recompute the goal `W<?0>: Constrain<'error>`, when
	`17`	`+// collecting ambiguities and overflows, we end up assembling a default`
	`18`	`+// error candidate w/o ambiguity, which causes the goal to pass, and ICE.`
	`19`	`+impl<'a, A: ?Sized> Tr<'a> for W<A> {}`
	`20`	`+struct W<A: ?Sized>(A);`
	`21`	`+impl<'a, A: ?Sized> Tr<'a> for A where A: Constrain<'a> {}`
	`22`	+//~^ ERROR conflicting implementations of trait `Tr<'_>` for type `W<_>`
	`23`	`+`
	`24`	`+trait Constrain<'a> {}`
	`25`	`+impl<A: Sized> Constrain<'missing> for W<A> {}`
	`26`	+//~^ ERROR use of undeclared lifetime name `'missing`
	`27`	`+`
	`28`	`+fn main() {}`

Lines changed: 21 additions & 0 deletions

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,21 @@`
	`1`	+error[E0261]: use of undeclared lifetime name `'missing`
	`2`	`+ --> $DIR/dont-canonicalize-re-error.rs:25:26`
	`3`	`+ \|`
	`4`	`+LL \| impl<A: Sized> Constrain<'missing> for W<A> {}`
	`5`	`+ \| - ^^^^^^^^ undeclared lifetime`
	`6`	`+ \| \|`
	`7`	+ \| help: consider introducing lifetime `'missing` here: `'missing,`
	`8`	`+`
	`9`	+error[E0119]: conflicting implementations of trait `Tr<'_>` for type `W<_>`
	`10`	`+ --> $DIR/dont-canonicalize-re-error.rs:21:1`
	`11`	`+ \|`
	`12`	`+LL \| impl<'a, A: ?Sized> Tr<'a> for W<A> {}`
	`13`	`+ \| ----------------------------------- first implementation here`
	`14`	`+LL \| struct W<A: ?Sized>(A);`
	`15`	`+LL \| impl<'a, A: ?Sized> Tr<'a> for A where A: Constrain<'a> {}`
	`16`	+ \| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ conflicting implementation for `W<_>`
	`17`	`+`
	`18`	`+error: aborting due to 2 previous errors`
	`19`	`+`
	`20`	`+Some errors have detailed explanations: E0119, E0261.`
	`21`	+For more information about an error, try `rustc --explain E0119`.

Comments

(0)