std::sort() will sometimes compare an element with itself (?)
so remove the assert here. Duplicate conformance requirements
should be already caught by checkGenericSignature() anyway.
This was manifesting as module interfaces printing generic parameters
as `τ_0_0` in some cases.
Note that the GSB has the same bug, so this test case will fail with
-requirement-machine=off. I don't plan on fixing the bug in the GSB
unless we need to.
Fixes rdar://problem/78977127.
If two associated types in an (invalid) protocol have the same name, use their
source location to order them instead of their memory address. This fixes a
non-deterministic test failure with the requirement machine enabled.
The GSB and RequirementMachine sometimes disagree on whether two
type parameters that have both been equated to the same concrete
type belong to the same equivalence class or not.
Since in practice, this distinction does not matter, relax the
assert here.
Rework Sendable checking to be completely based on "missing"
conformances, so that we can individually diagnose missing Sendable
conformances based on both the module in which the conformance check
happened as well as where the type was declared. The basic rules here
are to only diagnose if either the module where the non-Sendable type
was declared or the module where it was checked was compiled with a
mode that consistently diagnoses `Sendable`, either by virtue of
being Swift 6 or because `-warn-concurrency` was provided on the
command line. And have that diagnostic be an error in Swift 6 or
warning in Swift 5.x.
There is much tuning to be done here.
Many clients of the conformance lookup operations would prefer to get
an invalid conformance (== there is no conformance) rather than a
missing conformance. Parameterize the conformance lookup operations so
that most callers won't see missing conformances, by filtering them
out at the end. Opt-in those callers that do want to see missing
conformances so they can be diagnosed.
Start treating the null {Can}GenericSignature as a regular signature
with no requirements and no parameters. This not only makes for a much
safer abstraction, but allows us to simplify a lot of the clients of
GenericSignature that would previously have to check for null before
using the abstraction.
This is just a straight port of the existing code in the GSB, with minimal changes.
It could be made more efficient in the future by trafficking in Terms rather than
Types, avoiding some intermediate conversion and canonicalization steps.
We compute the canonical type by first simplifying the type term, and
then checking if it is a concrete type. If there's no concrete type,
we convert the simplified term back to an interface type and return
that; otherwise, we canonicalize any structural sub-components of
the concrete type that contain interface types, and so on.
Due to a quirk of how the existing declaration checker works, we also
need to handle "purely concrete" member types, eg if I have a
signature `<T where T == Foo>`, and we're asked to canonicalize the
type `T.[P:A]` where Foo : A.
This comes up because we can derive the signature `<T where T == Foo>`
from a generic signature like `<T where T : P>`; adding the
concrete requirement 'T == Foo' renders 'T : P' redundant. We then
want to take interface types written against the original signature
and canonicalize them with respect to the derived signature.
The problem is that `T.[P:A]` is not a valid term in the rewrite system
for `<T where T == Foo>`, since we do not have the requirement T : P.
A more principled solution would build a substitution map when
building a derived generic signature that adds new requirements;
interface types would first be substituted before being canonicalized
in the new signature.
For now, we handle this with a two-step process; we split a term up
into a longest valid prefix, which must resolve to a concrete type,
and the remaining suffix, which we use to perform a concrete
substitution using subst().
