================
@@ -170,102 +132,112 @@ struct alignas(ConstraintAlignment)
FoldExpandedConstraint {
const Expr *Pattern)
: Kind(K), Constraint(std::move(C)), Pattern(Pattern) {};
- template <typename AtomicSubsumptionEvaluator>
- bool subsumes(const FoldExpandedConstraint &Other,
- const AtomicSubsumptionEvaluator &E) const;
-
static bool AreCompatibleForSubsumption(const FoldExpandedConstraint &A,
const FoldExpandedConstraint &B);
+
+ llvm::FoldingSetNodeID ProfileForSubsumption() const;
----------------
AaronBallman wrote:
> This was just a (poorly named) attempt to produce a cache key for atomic
> constraint generically (we do it in two places but both places do it
> differently) - i just forgot to cleanup after myself
Ah, yeah, I think I figured that out after posting the comment. Removal makes
sense to me now.
> Completely unrelated but it might be a good in a follow up. But what we
> should profile there is more "how much time is spent finding the most
> constraint candidate"
Yeah, follow-up is fine by me. I think what would be interesting to profile is
"how much time is spent determining whether a constraint is met when
instantiating a template" and "how much time is spent checking each individual
constraint".
https://github.com/llvm/llvm-project/pull/132849
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
