================
@@ -1,2 +1,47 @@
// RUN: %clang_cc1 -fclangir -emit-cir -mmlir --mlir-print-ir-after-all
-clangir-enable-idiom-recognizer %s -o %t.cir 2>&1 | FileCheck %s
-check-prefix=CIR
// CIR: IR Dump After IdiomRecognizer: cir-idiom-recognizer
+
+// RUN: %clang_cc1 -std=c++17 -triple x86_64-unknown-linux-gnu -fclangir
-clangir-enable-idiom-recognizer -emit-cir -mmlir
--mlir-print-ir-after=cir-idiom-recognizer %s -o %t.cir 2>&1 | FileCheck %s
--check-prefix=RAISED '--implicit-check-not=cir.call
@_ZSt4findIPccET_S1_S1_RKT0_'
+// RUN: FileCheck %s --check-prefix=FINAL --input-file=%t.cir
+
+namespace std {
+template <class Iter, class T>
+__attribute__((pure)) Iter find(Iter, Iter, const T &) noexcept;
----------------
erichkeane wrote:
Doesn't libcxx use an inline namespace here too? Can we have a test that
validates those uses?
Also, can we look and see what MSVC's STL does to make sure we have a pattern
that shows that IT works?
Also, what about a `std::another_ns::find` where `another_ns` is not inline? We
should also probably check `anonymous` namespaces here too.
https://github.com/llvm/llvm-project/pull/208854
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits