[clang][NFC] Refactor replaceExternalDecls to use llvm::any_of

[snarang181]

This patch simplifies the declaration replacement logic in replaceExternalDecls by replacing a manual loop with an idiomatic use of llvm::any_of. This improves code readability and aligns with common LLVM coding style.

[llvmbot]

@llvm/pr-subscribers-clang

Author: Samarth Narang (snarang181)

Changes

This patch simplifies the declaration replacement logic in replaceExternalDecls by replacing a manual loop with an idiomatic use of llvm::any_of. This improves code readability and aligns with common LLVM coding style.

---

Full diff: https://github.com/llvm/llvm-project/pull/143275.diff

1 Files Affected:

• (modified) clang/include/clang/AST/DeclContextInternals.h (+4-8)

diff --git a/clang/include/clang/AST/DeclContextInternals.h b/clang/include/clang/AST/DeclContextInternals.h
index b17b7627ac90c..a1071f62d5fae 100644
--- a/clang/include/clang/AST/DeclContextInternals.h
+++ b/clang/include/clang/AST/DeclContextInternals.h
@@ -176,14 +176,10 @@ class StoredDeclsList {
     DeclListNode::Decls *Tail = erase_if([Decls](NamedDecl *ND) {
       if (ND->isFromASTFile())
         return true;
-      // FIXME: Can we get rid of this loop completely?
-      for (NamedDecl *D : Decls)
-        // Only replace the local declaration if the external declaration has
-        // higher visibilities.
-        if (D->getModuleOwnershipKind() <= ND->getModuleOwnershipKind() &&
-            D->declarationReplaces(ND, /*IsKnownNewer=*/false))
-          return true;
-      return false;
+      return llvm::any_of(Decls, [ND](NamedDecl *D) {
+        return D->getModuleOwnershipKind() <= ND->getModuleOwnershipKind() &&
+               D->declarationReplaces(ND, /*IsKnownNewer=*/false);
+      });
     });
 
     // Don't have any pending external decls any more.

[snarang181]

@ChuanqiXu9, requesting your review here.

[zwuis]

IIUC the FIXME comment means that the time complexity of this part of code is not good enough and we can make it faster.

[ChuanqiXu9]

Please remain the comments.

[snarang181]

I added the comment back. Is that what you meant?

[ChuanqiXu9]

Yes. We should keep both.

[snarang181]

IIUC the FIXME comment means that the time complexity of this part of code is not good enough and we can make it faster.

OK. In any case, the usage of llvm_any_of seemed better than using an explicit for-loop.

[ChuanqiXu9]

Here is another missing comment.

[snarang181]

OK. Should we keep the FIXME?

[snarang181]

Retain all now, I think.