[clang-scan-deps] Remove const from ModuleDeps loop to enable move.

[naveen-seth]

This changes the iteration from const to non-const so that std::move results in a true move rather than a copy.

[llvmbot]

@llvm/pr-subscribers-clang-modules

@llvm/pr-subscribers-clang

Author: Naveen Seth Hanig (naveen-seth)

Changes

This changes the iteration from const to non-const so that std::move results in a true move rather than a copy.

---

Full diff: https://github.com/llvm/llvm-project/pull/161109.diff

1 Files Affected:

• (modified) clang/tools/clang-scan-deps/ClangScanDeps.cpp (+1-1)

diff --git a/clang/tools/clang-scan-deps/ClangScanDeps.cpp b/clang/tools/clang-scan-deps/ClangScanDeps.cpp
index 0e2758d123edc..e41f4eb7999ae 100644
--- a/clang/tools/clang-scan-deps/ClangScanDeps.cpp
+++ b/clang/tools/clang-scan-deps/ClangScanDeps.cpp
@@ -420,7 +420,7 @@ class FullDeps {
     std::vector<ModuleDeps *> NewMDs;
     {
       std::unique_lock<std::mutex> ul(Lock);
-      for (const ModuleDeps &MD : Graph) {
+      for (ModuleDeps &MD : Graph) {
         auto I = Modules.find({MD.ID, 0});
         if (I != Modules.end()) {
           I->first.InputIndex = std::min(I->first.InputIndex, InputIndex);

[ChuanqiXu9]

Why is this a standalone PR? I feel this can be within other PR. The PR itself doesn't do anything.

[naveen-seth]

Ah I see, I've moved it to #155523 now (even though that is also NFC). Hope that’s fine anyway.
Otherwise I can add them to a more meaningful PR in the future.

[Bigcheese]

This PR does do something, it change a copy to a move of a large data structure, which has (minor) perf/memory impact. This seems fine to me as a standalone PR, and doesn't really fit as a side change in any other existing PR.