Speculative change for clang-analyzer.
Code should be the same but see if there is some interaction between
moved objects and NRVO that is troubling clang-analyzer.
Bug: chromium:1380478
Change-Id: I48a4d789840004ce2c5e16ab9434ba810129ca82
Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/100610
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
diff --git a/core/fpdfapi/parser/cpdf_object_walker.cpp b/core/fpdfapi/parser/cpdf_object_walker.cpp
index efe3fa5..577c1e0 100644
--- a/core/fpdfapi/parser/cpdf_object_walker.cpp
+++ b/core/fpdfapi/parser/cpdf_object_walker.cpp
@@ -142,7 +142,10 @@
// Schedule walk within composite objects.
stack_.push(std::move(new_iterator));
}
- return std::move(next_object_); // next_object_ now NULL after move.
+ // TODO(crbug.com/1380478): see if this skirts clang-analyzer issue.
+ // next_object_ will now be NULL after move.
+ auto result = std::move(next_object_);
+ return result;
}
SubobjectIterator* it = stack_.top().get();
