QDoc: Sort non-function nodes by name then erase duplicatesHEADdev

The order of node information in index files could change between runs
of QDoc. This appears to be due to the sorting method used before
erasing duplicate non-function nodes.

Task-number: QTBUG-136483
Change-Id: Ia58585c19e1c22172ee4c58c3ba054ec5d14a0d5
Pick-to: 6.9
Reviewed-by: Paul Wicking <[email protected]>

1 files changed, 2 insertions, 4 deletions

diff --git a/src/qdoc/qdoc/src/qdoc/aggregate.cpp b/src/qdoc/qdoc/src/qdoc/aggregate.cpp
index ebdf881d5..0a1aa2606 100644
--- a/src/qdoc/qdoc/src/qdoc/aggregate.cpp
+++ b/src/qdoc/qdoc/src/qdoc/aggregate.cpp
@@ -303,13 +303,11 @@ void Aggregate::normalizeOverloads()
 const NodeList &Aggregate::nonfunctionList()
 {
     m_nonfunctionList = m_nonfunctionMap.values();
+    // Sort based on node name
+    std::sort(m_nonfunctionList.begin(), m_nonfunctionList.end(), Node::nodeNameLessThan);
     // Erase duplicates
-    std::sort(m_nonfunctionList.begin(), m_nonfunctionList.end());
     m_nonfunctionList.erase(std::unique(m_nonfunctionList.begin(), m_nonfunctionList.end()),
                             m_nonfunctionList.end());
-
-    // Sort based on node name
-    std::sort(m_nonfunctionList.begin(), m_nonfunctionList.end(), Node::nodeNameLessThan);
     return m_nonfunctionList;
 }