diff --git a/src/App/MappedElement.cpp b/src/App/MappedElement.cpp
index 14eb6f99ba91..5df937adfe9e 100644
--- a/src/App/MappedElement.cpp
+++ b/src/App/MappedElement.cpp
@@ -22,3 +22,105 @@
 
 #include "PreCompiled.h"
 #include "MappedElement.h"
+
+using namespace Data;
+
+bool ElementNameComparator::operator()(const MappedName &a, const MappedName &b) const {
+    size_t size = std::min(a.size(),b.size());
+    if(!size)
+        return a.size()<b.size();
+    size_t i=0;
+    if(b[0] == '#') {
+        if(a[0]!='#')
+            return true;
+        // If both string starts with '#', compare the following hex digits by
+        // its integer value.
+        int res = 0;
+        for(i=1;i<size;++i) {
+            unsigned char ac = (unsigned char)a[i];
+            unsigned char bc = (unsigned char)b[i];
+            if(std::isxdigit(bc)) {
+                if(!std::isxdigit(ac))
+                    return true;
+                if(res==0) {
+                    if(ac<bc)
+                        res = -1;
+                    else if(ac>bc)
+                        res = 1;
+                }
+            }else if(std::isxdigit(ac))
+                return false;
+            else
+                break;
+        }
+        if(res < 0)
+            return true;
+        else if(res > 0)
+            return false;
+
+        for (; i<size; ++i) {
+            char ac = a[i];
+            char bc = b[i];
+            if (ac < bc)
+                return true;
+            if (ac > bc)
+                return false;
+        }
+        return a.size()<b.size();
+    }
+    else if (a[0] == '#')
+        return false;
+
+    // If the string does not start with '#', compare the non-digits prefix
+    // using lexical order.
+    for(i=0;i<size;++i) {
+        unsigned char ac = (unsigned char)a[i];
+        unsigned char bc = (unsigned char)b[i];
+        if(!std::isdigit(bc)) {
+            if(std::isdigit(ac))
+                return true;
+            if(ac<bc)
+                return true;
+            if(ac>bc)
+                return false;
+        } else if(!std::isdigit(ac)) {
+            return false;
+        } else
+            break;
+    }
+
+    // Then compare the following digits part by integer value
+    int res = 0;
+    for(;i<size;++i) {
+        unsigned char ac = (unsigned char)a[i];
+        unsigned char bc = (unsigned char)b[i];
+        if(std::isdigit(bc)) {
+            if(!std::isdigit(ac))
+                return true;
+            if(res==0) {
+                if(ac<bc)
+                    res = -1;
+                else if(ac>bc)
+                    res = 1;
+            }
+        }else if(std::isdigit(ac))
+            return false;
+        else
+            break;
+    }
+    if(res < 0)
+        return true;
+    else if(res > 0)
+        return false;
+
+    // Finally, compare the remaining tail using lexical order
+    for (; i<size; ++i) {
+        char ac = a[i];
+        char bc = b[i];
+        if (ac < bc)
+            return true;
+        if (ac > bc)
+            return false;
+    }
+    return a.size()<b.size();
+}
\ No newline at end of file
diff --git a/src/App/MappedElement.h b/src/App/MappedElement.h
index 7f6ee68f010c..9c37c3e8ba8a 100644
--- a/src/App/MappedElement.h
+++ b/src/App/MappedElement.h
@@ -98,6 +98,22 @@ struct AppExport MappedElement
     }
 };
 
+struct AppExport ElementNameComparator {
+    /** Comparison function to make topo name more stable
+     *
+     * The sorting decomposes the name into either of the following two forms
+     *      '#' + hex_digits + tail
+     *      non_digits + digits + tail
+     *
+     * The non-digits part is compared lexically, while the digits part is
+     * compared by its integer value.
+     *
+     * The reason for this is to prevent names with bigger digits (which usually means
+     * they come later in history) from coming earlier when sorting.
+     */
+    bool operator()(const MappedName &a, const MappedName &b) const;
+};
+
 }// namespace Data