Optimize SemistaticGraph using shallow copy semantics to share data w…

…ith the NormalizedComponent. Decreases injection time by ~15%.

CMakeLists.txt

@@ -5,17 +5,17 @@ if(NOT CMAKE_BUILD_TYPE)
   set(CMAKE_BUILD_TYPE "Release")
 endif()
 
-add_definitions("-std=c++11 -W -Wall")
+add_definitions(-std=c++11 -W -Wall)
 if(${CMAKE_BUILD_TYPE} STREQUAL "Debug")
-  add_definitions("-g -O2 -Werror -DFRUIT_EXTRA_DEBUG -D_GLIBCXX_DEBUG -ftemplate-backtrace-limit=0")
+  add_definitions(-g -O2 -Werror -DFRUIT_EXTRA_DEBUG -D_GLIBCXX_DEBUG -ftemplate-backtrace-limit=0)
 endif()
 
 include_directories(${CMAKE_SOURCE_DIR}/include)
 
 #set(CMAKE_CXX_COMPILER "/usr/bin/clang++")
 
 # Unsafe, only for debugging/benchmarking.
-#add_definitions("-DFRUIT_NO_LOOP_CHECK")
+#add_definitions(-DFRUIT_NO_LOOP_CHECK)
 
 
 set(INSTALL_LIBRARY_DIR lib CACHE PATH "Installation directory for libraries")

include/fruit/impl/data_structures/semistatic_graph.defn.h

@@ -31,20 +31,20 @@ inline SemistaticGraph<NodeId, Node>::node_iterator::node_iterator(typename std:
 
 template <typename NodeId, typename Node>
 inline Node& SemistaticGraph<NodeId, Node>::node_iterator::getNode() {
-  assert(itr->edgesBeginOffset != invalidEdgesBeginOffset);
+  assert(itr->edgesBegin != 1);
   return itr->node;
 }
 
 template <typename NodeId, typename Node>
 inline bool SemistaticGraph<NodeId, Node>::node_iterator::isTerminal() {
-  assert(itr->edgesBeginOffset != invalidEdgesBeginOffset);
-  return itr->edgesBeginOffset == 0;
+  assert(itr->edgesBegin != 1);
+  return itr->edgesBegin == 0;
 }
 
 template <typename NodeId, typename Node>
 inline void SemistaticGraph<NodeId, Node>::node_iterator::setTerminal() {
-  assert(itr->edgesBeginOffset != invalidEdgesBeginOffset);
-  itr->edgesBeginOffset = 0;
+  assert(itr->edgesBegin != 1);
+  itr->edgesBegin = 0;
 }
 
 template <typename NodeId, typename Node>
@@ -60,19 +60,19 @@ inline SemistaticGraph<NodeId, Node>::const_node_iterator::const_node_iterator(
 
 template <typename NodeId, typename Node>
 inline const Node& SemistaticGraph<NodeId, Node>::const_node_iterator::getNode() {
-  assert(itr->edgesBeginOffset != invalidEdgesBeginOffset);
+  assert(itr->edgesBegin != 1);
   return itr->node;
 }
 
 template <typename NodeId, typename Node>
 inline bool SemistaticGraph<NodeId, Node>::const_node_iterator::isTerminal() {
-  assert(itr->edgesBeginOffset != invalidEdgesBeginOffset);
+  assert(itr->edgesBegin != 1);
   return itr->edgesBeginOffset == 0;
 }
 
 template <typename NodeId, typename Node>
 inline void SemistaticGraph<NodeId, Node>::const_node_iterator::setTerminal() {
-  assert(itr->edgesBeginOffset != invalidEdgesBeginOffset);
+  assert(itr->edgesBegin != 1);
   itr->edgesBeginOffset = 0;
 }
 
@@ -83,9 +83,10 @@ inline bool SemistaticGraph<NodeId, Node>::const_node_iterator::operator==(const
 
 
 template <typename NodeId, typename Node>
-inline typename SemistaticGraph<NodeId, Node>::edge_iterator SemistaticGraph<NodeId, Node>::node_iterator::neighborsBegin(
-    SemistaticGraph<NodeId, Node>& graph) {
-  return edge_iterator{graph.edgesStorage.data() + itr->edgesBeginOffset};
+inline typename SemistaticGraph<NodeId, Node>::edge_iterator SemistaticGraph<NodeId, Node>::node_iterator::neighborsBegin() {
+  assert(itr->edgesBegin != 0);
+  assert(itr->edgesBegin != 1);
+  return edge_iterator{reinterpret_cast<InternalNodeId*>(itr->edgesBegin)};
 }
 
 template <typename NodeId, typename Node>
@@ -133,7 +134,7 @@ inline typename SemistaticGraph<NodeId, Node>::const_node_iterator SemistaticGra
     return const_node_iterator{nodes.end()};
   } else {
     auto itr = nodes.begin() + internalNodeIdPtr->id;
-    if (itr->edgesBeginOffset == invalidEdgesBeginOffset) {
+    if (itr->edgesBegin == 1) {
       return const_node_iterator{nodes.end()};
     }
     return const_node_iterator{itr};
@@ -147,7 +148,7 @@ inline typename SemistaticGraph<NodeId, Node>::node_iterator SemistaticGraph<Nod
     return node_iterator{nodes.end()};
   } else {
     auto itr = nodes.begin() + internalNodeIdPtr->id;
-    if (itr->edgesBeginOffset == invalidEdgesBeginOffset) {
+    if (itr->edgesBegin == 1) {
       return node_iterator{nodes.end()};
     }
     return node_iterator{itr};
@@ -156,6 +157,7 @@ inline typename SemistaticGraph<NodeId, Node>::node_iterator SemistaticGraph<Nod
 
 template <typename NodeId, typename Node>
 inline SemistaticGraphInternalNodeId SemistaticGraph<NodeId, Node>::getOrAllocateInternalNodeId(NodeId nodeId) {
+  // TODO: Optimize this, do a single lookup.
   InternalNodeId* internalNodeIdPtr = nodeIndexMap.find(nodeId);
   if (internalNodeIdPtr != nullptr) {
     return *internalNodeIdPtr;
@@ -164,49 +166,21 @@ inline SemistaticGraphInternalNodeId SemistaticGraph<NodeId, Node>::getOrAllocat
     std::size_t nodeIndex = nodes.size();
     nodeIndexMap.insert(nodeId, InternalNodeId{nodeIndex}, [](InternalNodeId x, InternalNodeId){ return x;});
     nodes.push_back(NodeData{
-#ifndef NDEBUG
+#ifdef FRUIT_EXTRA_DEBUG
     nodeId,
 #endif
-      Node(), ~std::size_t(0)});
+      Node(), 1});
     return InternalNodeId{nodeIndex};
   }
 }
 
 template <typename NodeId, typename Node>
-template <typename EdgeIter, typename Combine>
-void SemistaticGraph<NodeId, Node>::setNode(NodeId nodeId, Node node, EdgeIter edgesBegin, EdgeIter edgesEnd, Combine combine) {
-  InternalNodeId internalNodeId = getOrAllocateInternalNodeId(nodeId);
-  NodeData& nodeData = nodes[internalNodeId.id];
-#ifndef NDEBUG
-  nodeData.key = nodeId;
-#endif
-  if (nodeData.edgesBeginOffset == ~std::size_t(0)) {
-    // The node did not exist.
-    nodeData.node = node;
-  } else {
-    nodeData.node = combine(nodeData.node, node);
-  }
-  nodeData.edgesBeginOffset = edgesStorage.size();
-  for (EdgeIter i = edgesBegin; i != edgesEnd; ++i) {
-    edgesStorage.push_back(getOrAllocateInternalNodeId(*i));
-  }
-}
-
-template <typename NodeId, typename Node>
-template <typename Combine>
-void SemistaticGraph<NodeId, Node>::setTerminalNode(NodeId nodeId, Node node, Combine combine) {
-  InternalNodeId internalNodeId = getOrAllocateInternalNodeId(nodeId);
+void SemistaticGraph<NodeId, Node>::changeNodeToTerminal(NodeId nodeId) {
+  assert(nodeIndexMap.find(nodeId) != nullptr);
+  InternalNodeId internalNodeId = nodeIndexMap.at(nodeId);
   NodeData& nodeData = nodes[internalNodeId.id];
-#ifndef NDEBUG
-  nodeData.key = nodeId;
-#endif
-  if (nodeData.edgesBeginOffset == ~std::size_t(0)) {
-    // The node did not exist.
-    nodeData.node = node;
-  } else {
-    nodeData.node = combine(nodeData.node, node);
-  }
-  nodeData.edgesBeginOffset = 0;
+  assert(nodeData.edgesBegin != 1);
+  nodeData.edgesBegin = 0;
 }
 
 } // namespace impl

include/fruit/impl/data_structures/semistatic_graph.h

@@ -19,6 +19,10 @@
 
 #include "semistatic_map.h"
 
+#ifdef FRUIT_EXTRA_DEBUG
+#include <iostream>
+#endif
+
 namespace fruit {
 namespace impl {
 
@@ -50,20 +54,20 @@ class SemistaticGraph {
   // nodeIndexMap contains all known NodeIds, including ones known only due to an outgoing edge ending there from another node.
   SemistaticMap<NodeId, InternalNodeId> nodeIndexMap;
 
-  static constexpr std::size_t invalidEdgesBeginOffset = ~std::size_t(0);
-
   struct NodeData {
-#ifndef NDEBUG
+#ifdef FRUIT_EXTRA_DEBUG
     NodeId key;
 #endif
 
     Node node;
-    // (edgesStorage.begin() + edgesBeginOffset) is the beginning of the edges range.
     // If edgesBeginOffset==0, this is a terminal node.
-    // If edgesBeginOffset==invalidEdgesBeginOffset, this node doesn't exist, it's just referenced by another node.
-    std::size_t edgesBeginOffset;
+    // If edgesBeginOffset==1, this node doesn't exist, it's just referenced by another node.
+    // Otherwise, reinterpret_cast<InternalNodeId*>(edgesBegin) is the beginning of the edges range.
+    std::uintptr_t edgesBegin;
   };
 
+  std::size_t firstUnusedIndex;
+
   std::vector<NodeData> nodes;
 
   // Stores vectors of dependencies as contiguous chunks of elements.
@@ -73,6 +77,9 @@ class SemistaticGraph {
 
   InternalNodeId getOrAllocateInternalNodeId(NodeId nodeId);
 
+  // TODO: Remove.
+  void printEdgesBegin(std::ostream& os, std::uintptr_t edgesBegin);
+
 public:
 
   class edge_iterator;
@@ -95,7 +102,7 @@ class SemistaticGraph {
 
     // Assumes !isTerminal().
     // neighborsEnd() is NOT provided/stored for efficiency, the client code is expected to know the number of neighbors.
-    edge_iterator neighborsBegin(SemistaticGraph<NodeId, Node>& graph);
+    edge_iterator neighborsBegin();
 
     bool operator==(const node_iterator&) const;
   };
@@ -159,6 +166,14 @@ class SemistaticGraph {
   SemistaticGraph(SemistaticGraph&&) = default;
   SemistaticGraph(const SemistaticGraph&) = default;
 
+  // Creates a copy of x with the additional nodes in [first, last). The requirements on NodeIter as the same as for the 2-arg
+  // constructor.
+  // The nodes in [first, last) must NOT be already in x, but can be neighbors of nodes in x.
+  // The new graph will share data with `x', so must be destroyed before `x' is destroyed.
+  // Also, after this is called, `x' must not be modified until this object has been destroyed.
+  template <typename NodeIter>
+  SemistaticGraph(const SemistaticGraph& x, NodeIter first, NodeIter last);
+
   SemistaticGraph& operator=(const SemistaticGraph&) = default;
   SemistaticGraph& operator=(SemistaticGraph&&) = default;
 
@@ -174,18 +189,10 @@ class SemistaticGraph {
   node_iterator find(NodeId nodeId);
   const_node_iterator find(NodeId nodeId) const;
 
-  // Sets nodeId as a non-terminal node with outgoing edges [edgesBegin, edgesEnd).
-  // If the node already exists, combine(oldNode, newNode) is called and the result (also of type Node) is used as the node value.
-  template <typename EdgeIter, typename Combine>
-  void setNode(NodeId nodeId, Node node, EdgeIter edgesBegin, EdgeIter edgesEnd, Combine combine);
-
-  // Sets nodeId as a terminal node.
-  // If the node already exists, combine(oldNode, newNode) is called and the result (also of type Node) is used as the node value.
-  // For cases where the node is known to exist, use the setTerminal() method on an iterator, it's faster.
-  template <typename Combine> 
-  void setTerminalNode(NodeId nodeId, Node node, Combine combine);
+  // Changes the node with ID nodeId (that must exist) to a terminal node.
+  void changeNodeToTerminal(NodeId nodeId);
 
-#ifndef NDEBUG
+#ifdef FRUIT_EXTRA_DEBUG
   // Emits a runtime error if some node was not created but there is an edge pointing to it.
   void checkFullyConstructed();
 #endif