From dbd4856faee98cb325053ca6fb8a2aca923ebe4c Mon Sep 17 00:00:00 2001
From: Davi Reis <davi@davi-macbookpro.local>
Date: Wed, 7 Mar 2012 01:48:20 -0500
Subject: [PATCH] Removed unnecessary seed mod which was breaking on presence
 of poor hash functions.

---
 cxxmph/bm_index.cc   | 10 ++++------
 cxxmph/bm_map.cc     | 42 ++++++++++++++++++++----------------------
 cxxmph/mph_index.h   |  3 +--
 cxxmph/mph_map.h     |  1 -
 cxxmph/seeded_hash.h |  9 ++-------
 5 files changed, 27 insertions(+), 38 deletions(-)

diff --git a/cxxmph/bm_index.cc b/cxxmph/bm_index.cc
index e700840..924231c 100644
--- a/cxxmph/bm_index.cc
+++ b/cxxmph/bm_index.cc
@@ -129,17 +129,15 @@ class BM_STLIndexSearch : public SearchUrlsBenchmark {
    index.swap(index_);
    return true;
   }
-  std::unordered_map<StringPiece, uint32_t> index_;
+  unordered_map<StringPiece, uint32_t> index_;
 };
 
 int main(int argc, char** argv) {
-/*
   Benchmark::Register(new BM_MPHIndexCreate("URLS100k"));
   Benchmark::Register(new BM_STLIndexCreate("URLS100k"));
-  Benchmark::Register(new BM_MPHIndexSearch("URLS100k", 100*1000*1000));
-  Benchmark::Register(new BM_STLIndexSearch("URLS100k", 100*1000*1000));
-*/
-  Benchmark::Register(new BM_CmphIndexSearch("URLS100k", 100*1000*1000));
+  Benchmark::Register(new BM_MPHIndexSearch("URLS100k", 10*1000*1000));
+  Benchmark::Register(new BM_STLIndexSearch("URLS100k", 10*1000*1000));
+  Benchmark::Register(new BM_CmphIndexSearch("URLS100k", 10*1000*1000));
   Benchmark::RunAll();
   return 0;
 }
diff --git a/cxxmph/bm_map.cc b/cxxmph/bm_map.cc
index 42516b6..8334604 100644
--- a/cxxmph/bm_map.cc
+++ b/cxxmph/bm_map.cc
@@ -4,29 +4,29 @@
 #include "bm_common.h"
 #include "mph_map.h"
 
-  using cxxmph::mph_map;
-  using std::string;
-  using std::unordered_map;
+using cxxmph::mph_map;
+using std::string;
+using std::unordered_map;
 
-  namespace cxxmph {
+namespace cxxmph {
 
-  template<class Container, class T>
-  const T* myfind(const Container& mymap, const T& k) {
-    auto it = mymap.find(k);
-    if (it == mymap.end()) return NULL;
-    return &it->second;
-  }
+template<class MapType, class T>
+const T* myfind(const MapType& mymap, const T& k) {
+  auto it = mymap.find(k);
+  if (it == mymap.end()) return NULL;
+  return &it->second;
+}
 
-  template <class MapType>
-  class BM_CreateUrls : public UrlsBenchmark {
-   public:
-    BM_CreateUrls(const string& urls_file) : UrlsBenchmark(urls_file) { }
-    virtual void Run() {
-      MapType mymap;
-      for (auto it = urls_.begin(); it != urls_.end(); ++it) {
-        mymap[*it] = *it;
-      }
+template <class MapType>
+class BM_CreateUrls : public UrlsBenchmark {
+ public:
+  BM_CreateUrls(const string& urls_file) : UrlsBenchmark(urls_file) { }
+  virtual void Run() {
+    MapType mymap;
+    for (auto it = urls_.begin(); it != urls_.end(); ++it) {
+      mymap[*it] = *it;
     }
+  }
 };
 
 template <class MapType>
@@ -90,14 +90,12 @@ int main(int argc, char** argv) {
   /*
   Benchmark::Register(new BM_CreateUrls<mph_map<StringPiece, StringPiece>>("URLS100k"));
   Benchmark::Register(new BM_CreateUrls<unordered_map<StringPiece, StringPiece>>("URLS100k"));
-  */
   Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000 * 1000, 0));
   Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur2StringPiece>>("URLS100k", 10*1000 * 1000, 0));
   Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000 * 1000, 0.9));
   Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur2StringPiece>>("URLS100k", 10*1000 * 1000, 0.9));
-  /*
+*/
   Benchmark::Register(new BM_SearchUint64<unordered_map<uint64_t, uint64_t>>);
   Benchmark::Register(new BM_SearchUint64<mph_map<uint64_t, uint64_t>>);
-  */
   Benchmark::RunAll();
 }
diff --git a/cxxmph/mph_index.h b/cxxmph/mph_index.h
index 3ee9090..d2e4a01 100644
--- a/cxxmph/mph_index.h
+++ b/cxxmph/mph_index.h
@@ -138,8 +138,7 @@ bool MPHIndex::Reset(ForwardIterator begin, ForwardIterator end) {
   std::vector<uint32_t> queue;
   while (1) {
     // cerr << "Iterations missing: " << iterations << endl;
-    for (int i = 0; i < 3; ++i) hash_seed_[i] = random() % m_;
-    // for (int i = 0; i < 3; ++i) hash_seed_[i] = random() + i;
+    for (int i = 0; i < 3; ++i) hash_seed_[i] = random();
     if (Mapping<SeededHashFcn>(begin, end, &edges, &queue)) break;
     else --iterations;
     if (iterations == 0) break;
diff --git a/cxxmph/mph_map.h b/cxxmph/mph_map.h
index e574c7c..7541c45 100644
--- a/cxxmph/mph_map.h
+++ b/cxxmph/mph_map.h
@@ -186,7 +186,6 @@ MPH_MAP_METHOD_DECL(iterator, find)(const key_type& k) {
 }
 
 MPH_MAP_METHOD_DECL(my_int32_t, index)(const key_type& k) const {
-  assert(slack_.empty());
   if (index_.size() == 0) return -1;
   return index_.index(k);
 }
diff --git a/cxxmph/seeded_hash.h b/cxxmph/seeded_hash.h
index 64cb74d..0f9adfa 100644
--- a/cxxmph/seeded_hash.h
+++ b/cxxmph/seeded_hash.h
@@ -19,13 +19,6 @@ struct seeded_hash_function {
   }
 };
 
-struct seeded_identity_function {
-  template <class Key>
-  uint32_t operator()(const Key& k, uint32_t seed) const {
-    return k ^ seed;
-  }
-};
-
 struct Murmur2 {
   template<class Key>
   uint32_t operator()(const Key& k) const {
@@ -78,6 +71,7 @@ template <> struct seeded_hash<std::hash<short> >
 { typedef seeded_hash_function<Murmur2> hash_function; };
 template <> struct seeded_hash<std::hash<unsigned short> >
 { typedef seeded_hash_function<Murmur2> hash_function; };
+/*
 template <> struct seeded_hash<std::hash<int> >
 { typedef seeded_hash_function<Murmur2> hash_function; };
 template <> struct seeded_hash<std::hash<unsigned int> >
@@ -90,6 +84,7 @@ template <> struct seeded_hash<std::hash<long long> >
 { typedef seeded_hash_function<Murmur2> hash_function; };
 template <> struct seeded_hash<std::hash<unsigned long long> >
 { typedef seeded_hash_function<Murmur2> hash_function; };
+*/
 
 }  // namespace cxxmph