unifying the interface

Test/test_filter.cpp

@@ -42,10 +42,10 @@ TEST_F(test_filter, test_should_compress_all_zero) {
         near_zero_value, near_zero_value, near_zero_value};
     multiverso::Blob input_blob(array, sizeof(float) * size);
     multiverso::Blob compressed_blob;
-    auto compressed = try_compress(input_blob, &compressed_blob);
+    auto compressed = TryCompress(input_blob, &compressed_blob);
     ASSERT_EQ(compressed, true);
     auto de_compressed_blob =
-        de_compress(compressed_blob, size * sizeof(float));
+        DeCompress(compressed_blob, size * sizeof(float));
     ASSERT_TRUE(is_same_content_with_clipping(array, de_compressed_blob, size));
 }
 
@@ -55,10 +55,10 @@ TEST_F(test_filter, test_should_compress_most_zero_a) {
         near_zero_value, none_zero_value};
     multiverso::Blob input_blob(array, sizeof(float) * size);
     multiverso::Blob compressed_blob;
-    auto compressed = try_compress(input_blob, &compressed_blob);
+    auto compressed = TryCompress(input_blob, &compressed_blob);
     ASSERT_TRUE(compressed == true);
     auto de_compressed_blob =
-        de_compress(compressed_blob, size * sizeof(float));
+        DeCompress(compressed_blob, size * sizeof(float));
     ASSERT_TRUE(is_same_content_with_clipping(array, de_compressed_blob, size));
 }
 
@@ -67,10 +67,10 @@ TEST_F(test_filter, test_should_compress_most_zero_b) {
     auto array = new float[size]{1, 2, 0, 0, 0, 0, 0, 3};
     multiverso::Blob input_blob(array, sizeof(float) * size);
     multiverso::Blob compressed_blob;
-    auto compressed = try_compress(input_blob, &compressed_blob);
+    auto compressed = TryCompress(input_blob, &compressed_blob);
     ASSERT_TRUE(compressed == true);
     auto de_compressed_blob =
-        de_compress(compressed_blob, size * sizeof(float));
+        DeCompress(compressed_blob, size * sizeof(float));
     ASSERT_TRUE(is_same_content(array, de_compressed_blob, size));
 }
 
@@ -79,7 +79,7 @@ TEST_F(test_filter, test_should_not_compress_half_zero) {
     auto array = new float[size]{1, 0, 2, 0, 3, 0, 0, 4};
     multiverso::Blob input_blob(array, sizeof(float) * size);
     multiverso::Blob compressed_blob;
-    auto compressed = try_compress(input_blob, &compressed_blob);
+    auto compressed = TryCompress(input_blob, &compressed_blob);
     ASSERT_TRUE(compressed == false);
 }
 
@@ -88,7 +88,7 @@ TEST_F(test_filter, test_should_not_compress_most_none_zero) {
     auto array = new float[size]{1, 2, 3, 4, 0, 0, 0, 5};
     multiverso::Blob input_blob(array, sizeof(float) * size);
     multiverso::Blob compressed_blob;
-    auto compressed = try_compress(input_blob, &compressed_blob);
+    auto compressed = TryCompress(input_blob, &compressed_blob);
     ASSERT_TRUE(compressed == false);
 }
 
@@ -99,7 +99,7 @@ TEST_F(test_filter, test_should_not_compress_all_none_zero) {
         none_zero_value, none_zero_value, none_zero_value};
     multiverso::Blob input_blob(array, sizeof(float) * size);
     multiverso::Blob compressed_blob;
-    auto compressed = try_compress(input_blob, &compressed_blob);
+    auto compressed = TryCompress(input_blob, &compressed_blob);
     ASSERT_TRUE(compressed == false);
 }
 

include/multiverso/table/sparse_matrix_table.h

@@ -53,8 +53,8 @@ class SparseMatrixServerTable : public MatrixServerTable<T> {
     void ProcessGet(const std::vector<Blob>& data,
         std::vector<Blob>* result) override;
  private:
-     void update_state_on_add(int worker_id, Blob keys);
-     void update_state_on_get(int worker_id, int* keys, int key_size,
+     void UpdateAddState(int worker_id, Blob keys);
+     void UpdateGetState(int worker_id, int* keys, int key_size,
        std::vector<int>* out_rows);
      int get_global_row_id(int local_row_id) {
        return row_offset_ + local_row_id;

include/multiverso/util/quantization_util.h

@@ -41,7 +41,7 @@ namespace multiverso {
       for (auto i = 0; i < blobs.size(); ++i) {
         auto& blob = blobs[i];
         Blob compressed_blob;
-        auto compressed = try_compress(blob, &compressed_blob);
+        auto compressed = TryCompress(blob, &compressed_blob);
         // size info (compressed ? size : -1)
 #pragma warning( push )
 #pragma warning( disable : 4267)
@@ -66,13 +66,13 @@ namespace multiverso {
           size_blob.As<int>(i - 1) : blobs[i].size();
         auto& blob = blobs[i];
         outputs->push_back(is_compressed ?
-          std::move(de_compress(blob, size)) : blob);
+          std::move(DeCompress(blob, size)) : blob);
       }
     }
 
 protected:
 
-  bool try_compress(const Blob& in_blob,
+  bool TryCompress(const Blob& in_blob,
     Blob* out_blob) {
     CHECK_NOTNULL(out_blob);
 #pragma warning( push )
@@ -121,7 +121,7 @@ protected:
     return true;
   }
 
-  Blob de_compress(const Blob& in_blob, size_t size) {
+  Blob DeCompress(const Blob& in_blob, size_t size) {
 #pragma warning( push )
 #pragma warning( disable : 4127)
     CHECK(sizeof(data_type) == sizeof(index_type));

src/table/sparse_matrix_table.cpp

@@ -28,6 +28,7 @@ void SparseMatrixWorkerTable<T>::Get(int row_id, T* data, size_t size,
   if (row_id >= 0) CHECK(size == num_col_);
   row_index_[row_id] = data;  // data_ = data;
   Blob keys(&row_id, sizeof(int) * 2);
+  // TODO[qiwye]: to make worker_id an Option.
   keys.As<int>(1) = worker_id;
   WorkerTable::Get(keys);
   Log::Debug("[Get] worker = %d, #row = %d\n", MV_Rank(), row_id);
@@ -41,7 +42,7 @@ void SparseMatrixWorkerTable<T>::Get(const std::vector<int>& row_ids,
   for (int i = 0; i < row_ids.size(); ++i) {
     row_index_[row_ids[i]] = data_vec[i];
   }
-  Blob keys(row_ids.data(), sizeof(int)* (row_ids.size() + 1));
+  Blob keys(row_ids.data(), sizeof(int) * (row_ids.size() + 1));
   keys.As<int>(row_ids.size()) = worker_id;
   WorkerTable::Get(keys);
   Log::Debug("[Get] worker = %d, #rows_set = %d\n", MV_Rank(),
@@ -82,6 +83,7 @@ int SparseMatrixWorkerTable<T>::Partition(const std::vector<Blob>& kv,
       }
 
       // space for workder_id
+      // TODO[qiwye]: to make worker_id an Option.
       for (auto &kv : count) {
         ++kv.second;
       }
@@ -102,6 +104,7 @@ int SparseMatrixWorkerTable<T>::Partition(const std::vector<Blob>& kv,
       }
 
       // append workder_id
+      // TODO[qiwye]: to make worker_id an new Blob.
       for (auto& kv : *out) {
         kv.second[0].As<int>(kv.second[0].size<int>() - 1) = keys[keys_size];
       }
@@ -158,7 +161,7 @@ SparseMatrixServerTable<T>::SparseMatrixServerTable(int num_row, int num_col,
 }
 
 template <typename T>
-void SparseMatrixServerTable<T>::update_state_on_add(int worker_id,
+void SparseMatrixServerTable<T>::UpdateAddState(int worker_id,
   Blob keys_blob) {
   size_t keys_size = keys_blob.size<int>();
   int *keys = reinterpret_cast<int*>(keys_blob.data());
@@ -180,7 +183,7 @@ void SparseMatrixServerTable<T>::update_state_on_add(int worker_id,
 }
 
 template <typename T>
-void SparseMatrixServerTable<T>::update_state_on_get(int worker_id, int* keys,
+void SparseMatrixServerTable<T>::UpdateGetState(int worker_id, int* keys,
   int key_size, std::vector<int>* out_rows) {
 
   if (worker_id == -1) {
@@ -241,7 +244,7 @@ void SparseMatrixServerTable<T>::ProcessAdd(
   // must contain option that has worker id
   CHECK(data.size() == 3);
   UpdateOption option(data[2].data(), data[2].size());
-  update_state_on_add(option.worker_id(), data[0]);
+  UpdateAddState(option.worker_id(), data[0]);
   MatrixServerTable<T>::ProcessAdd(data);
 }
 
@@ -263,7 +266,7 @@ void SparseMatrixServerTable<T>::ProcessGet(
   std::vector<int> outdate_rows;
 #pragma warning( push )
 #pragma warning( disable : 4267)
-  update_state_on_get(workder_id, keys, keys_size, &outdate_rows);
+  UpdateGetState(workder_id, keys, keys_size, &outdate_rows);
 #pragma warning( pop ) 
   Blob outdate_rows_blob(sizeof(int) * outdate_rows.size());
   for (auto i = 0; i < outdate_rows.size(); ++i) {