Remove compile warning on clang 8.0 in BufferArg.h

* CHECK_XX in glog need the left and right values are both signed or
  unsigned integer. So we could use integer literal syntax in C/C++
  to make literal unsigned or signed. Add suffix 'UL' means unsigned
  long.

Author: reyoung

paddle/function/BufferArg.h

@@ -189,8 +189,8 @@ class SequenceIdArg : public BufferArg {
   SequenceIdArg(const TensorShape& shape, ArgType argType = UNSPECIFIED)
       : BufferArg(VALUE_TYPE_INT32, shape, argType) {
     bufferType_ = TENSOR_SEQUENCE_ID;
-    CHECK_EQ(shape_.ndims(), (size_t)1);
-    CHECK_GT(shape_[0], 1);
+    CHECK_EQ(shape_.ndims(), 1UL);
+    CHECK_GT(shape_[0], 1UL);
     numSeqs_ = shape_[0] - 1;
   }
 
@@ -199,7 +199,7 @@ class SequenceIdArg : public BufferArg {
                 ArgType argType = UNSPECIFIED)
       : BufferArg(buf, VALUE_TYPE_INT32, shape, argType) {
     bufferType_ = TENSOR_SEQUENCE_ID;
-    CHECK_EQ(shape_.ndims(), (size_t)1);
+    CHECK_EQ(shape_.ndims(), 1UL);
     numSeqs_ = shape_[0] - 1;
   }
 
@@ -280,9 +280,9 @@ class SparseMatrixArg : public BufferArg {
         type_(static_cast<SparseDataType>(type)) {
     bufferType_ = TENSOR_SPARSE;
     CHECK((valueType == VALUE_TYPE_FLOAT) || (valueType == VALUE_TYPE_DOUBLE));
-    CHECK_EQ(shape_.ndims(), (size_t)2);
-    CHECK_EQ(row_.shape().ndims(), (size_t)1);
-    CHECK_EQ(col_.shape().ndims(), (size_t)1);
+    CHECK_EQ(shape_.ndims(), 2UL);
+    CHECK_EQ(row_.shape().ndims(), 1UL);
+    CHECK_EQ(col_.shape().ndims(), 1UL);
     if (format_ == T_SPARSE_CSR) {
       CHECK_EQ(nnz, col.shape()[0]);
     } else if (format_ == T_SPARSE_CSC) {
@@ -304,7 +304,7 @@ class SparseMatrixArg : public BufferArg {
         type_(static_cast<SparseDataType>(type)) {
     bufferType_ = TENSOR_SPARSE;
     CHECK((valueType == VALUE_TYPE_FLOAT) || (valueType == VALUE_TYPE_DOUBLE));
-    CHECK_EQ(shape_.ndims(), (size_t)2);
+    CHECK_EQ(shape_.ndims(), 2UL);
 
     /// len of row_ : height + 1 (CSR) or nnz (CSC), buf_ == nullptr
     row_ = (format_ == T_SPARSE_CSR
@@ -325,7 +325,7 @@ class SparseMatrixArg : public BufferArg {
     CHECK(buf_);
     CHECK(valueType_ == DataType<real>::value);
     // CHECK(deviceType_ == DType);
-    CHECK_EQ(2, shape_.ndims());
+    CHECK_EQ(2UL, shape_.ndims());
     return typename Tensor<real, DType>::SparseMatrix(
         reinterpret_cast<real*>(buf_),
         reinterpret_cast<int*>(row_.data()),