Note: This is still WIP.
This is an ongoing effort to implement Protobuf column storage in C++. The format and algorithm are illustrated in Google's Dremel paper. Parquet adapts the same idea.
Required libraries:
- Protobuf 3
- Google test
- Glog
pb::Document doc;
// Full field name to serialized data.
std::map<std::string, std::string> serialized = proto_column::Serialize(doc);
for (const auto& column_and_data : serialized) {
const std::string& serialized = column_and_data.second;
// ...
}
TODO