Protobuf Naming Conventions

Question

Besides the rather short Google provided style guide , here are my thoughts on the Google Protocol Buffer message names.

• Use the "Message" at the end of the message type names.

  • This makes it easy to see in the source code that the class is a class generated by the proto-buf. This also has the advantage that if I have a class with a rich domain, then it can have a real name, for example AddressBookMessage for the protobuf class and AddressBook for the real class.

• For Java users, it seems that java_outer_classname end in Protos is standard.

  • I did not notice this at the initial stage, so my current protobuf classes are in com.example.project.protobuf.MyProtos , but I see no reason to keep it there, since we need to have a containing class, so it can be com.example.protobuf.MyProtos to com.example.protobuf.MyProtos if there are no classes in the top package of the project.

• Start enums at 0 according to C / C ++.

• Use a unique name for the repeated field.

  • Most of the generated methods sound better with a unique field name, even if it is repeated, for example. message-> add_child (), not message-> add_children () if you had a repeated child field.

Are there any other standards that people use or differ from?

Answer 1

Disclaimer: A response from Googler using protobufs on a daily basis. I in no way represent Google.

• Do not do this. Compiled protocol buffers are simply a class definition indicated in the language you are using, with some improvements. the addition of “Message” is additional verbosity. Usually you just use protocol buffers without other class definitions, and even if you use other class definitions, just import java_outer_classname and make the exact point. You can even put the full path in the code to erase a single import line, no problem.

• Although this is not officially indicated, it sounds good, because usually you put several protons in a folder.

• You usually start with 0. See the protocol buffer language guide.

• Yes. Read the following to feel its use: https://developers.google.com/protocol-buffers/docs/javatutorial

Answer 2

I do not agree with answer 4. In a related article I can find only such examples:

 repeated PhoneNumber phones = 4; repeated Person people = 1; 

And even at https://developers.google.com/protocol-buffers/docs/proto3 we find the plural:

 repeated Result results = 1; repeated string snippets = 3;