floodsub::lib::Message vs floodsub::rpc_proto::Message

[jamesray1]

Why do you use Message in floodsub::lib when it differs from floodsub::rpc_proto::Message?

floodsub::lib:

/// A message received by the floodsub system.
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
pub struct Message {
    /// Remote that sent the message.
    pub source: Multiaddr,

    /// Content of the message. Its meaning is out of scope of this library.
    pub data: Vec<u8>,

    /// List of topics of this message.
    ///
    /// Each message can belong to multiple topics at once.
    pub topics: Vec<TopicHash>,
}

floodsub::rpc_proto::Message:

#[derive(PartialEq,Clone,Default)]
pub struct Message {
    // message fields
    from: ::protobuf::SingularField<::std::vec::Vec<u8>>,
    data: ::protobuf::SingularField<::std::vec::Vec<u8>>,
    seqno: ::protobuf::SingularField<::std::vec::Vec<u8>>,
    topicIDs: ::protobuf::RepeatedField<::std::string::String>,
    // special fields
    unknown_fields: ::protobuf::UnknownFields,
    cached_size: ::protobuf::CachedSize,
}

Related: #473.

[tomaka]

We don't expose the message struct of the protobuf library because we don't want users to have to be aware of the fact that we use this library as part of the implementation.
By doing so, we can replace protobuf with something else if necessary, without having to modify the Message struct. And therefore the users of floodsub also will not need to make any adjustement.