While Fudge as a project targets the low-level encoding of hierarchical, self-describing, binary data, we realize that the most common use case for this type of framework is to pass objects around in a distributed system. Fudge Proto seeks to solve this problem, by allowing you to:
- Specify the schema for data to be encoded in Fudge messages
- Automatically generate language-specific classes for Java and C# that represents that schema
- Support encoding and decoding to and from language-specific classes in an efficient way.
The easiest way to view the Fudge Proto project is that it’s a system akin to Google Protocol Buffers or Thrift, but the data encoding adheres to the Fudge specification (and is thus self-describing).
The Basics
The content of a Fudge message is defined within a .proto file, very similar to basic
Google Protocol Buffer (GPB) files.
Although Fudge encoding and GPB are not compatible at the wire level, it is intended that existing GPB
.proto files be usable with the proto-Fudge tools with minimal modification.
A basic .proto file may contain a taxonomy, message or enumeration definition.
For example, a basic message containing a person’s name, date of birth and contact details might be described as:
message Person {
required string name;
optional date dateOfBirth;
optional repeated emailAddress;
enum PhoneType {
MOBILE = 0;
HOME = 1;
WORK = 2;
}
message PhoneNumber {
required PhoneType type;
required string number;
}
optional repeated PhoneNumber telephoneNumber;
}
The main difference from the GPB format is that the field ordinals can be omitted. Field ordinals can be
specified using a notation such as required string name = 1; for a field definition, but the reduction
of message sizes in this manner is best
handled using a taxonomy to reduce commonly used (or long field names) to 2-byte ordinals.
If targeted to an implementation such as Java, C#, or C++ would result in a class with methods to encode/decode an instance to/from a Fudge encoded message and access the elements by the symbolic names above. A language such as C that supports structured data would result in a data structure definition and conventional functions to perform the Fudge encoding/decoding. Additional methods or functions may be generated for equality comparisons, hash codes, and string representations.
Getting Started
- The GitHub home of Fudge-Proto
- The Downloads
To get started you will need the pre-built Jar file, available from releases, in your class path and use the Command Line Reference or Ant if you are working with Java. You will also need the Antlr 3 Java library.
Or, if you want to get involved with the code, check out the repository from GitHub.