Our data distribution platform, based on NDN technology.
Offline sync is a hard problem. We want to be able to retrieve data from the internet, from USB thumbdrives, over the local network with services like Avahi, and in the future, maybe even functionality like mesh networking.
The Named Data project takes an interesting approach to this problem space allowing us to send and recieve "data packets".
The missing glossary
Here are the words you need to know:
Data packet - A data packet, in NDN terminology, is the core piece of data exchange in NDN. Much like BitTorrent's chunks, users exchange data packets which are named. Data packets are currently limited to 8800 bytes, so approaches like chunking are required to break up larger pieces of data into smaller data packets.
Name - A global identifier for a specific data packet. Should include a version to identify the specific piece of data. If you are aware of a name, you can ask the "NDN network" for a data packet that maps to that name. Names are hierchical in a DNS-like fashion, e.g. in
Producer - A program that knows how to provide a data packet for a name within its scope. It should register a prefix with the NFD, e.g. /foocorp/, and returns or delegates incoming interests, returning data packets.
Interest - A request for a data packet. In the most basic case, is just the name of the data packet. They can have flags attached to them. You should note that an Interest describes a 'need' for data, the actual data forfilling that need may come from another Name, that said the Interest's Name has to be a prefix of the Data Name. i.e.
/foocorp/alice/my_data.txt/v1can be satisfied by a Data Packet on the Name
Command Interest - The Routing Deamon (NFD) can be configured at runtime, routes, strategies, filters, and much more can be altered via the Management API that is activated by sending special kinds of Signed Interests called Command Interests.
Consumer - A program that requests a named data packet using an interest.
NFD - The Name Forwarding Daemon. A piece of software on the client machine that is aware of all producers and consumers and pairs the two up based on what it believes is the best route.
Named Data - Data is said to be named when there is a direct, unalterable mapping between a chunk of data and a name. NDN Data packets have that property, you cannot have 2 different data packets with the same name (or very bad things will occur). This concept is important, because sets of named data are easily mapped to NDN Packets, other sets (like random files on HTTP) must first be translated into an inmutable naming scheme.
Chunk - Since NDN packets are limited in size, approaches are taken to split larger files up into small data packets. A standard approach is called "chunking", where files are chunked into N-sized packets, and each data packet has a suffix clarifying its place in the sequence. NDN has standard naming conventions for this approach.
Face - A connection to the NFD program. Effectively an API around a UNIX socket, WebSocket, TCP, UDP or anything that establishes a connection with NFD (even ethernet, there is in fact a libpcap based transport). In NDN terms a face can be (without distinctions) another router (another NFD) or an application. Through the Face API you can register prefixes, send data and express interests, among other things.
EDD - Acronym for ‘EOS Data Distribution’, the name of the project.
Coding style and NDN API references
This package follows PEP 8.
This package is licensed under the LGPLv3+.