þreader concept/project description

þreader will be a web application focusing initially on taking discussions on the web (either via feeds (Atom, RSS) or screen-scraping) which have explicit or implicit thread structure, pulling them into a local store for fast access, and presenting a UI optimized for efficient, comfortable, thread-aware reading.

My personal goal is to “bring back the glory days of Usenet”, or rather my own experience of using a good, offline, threaded, NNTP client. þreader as initially implemented will not have any NNTP (or email) features, but that could come in the future.

Student project info

I intend to use þreader as my project for the Fall 2011 courses MP451 Open Source Software and CS459 Human-Computer Interaction at Clarkson University.

As this is an open-source project, all materials will be made public; where applicable, they will be reusable under the MIT license.

Core functionality

Fundamentally, þreader will be a feed reader. It will initially be a web application, in that the user interface runs in a web browser (mainly for portability).

It will have an “unread flags” type user interface: every new post found in feeds is initially marked unread, then marked read once the user has viewed it. The user interface will present a single post, possibly in context, together with a graphical display of the thread structure involving that post.

Feature ideas / considerations

These are some simple ideas for the type of functionality which distinguishes þreader from other feed or threaded-discussion readers.

• Recognize that time-ordering is important as well as threading. Support messages in multiple threads/replies to multiple messages.
• Infer thread structure from unthreaded content:
  • Interpret “conversations among blogs”: If a blog post links to another in the first paragraph, then it may be a reply.
  • In a single blog with categories/tags, the categories may constitute threads.
  • Support extracting threaded discussions from the HTML of sites which have threads but do not expose them in feeds, such as Reddit.
• Support following threads through automatic feed subscription:
  • In “conversations among blogs”, subscribe to replied-to feeds to get a wider view of the conversation.
  • When a blog's comments are of interest, subscribe to per-post comment feeds automatically, and consider that feed's entries to be replies to the post.

Implementation plans

The entire system will be written in JavaScript; I wish to provide for easy revision of decisions to place particular functionality on the client or server side.

The server/back-end component will, unless I find reason to do otherwise, be built on the node.js platform. The client component will be written in HTML and JavaScript and communicate with the server in the usual current AJAX/JSON/Comet rich webapp style.