Brennan’s Blog

I have finally figured out what all has to be done to publish a BitTorrent feed. And it seems that there is some interesting support in Internet Explorer 7. My plan is to create screencasts on various topics and publish them exclusively through torrent feeds to reduce the bandwidth I use from my hosted server.

I am hosting the tracker files on my FreeBSD server using TrackerBT and to start the seeding I am using RTorrent. And to create the torrent file I use µTorrent (pronounced micro-torrent). To ensure that RTorrent would stay alive to seed the torrent I am also using Screen, a utility for a Unix environment (FreeBSD, Linux, MacOS X) which allows me to leave RTorrent running after I have closed the remote connection to the server.

It just took me a while to get all of these pieces together...

Just to make it interesting, both TrackerBT and RTorrent provide virtually no documentation. I also had several assumptions about how the whole BitTorrent system worked. Many of those assumptions were a bit off base. First off, the tracker does remarkably very little as a part of the torrent process. It simply acts as the central node to coordinate communications between nodes. My assumption was that somehow the torrent file would tell the tracker where the source file existed so that the tracker would also seed that file, as the first BitTorrent client in the swarm.

What I learned to do was generate the torrent file using µTorrent and then upload the torrent file and the source file to the server. I placed the files where they are available on a website and adjusted the web server to recognize .torrent files as application/x-bittorrent. Then I had to get at least one BitTorrent client running to seed the source file. I did so with RTorrent.

rtorrent File.zip.torrent File.zip

Running that command would start RTorrent, load the torrent file and scan the source file to ensure it matched the checksum values defined in the torrent file. Once that is confirmed it contacts the tracker which I set when I create the torrent file with µTorrent. Once that first client announces that it is seeding the file other clients are able to start pulling bits from that client.

When I had that all working I created an RSS feed (version 2.0) so that I could publish the feeds as enclosures as a part of the blog entries. I tried using the RSS Editor for Firefox, but it was not generating a valid RSS feed. So I just grabbed the feed from dnrTV and updated it for my first torrent feed. I put my new video feed in place and added the following link information to the header of my Wordpress blog.

<link rel="alternate" type="application/rss+xml" title="Brennan&#8217;s Video Feed"
  href="http://brennan.offwhite.net/blog/videofeed.xml" />

Now when you use the most current versions of Firefox or IE7 you will see the orange feed icon appear. For Firefox it shows in the location bar while IE7 always has a feed icon showing, but it is gray if the current page is not listing feed information. And with IE7 you can even view the RSS feed directly and it shows a nicely formatted screen which allows you to subscribe to the feed. The enclosure is also shown as an attachment with paperclip icon as you can see in the image below.

Currently the available blog readers routinely check for updates for these RSS feeds, but the combination of RSS, enclosures (file attachments) and torrent feeds is still fairly new. To make these video blogs as convenient as using Tivo, the torrent feed would automatically be loaded into your favorite BitTorrent client to immediately start downloading the content so it is ready when you want it. As yet I have only seen one project which seeks to fill this void, and that is PwopCatcher, from the creators of dnrTV. It is still marked as Alpha, and it personally did not work in my first attempts so it will be a little while longer before we will have our Tivo for torrents. But I expect we should have it very soon.