What is Trackback?
The target of Trackback is almost the same as of Pingback, but it differs about the protocol used. As in Pingback the notification of references is realized by XMlRpc, Trackback uses a REST model via HTTP. Additionaly Trackback allows to send some information about the notified reference.
The complete specification is given by sixapart.com.
The Trackback client
The Trackback client sends a Trackback to a Trackback server which links its post to the specified reference. Important about this is how exactly the Trackback must be sent (as defined in the specification):
A POST request which conforms to the following points must be send to the server:
- the field Content-Type in the HTTP header must contain "application/x-www-form-urlencoded" and should additionally contain the used charset;
- the POST request must contain the URL of the referencing resource;
- the POST request can contain title, blog_name and excerpt of the referencing resource.
The charset used must always conform to the charset specified.
The servers response should is xml encoded and should specify error 0 if the Trackback was registered successfully and error 1 with given errorMessage, if problems occured.
Trackback link discovery
As for Pingback a link autodiscovery algorithm is defined for Trackback. It looks for rdf contained in the html code, and contains a trackback:ping field which gets extracted. The Autodiscovery algorithm of Trackback is described in a more detailed way in the specification.
Trackbacks in BibSonomy
Trackbacks are send like Pingbacks after a publication or a bookmark was edited or added. Spammers are ignored and Trackbacks are only send for public posts.
In the #ThreadedPingback client of BibSonomy the post is checked for containing bookmark URLs or BibTex URLs which are used to pull the website and check it for Pingback or Trackback metadata.
If found, the appropriate clientservice is used.