I can provide a simple example for using SUP. Assuming you are
interested in a number of conversations stored in your system you check
the CV SUP every minute (every "period" from the JSON response) and for
each resource ID (You should POST the conversations that you are interested in to CV and retrieve the ID of the conversation and store that in your database) you see if
you are interested in the updates.
If you are interested in the updates for
http://uk.techcrunch.com/2009/08/19/greactions-shows-blog-comments-right-inside-google-reader/ (this conversation has the id ce118870c21c80701e9dbcd42dd2fad8) and the SUP file contains this id, then you should get the newest reactions from the reactions endpoint using the since argument.
* At this moment we are not pushing data to the
client.