Why not create a simple relational database and redundantly log to that database whenever you log to a file? Then you can simply query your database from the webserver and get a customizable and real-time view of log activity without needing to re-parse your log files every time for every user. This approach would provide the ability to easily perform longer-term analysis of the log material as well as redundant backup of log information -- it also allows you to distribute the webserver and database on different platforms from your applications.
Of course I know nothing about RSS feeds and this doesn't really answer your question, but it is perhaps an alternative you haven't considered. :)