I realized that my first incarnation might have been a bit unrefined in some respects. This new version is much easier to run an XSLT parser over and eliminates a lot of the redunancies. Feel free to fetch the new version from the same spot. I also got an email from Wari Wahab (the author of pyBlosxom) today suggesting that I make some of my other changes with the browser specific flavors as a renderer. I’m going to look into that this weekend. If that works then I’ll probably just have some special code for the HTML version of the page where it generates the XSLT then sends it through xsltproc serverside before sending off to the older clients.