Lupin wrote: Neat! A couple notes: First, since this changes the HTML output of section headings this has broken several of the parser test cases. When changing parser behavior it's a really good idea to run the parser tests before and after. If anything's stopped working, you should determine whether it's because the new code is broken or because the output has legitimately changed; if it's broken, fix it. :) If it's a legitimate change, fix the test case output in parserTests.txt so it runs clean. Also, it's best to put in some test cases for the behavior you're changing if they aren't already there. This can ensure that you've coded the new feature correctly and that any future regressions will be caught. Offhand I tried tossing some examples of markup and other non-plaintext into headers to see what the output would be; things appear to be getting double-escaped. (eg, "<" turns into "&amp;lt;" in the HTML, showing "&lt;" in the mouseover popup) It might be good to de-escape here. (There should be a method on Sanitizer to intepret HTML entities into plaintext.) -- brion vibber (brion @ pobox.com)