Jeni's Musings

Yes, I’m doing some CSS parsing, but it’s pretty straight-forward: all the CSS is generated from the WordprocessingML-to-XHTML stylesheet, so it’s very predictable, and I’m not doing all the things that I should be regarding ignoring the contents of string literals and so on. Also, the selectors that I’m dealing with are all very simple (just element names or class names with preceding .) so working out how to apply declarations to individual elements isn’t as complicated as it would be in the general case.

I agree with your observations about being able to isolate particular parts of a pipeline being incredibly useful for debugging.