There's a moment when you're looking at a reStructuredText parser, you have a basic idea of the spec, and you know with absolute certainty it's not parsing it correctly. It gets the basics right, sure. But now you know why the tools that use it have disclaimers like "Any para lists or other directives must be at the end of the string". And you're thinking "how hard would it *really* be to write one?". After all, I'm now extracting all the docstrings from the code... Is it really that much more? Shit...