I’ve written previously on the trade-offs that microformats make in vocabulary design. I’m still not sure how I feel about the short-string issue, but it appears no one is waiting for me to make up my mind, so I figured I’d try it out.
So far I like it that microformats depend on well-understood technology, and that a lazy dev could point their HTML rendering component at the page if they didn’t feel like writing support. OTOH, parsing can be a bear. I think I may have cheated a little in that I insisted on XHTML, but even that had its own problems, since it’s tied to a DTD. Either you write your own entity resolver or the XML parser downloads three DTD files from w3.org. Rotten. I let my example client do just that in the interest of brevity.
On the lesscode side of things, the draft includes Python implementations of a client and server. With margins, the IETF txt format ends up allowing something less than the usual number of columns, so it cramped my Python style a little bit. I think they ended up pretty short, but one of the nice things about Python is that someone can always show you a way to do it that’s shorter and clearer. So, let’s have it! (no generators allowed, to keep things adaptable to those other languages :)