Ezra's Research

The first two things that come to mind for me (mostly because I have to teach them to someone else a couple of times a week) are how character entity references actually work, and thus why &yen 50 is fine but &yen50 has to be &yen;50, and the fact that <img src=""> isn't an image with an empty source, it's an image with a relative URL for the src that evaluates to the current page's URL, and yes it'll be requested again if we don't already have a cached copy of the page without a Vary: header to tell us that we won't get an image if we re-request with our "images, please" Accept: header, and yes if that will mess up your form handling code that can't be bothered to tell POST from GET then you really ought to use <img> to have no src attribute instead, or insert the whole image when you know what the URL for it will be, and no I don't care to hear about how it works in a browser that fails to support content negotiation in any useful way...

Um, I learned that <img src=""> in the page you return in response to a POST is a really bad idea. And some stuff about why your comment preview page winds up decoding entities, so that someone who carefully previews will then submit something broken.

Of course, my bar isn't very high, just my blood pressure.