Server-side parsing of htm4l documents

I'm pretty sure this is not a good idea in general, though you could certainly do some powerful things with it. If the server checked for an out-of-date .html file and re-generated it with m4, there would be less of a performance hit. If people start to use htm4l, someone, somewhere, is surely going to hack their server to do the macro processing. I'd be interested to hear how it goes and if it's useful for anything. The server could define macros encoding the various information it has (User-Agent, Referer etc.) and the htm4l file could produce the appropriate html. For example, it could recognize abilities of the browser and not bother giving it information it couldn't handle it (e.g., about frames).