phil ringnalda

Has anyone ever given any thought to how RSS and Atom autodiscovery ought to work in frames and framesets? (Probably not, since most autodiscovery happens with a single URI, not with browsing.)

Given the URI for a frameset, you’re either going to discover feeds from <link>s in the frameset doc only, or if someone’s bugged you about their framed weblog, you might recurse into the default frame documents as well.

But suppose instead you are a browser: you load a frameset, and gather the <link>s from the frameset and the default frames, and then there’s subframe navigation: do you wipe out everything you’ve discovered, or only things discovered in the replaced frame(s), or do you keep appending everything you discover until you replace the top-level document?

I can make a case for either way, just the current set of frames and framesets as the equivalent of what you would get if it wasn’t framed, or accumulated as being more useful (you hit the front page of Frames-R-Us, then the Oak category and then the Gilt category, at which point you can choose among the main feed, the New Oak feed, or the New Gilt feed), but I can’t come up with a real reason to say that either is right.

(To save you the trouble of checking on what Firefox does now, it only sees what’s in the top frameset, a fact I apparently knew in mid-December when I testcased it, but had forgotten until I was doing due diligence before filing a new bug just now.)