James M Snell wrote: >> how about s/IRI/URI/ ? > Whenever RFC4287 says "atomUri" it means IRI. I understand > what you're saying, but the value of the href value may, in > fact, be an IRI. The only case where I watched this before were some XMPP RFCs, and there it always worked in a way that clients never need to worry about it: Either it is already an URI, or it's sent to a server who's supposed to know how to translate IRIs to URIs. Clients don't need to know how this works. Apparently different for atom, I missed that point in RFC 4287. You've of course no reason to be more restrictive for rel="license". > An analogous scenario is when I distribute some piece of open > source software under the Apache license. The zip I > distribute contains the ASF LICENSE file. It also happens to > contain jars for a number of dependencies from other > projects, all of which are individually licensed. Okay, but your license would cover something "real" in the zip, files you have created. It's not only about your arrangement of content collected from 3rd parties. Or is that precisely the idea of the rel="license" for a feed ? Then folks like me could need a hint that what they really want is a rel="license" for each of their own entries, because there is no global default. Frank _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf