As I mentioned on a comment on the above mentioned post, I like the proposed use of HTTP headers.
However, I would like to see an alternative place to store this kind of information other than the HTTP headers, because “advanced” HTTP header manipulation is out of reach for many users and in many cases: form complex CMSs, to simple hosting solutions like Amazon S3 (I tried to set a “Link:” header and it looks like s3 will not accept it).
I think that it would be useful if the discovery mechanism described in the draft would fallback to something like an extended sitemap.xml that includes rel=pub resources for each URL. This could even be an extension of the current feed mechanism.
It’s not as elegant as the HTTP headers, but it should be much easier for many publishers to implement, IMHO.
- update: Julien replied that this sounds like a good suggestion.