The other day, Don Park posted about Fixed-URI for Site Metadata, and I also just saw a post from Danny Ayers on the comments that followed, The Bird is Back!. This topic has been making the rounds again.

I posted the following comments on Don's site:

For the iCite net, I am using either .icu file extensions or directories named /icu/ as standards for "guess" pings for iCites (think metadata files), *and* LINK elements in pages to point to specific iCites. I think both approaches have their place.

With the iCite net, the idea is also that each iCite points to its container (the virtual webserver, basically), which tracks every iCite in that container. This is kind-of like a webserver that spiders its own files and keeps an index of the metadata files it serves.

So, two things about this that I think might be useful:

iCites (think metadata files) can be on a different server or in a totally different directory location than the website they are about. (And, websites can be on multiple servers or across multiple domains, and defined by metadata to be a single website)

2. each iCite (think metadata file) provides a reference to an index of other metadata in the same container / server.

The two usages are, 1) make the location of metadata as flexible as possible, and 2) provide good hints as to how to find more metadata in the same container / on the same server.

***

(Note: I also posted a lot about this topic a couple months ago in Autodiscovery: spider food, interactive query, processing instruction?.)