« Web Site Metadata | Main | Web vs. Web Architecture »

Tuesday, February 03, 2009


Feed You can follow this conversation by subscribing to the comment feed for this post.

Peter Keane

It has always struck me that sitemaps (perhaps not as implemented, but the general approach) was bringing us closer to the interoperability/discoverability ideals of the Semantic Web effort than RDF, etc. It's already working in fact -- since presumably site maps feed the indexes that allow us to search the web. I'm curious to hear/read/think more about Atom-based sitemaps as well -- seems quite natural. And feed-discovery is so well established (head links) that we have a foundation to build on.

Subbu Allamaraju

Better yet - something along the lines of

[atom:link type="application/vnd.google-earth.kml+xml" href="http://www.example.com/example1.kml"/]


@subbu: in theory, yes. in practice, the sitemaps format is a given and it does have this decidedly weird idea of using [url][loc]URI[/loc][/url] instead of just saying [url href="URI"/] (what almost any XML person would do). i guess it must have been designed by one of those people who do not like attributes.

( using [] instead of real XML because, believe it or not, typepad is not capable of dealing with markup in comments. typepad is so badly broken in so many different ways. )


@peter: yes, sitemaps (the general approach) are good, because they allow machine-based discovery. it is just that the current format and spec are decidedly sub-optimal, which is sad, because like you say, it is a good approach. sitemaps.org does mention feeds, but that is more in passing and they don't define any extensions, so it is not really all that useful in practice.

i think in the end, the interesting question would be how big of an overlap there is between sitemaps XML and atom. both list URIs (well, almost; atom id's don't even have to be URIs), but sitemaps in random order with minimal and optional metadata (@lastmod, @changefreq, and @priority), and atom in time-ordered fashion and with different and richer metadata.

i think the big difference is that atom is well thought-through and well-specified, whereas sitemaps are a quick hack and the "specification" is more a rough draft. a new one should be coming out soon, i am curious to see how it will change the format (if at all). hopefully, it will at least more completely specify the format and its extensibility.

The comments to this entry are closed.