Please send comments about this service to Doug Burke.

Recent Changes

February 24, 2009

The server has been re-located to a different machine; the data should still be available as long as you use the purl URL's - e.g. http://purl.org/net/djburke/demo/astromoat/tag/agn.

Technically this is no-longer a server but just a set of files served by Apache; the main consequences of this are:

1. Resources are no longer re-directed using the 303 See Other response code, which needs fixing.
2. Direct access to a particular MIME type has changed from the MOAT standard of appending "/<format>" to appending ".<format>".
3. The search interfaces are no longer available.

Please let me know if there are any problems.

November 20, 2008

The Location field of the 303 redirects now include the host name, previously they looked like /tag/moon/rdf. This change was made to improve the validation results provided by the Vapour linked-data validator.

Possible improvements

1. Support some of the cache control headers of HTML (such as ETags). Actually, this is now supported as we are using static files and Apache (February 24 2008).
2. Use Apache's rewrite rules to restore support from some of the lost functionality when moving from a HAppS server to Apache.
3. Improve the ordering of searches rather than use an alphabetical listing.
4. Support some form of a suggestion mechanism for a given resource, be it a simple tag-cloud analysis or something altogether more "semantic" based on the contents of the resource (e.g. if a user wants to add tags to some text).
5. Support upload of new meanings and tags using the import method from the MOAT server specifications.
6. Check that the MIME-type checks for the content negotiation are correct.
7. Support XHTML output for the tag information, and include the tag information using RDFa.

Technology behind the server

The MOAT server was originally written in Haskell using version 0.9.2 of the HAppS web-serving framework. It didn't actually need a high-performance dynamic web-serving framework and could probably have been done with a bunch of static files and some Apache configuration, but was more fun.

It is now just a set of files served by Apache.