My colleague William Henry may be onto something with his idea for using RSS as a WSDL discovery mechanism, avoiding the overhead of a UDDI-based solution when you don't need it.
I can easily imagine StrikeIron and XMethods and others adding RSS feeds that include pointers to the WSDLs they list, as an additional mechanism for publishing links to the services they list.
These types of Web pages are service directories, after all, and pages like these already provide a good alternative to a registry/repository solution for lightweight browsing and discovery.
(In fact these types of pages have survived while the public UDDI will be shut down next week.)
I can easily also imagine Web services and SOA customers starting out their small and "bottom-up" projects by putting an existing intranet Web server to work hosting a page with service descriptions and RSS feeds used the way William's Celtix demo works.
Comments (1)
Agreed, in my most recent project I decided to drop the centralised UDDI approach and simply publish the WSDL on a lightweight web server alongside HTML documentation for the service. RSS is even neater approach, is amost a defacto standard XML updates, comes with libs for almost every language out there and makes it for *really* easy for developers to subscribe and react to content updates.
Using RSS in this way gives you a very simple and decentralisesed approach to advertising services. To provide a centralised repository of services simply use an RSS aggregator. Too publish your services to the repository just subscribe. Yes, the repository has to poll the subscribed services and yes it's a scaling challange but just look at the host of RSS feed repositories available in the blogsphere and you quickly realise that while it may not be neat, it works and works well.
my 2cs worth,
Paul
ps: If you haven't already guessed I really never did like UDDI :-)
Posted by Paul Power | January 6, 2006 2:57 PM
Posted on January 6, 2006 14:57