Adam Blum has written an article about using UDDI as an Extended Web Services Registry. I’m currently involved in a customer project where a lot of these issues, especially surrounding versioning, play an important role.
In passing, Adam’s article highlights one of the really annoying details of UDDI:
Examples of this might be classifications of average response time (fast, medium, slow), categories of average throughput (>10 Mbps, 1-10Mbps, <1Mbps), and overall reliability (99.99%, 99.9%, 99%).
Hidden in this is the fact that with UDDI, there is no way to store a value like
10 and then later do a search for
<20 — you can only do exact or wildcard searches.