> Thanks Simon. If interoperability is still an issue, I think that Web Service isn't > yet ready for enterprise-class deployment. The promise and power it brings would > fall apart..
TCP/IP has been around for, what, more than 3 decades, and there have been interop quirks as recent as a year ago. Yet, I think you'll agree, it's incredibly useful. Web service interop will never be *perfect*. Specs will never be complete and unambiguous, and there will always be bugs in implementations. The questions one has to ask are things like * Is interop good enough to get a sufficient ROI for development/debug? * Are there certain features that can be avoided to maximize interoperability? I have been using SOAP since the 0.9 rev of the spec, and I believe that each project that has used it has made business sense. To maximize interop, I avoid complex types wherever possible, sticking to simple types and 1-dimensional arrays thereof. I have also found that either writing the implementation or using an open source implementation has been invaluable, since code can be fixed or workarounds added. The SOAP Toolkit has been excruciating to work with at times, although once I bit the bullet and wrote a serialization/deserialization framework to use with the low-level API, things started to go more smoothly. And .NET is nice to work with because SOAP Extensions provide a hook to do any processing necessary to finesse subtle problems. At the same time, unlike some folks I have come across, I have not fallen into the trap of thinking that everything should connect via web services. I still do integration with COM+, EJBs and, when the infrastructure is already in place, CORBA. And I still do things like have flat files (not XML) dropped on my FTP server, which I then transform and load. In other words, I think web services are useful now, but so are many of the things I've been doing over the last 17 years! Scott Nichol Do not reply directly to this e-mail address, as it is filtered to only receive e-mail from specific mailing lists.
