The Cost of Simple Infrastructure
I am still very puzzled about a post by Dave Winerís Scripting News last Saturday. Itís taken me this long to poke around and think about this matter. I’m not done yet but here are my first collected thoughts on the matter. I want to be clear that I am not criticizing Userland or Dave Winer on this matter. This is everybodyís problem.
The headline reads:
Hereís the difference:
“In SOAP, procedures take named parameters and order is irrelevant. In XML-RPC order is relevant and parameters do not have names.”
Here is the key impact on UserLandís core productóFrontier:
“This has forced a change in Frontier’s implementation of SOAP.”
The bottom line for Frontier servers and Radio UserLand users:
“This will require a new release of Frontier and Radio, which will happen in the next few days. For right now all your SOAP is broken. We’ll get you back on the air shortly.”
How does this happen? Userlandósupposedly the key independent software vendor involved with the development of SOAPóis somehow forced to rearchitect its core infrastructure because of a not-so-subtle distinction between supposedly interoperable technologies. While today, this kind of change may only take a few weeks (not a few days) to ripple through the installed base of Frontier servers. What if there were a million Frontier servers instead of a few hunderd or a few thousand? What if there were three or four million Radio clients? This kind of change would not be insignificant.
I suspect we havenít seen the last of impact of the difference between SOAP and XML-RPC. Or even SOAP vs SOAP. Simplicityóas in Simple Object Application Protocol (SOAP)ócomes at a cost; compliance is partially defined and optional.
