dave continues to be dead wrong about the value of idls for dynamic languages. perl's excellent SOAP::Lite contains specific support for wsdl, and i've already beaten the horse about my use of an idl in php. dave's alidl is a dud because it doesn't provide any type information.

the reason i rolled my own idl-based xml-rpc implementation for php is because coding for the usefulinc xml-rpc implementation is utterly gross. you end up writing verbose code to create the equivalent of what the idl gives you, and then your actual method implementation has to go through all sorts of grotty work to create values.

compare the non-idl method of writing the xml-rpc validation suite to the idl-driven method (plus the idl).

i have very little doubt that the use-case for writing client code would be any less compelling.

« january 23, 2002 8:53pm january 24, 2002 6:14pm »