You are right that my style of communication isn't always nice, in fact, I'm pretty sure that most everyone here already knows that. I blow things way out of proportion, I make incorrect assumptions, etc.
On the other hand, I'm very receptive to anything indicating a bug. I think the main issue here is fairly simple: TWiST and tWSDL are for machine communication. This is vastly different than human communication. A guiding principle is to make things as simple as possible for clients (software), without regard to humans. But by simple, I don't mean accommodating. The best starting point for that is the WS-I Basic Profile. However, tWSDL has limitations above and beyond those outlined in that document.
OpenACS packages tend to be standardized for another purpose: pleasing humans. I chose to please the machines first, because even though I can argue with humans, machines refuse to even respond if they are not addressed correctly. I also have to please myself. I would rather TWiST do one thing very well, without errors, than to take on another layer of development too soon. I'm also starting to think that any OpenACS package should start over at the same level as TWiST, not above it. Then the OpenACS API could amount to a graphical interface to develop and deploy and monitor web services. Essentially you would have one page per equivalent TWiST API and use the guts of this API as a script on a Tcl page, with minor changes.
