Pekka Savola wrote: > Some of this would in fact usually be documenting the product bugs > that were caused by ambiguous or incorrect specification. You're > right that once you've figured out a way to fix a spec problem in YOUR > implementation, there may be marginal interest in fixing it in the > spec. However, I believe doing so will reduce the load on customer > support (and ultimately also engineering which will need to answer the > escalated support issues), because many of the support issues are > actually caused by interoperability between different products or > vendors. I'm not saying don't fix interoperability problems that actually have impact in a two-step approach. In fact if there *are* interoperability problems, that's an argument to recycle the spec, IMHO. I am saying that having that as a formal requirement demonstrably deters people from actually doing the work. Doing that function survey is pain in the butt. Eliot _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf