On Apr 5, 2010, at 2:01 AM, Robert Kisteleki wrote: > > I disagree with this. I think that spending hours and hours on specifying, > debating, coding, testing, etc. a feature that is used in 0.02% (or 0.003%, > depending on how you count) of the cases is not a good investment, imo. > > Such an underused feature will not have a well tested codebase, with all > possible bugs ironed out, which is a recipe for security problems.
My point was, it's all about perspective. Whether you like it or not, not accommodating those networks when adding new capabilities, when those routes are perfectly legitimate (i.e., within spec of a standards track document) and work just fine today, is inappropriate. As I've stated before, if we do things in SIDR that introduce new constraints on what and how people are doing something today that's completely within the bounds of legitimacy, then we ought to have a darn good reason to do so. That said, it might be less resource intensive to just contact those folks originating AS_SETs and convince them they don't need'm - assuming they don't. -danny _______________________________________________ sidr mailing list [email protected] https://www.ietf.org/mailman/listinfo/sidr
