At 12:55 PM 7/2/96, Larry Masinter wrote: >Most of these are OK, but "Stableness of versions" is a repository >requirement, not a protocol requirement. We might want to call out >some protocol requirement where a client can DETERMINE whether the >repository supports stable versions independent of other kinds of >modifications. It seems to me that the desire that a versioned URL be stable is inherent to the notion of versioning, and can be a protocol requirement (ie. assumption that users of the protocol are entitled to make). However, since URLs are not guaranteed to be stable, we still have a problem, because if the base URL is re-used, the versions should obviously not have to be persistent. Perhaps we can simply say "If the document addressed by a base URL (we don't have a proper name for this concept) remains the same, then the versions of that document will be stable. I think this is important, as a stability guaranteee (even a weak one) allows versioning to be used to provide intelligent caching, distribution, and so on. And a server implementer is free to strengthen the guarantee if they want to, by not re-using base URLs. ----------------------------------------------+---------------------------- David Durand dgd@cs.bu.edu | david@dynamicDiagrams.com Boston University Computer Science | Dynamic Diagrams http://cs-www.bu.edu:80/students/grads/dgd/ | http://dynamicDiagrams.com/