I apologize for the delay, but draft-deason-afs3-type-time-01 has been published:
<http://datatracker.ietf.org/doc/draft-deason-afs3-type-time/> The largest change is that there are now three types defined: AFSRelTimestamp (no resolution), AFSTime, and the new AFSTimestamp (no resolution). There is now extensive text explaining the details and reasoning for the "resolution" field in AFSTime. While we can still discuss/argue about it here, I just want to bring up the possibility that we _could_ to some extent defer such things for the I-Ds that actually make use of these time types (since this document leaves the decision of AFSTime vs AFSTimestamp completely up to the using RPC/structure). It could even be split into a separate I-D, if that is desired. Other changes include the addition of sample conversion code per Jeff, and text recommending keeping the special value of timestamp 0 per Simon. I think that the actual meaning of timestamp 0 is up to each RPC/structure, so in this document, the special case of 0 is just a recommendation for future RPC/structure definitions. There are also some language changes regarding the definition of the "resolution" field, though the actual meaning is basically the same. The document also now specifies that a resolution of 0 represents an "unknown resolution" per Tom, though implementations SHOULD treat it as a 1-second res. The document now also requires that the resolution is always 1 second or less, to ensure that we are sane w.r.t. existing implementations. Any and all feedback and review is welcomed. -- Andrew Deason [email protected] _______________________________________________ AFS3-standardization mailing list [email protected] http://lists.openafs.org/mailman/listinfo/afs3-standardization
