Standards
Keith Moore
moore at cs.utk.edu
Wed Feb 19 12:12:02 CET 2003
> However, I think there are different ways to ask the above
> question that may be as, or more, helpful:
>
> * How many documents do we hold before they are accepted
> at Proposed in the hope of getting them [more] right,
> thereby indirectly encouraging vendors to deploy things
> while they are still I-Ds ?
Or - to what extent does delaying publication of specifications
in order to fix them encourage implementors to improve their products?
> * How many protocols are implemented and maintained, in
> practice, as "proposed plus oral tradition" or "proposed
> plus conventional wisdom from interoperability
> experience"? If this number is non-trivial, it is bad
> news because it implies that we haven't provided
> sufficient documentation for someone to implement a
> standard properly unless he or she is a member of the
> community, communicating with others.
Another thing I've thought for a long time is that forcing review/revisions of
entire documents for the purpose of minor bug fixing (which is usually what
happens at DS/FS stage) is just too painful. Maybe we should make it possible
to declare a DS/FS that consist of the original document(s) as modified by
some other document(s) - which might consist of changes and/or implementation
advice gleaned from experience.
Another way to look at it is: is revising lengthy documents the best way to
document the oral tradition that is currently necessary for successful
implementation of a protocol? Seems like the diffs are much more useful...
Keith
More information about the Problem-statement
mailing list