Right. IETF specs cannot have an official reference implementation. The best we can do, in this case, is to have a number of implementations available that strive to a) implement the spec as completely as possible and b) interoperate with one another as best as possible. The "reference implementation" then becomes a union of everything these available alternatives seem to get right.
We made a mistake calling our stuff a "reference implementation" but it was an honest one born out of frustration trying to come up with a name our lawyers would actually clear (naming the damn thing was actually holding up our ability to open source it). - James Peter Saint-Andre wrote: > Elliotte Harold wrote: > >>> Of course, this requires the reference implementation to be developed >>> with the same authority that the spec writers have. That's not at all >>> the case here, so I suspect "reference implementation" is a false >>> statement. This will not be the reference implementation, and shouldn't >>> call itself one. > > Is it ever the case (at least in the IETF context) that the developers > of a reference implementation have the same authority that the spec > authors have? As far as I can see, the authority of the spec authors (if > there is such authority) is granted by the Internet Standards Process > itself. But the Internet Standards Process talks about specifications > rather than implementations, so I don't see how any developers could > have the same kind of authority (even if they were the spec authors). > Ergo we can never have an "authorized" reference implementation for any > IETF standard. > > Peter > > -- > Peter Saint-Andre > Jabber Software Foundation > http://www.jabber.org/people/stpeter.shtml >
