> supper. Ok, I've no problem with the use of http: being deprecated
when
> the protocol's not involved, but I don't think this calls for an
entirely
> new naming scheme or any other drastic action.
Yeah, I actually agree. It's not an issue of being draconian; it's just
that http: identifiers are generally used to identify things that people
think of as hypermedia. And http: identifiers are clearly bound to one
specific and limited use-case -- synchronous retrieval via GET over TCP.
There is no good reason to extend the range beyond that. People MAY do
so; but it's not very smart -- and it sort of defeats the whole purpose
of having a scheme in the first place.
> Anyhow, I propose a compromise : let's restrict the use of http://
naming
> to *within* documents found by that protocol...
I'm not sure I understand? I think it is OK to use http: URIs anywhere
that URIs of any kind are permitted.
