> > > It might even mean that we can have a URL literal type,
> >
> > I trust that you will think long and hard about that.
>
> Agreed. Saying "URL literal type" is rather bold since "URL" is an
> open-ended story. It is certainly nice to think of them as opaque
> filenames for "opening" them and doing IO on tehm but one major
> headache is the extensibility: the scheme part, especially. Check out
> http://www.w3.org/Addressing/schemes.html for the latest list. Each
> scheme carries with it own semantics for how the URL should be
> understood and which methods can be applied on it. So URLs are not
> literals, they have structure, and only thinking of them as filenames
> may be too simplistic.
But the structure you speak of exists only on the server. A URL as
accessor reference doesn't really need to know anything about the opening
of that path other than the fact that it is a URL. This renders it pretty
useless as a structure to be interpreted *as* a structure as far as the
client is concerned. But I agree, if only to not have to configure proxy
settings to get 'Configure' to work. :/
So these are actually half-digested-half-baked beans. The order of
half-ities shouldn't be given any more thought ... damn, too late.
