Lists Home |
Date Index |
Mark Baker wrote:
> > Is this thread drifting toward a discussion of how
> > XML infrastructure should evolve to enable component
> > support below the level of the plugin? IOW, if what
> > is needed is processors, in effect, MS behaviors
> > attached through CSS stylesheets work reasonably
> > well. My opinion at this time is that using
> > namespaces to infer or direct behaviors is overkill
> > and more complicated than is needed.
> We're using namespaces as a more granular form of media type. Media
> types have served us well so far, and they don't infer any type of
> behaviour. Both just assert "this is <some format>", and that can be
> used in conjunction with a mapping from <some format> to <some chunk
> of software>.
I agree. The I-D is fairly straightforward, and as Don Box notes, has been
used in a more specific fashion (SOAPHeader). What Simon proposes is a
clean general way to attach a URI or set of URIs to a MIME message in a way
which is intended to describe the content.
> Media types are just too coarse grained. They were designed to be this
> way because they were usable for any content format, so the content had
> to be almost (e.g. charset) entirely opaque. Do you think it would have
> been designed differently if the designers knew that there was only one
> possible syntax? I do.
> > Are we our own worst enemy for the keep it simple
> > requirement?
> If simplicity was our only objective, life would be a bit boring, no?
> We can progress *and* do it with simplicity in mind.
Yes, this proposal seems quite straightforward.