Scripting Media Types

[ned.freed at mrochek.com]

> On Sat February 12 2005 15:16, ned.freed at mrochek.com wrote:

> > There is no defined process for deprecation of a media type. The closest
> > thing we have is marking the type as being obsolete, and the media types
> > registration procedure specifically calls for this to be done "when use of
> > the type is no longer appropriate". This IMO matches the present case
> > exactly.

> Ned, you're the expert on the registration procedure, but isn't the
> current discussion about a non-type whose use was never appropriate
> rather than a registered type whose use has become inappropriate?

Actually, there was a time when this particular registration would have sailed
right on through.

> I believe that's an important pair of distinctions;

I disagree. I find neither distinction to be in any way important.

> we've seen in this
> case an argument along the lines of "well, text/blah doesn't meet the
> requirements for a text type but was registered as one anyway" -- I
> suspect that yet another inappropriate registration would make that
> sort of argument more likely in the future.

Ah yes, the slippery slope argument. I absolutely reject this, for two
reasons:

(1) Registering a type as obsolete and with a strong admonition that it
    not be used for anything new and for any existing use to convert to
    the correct type ASAP doesn't exactly create a climate that
    engourages inappropriate registrations. If anything, it does the
    exact opposite by making it clear that the text top level type
    shouldn't be abused in this way.

    There's a lot to be said for examples, even (or especially) negative ones.

(2) Been there, done that. The reason media type usage got to be such a
    mess is because the initial go at registration procedures was overly
    restrictive. This led to everyone just using all sorts of types
    willy-nilly. I hasten to add that this is not anyone's fault - at the
    time the registry was first designed we really had no idea how
    things would play out, and the concern that led to the ovverly
    restrictive design was reasonable. But now we know better, and we know
    that registering things both good and bad is a much more effetive
    way of dealing with them than pushing them away in hopes they'll
    vanish.

				Ned