On Sat February 12 2005 16:02, Bjoern Hoehrmann wrote:
Well, my concern is that it would be misleading to suggest people use application/ecmascript instead of text/ecmascript if the overwhelming majority can't do that at the moment
Why do you say that it would be misleading to suggest use of a type registered in an appropriate part of the tree rather than to use an unregistered tag which impinges on part of the IANA tree namespace which is inappropriate for the content; isn't that exactly the point of a registration -- to provide an appropriate standardized tag?
, so my preferred approach is to register these types now as COMMON (or LIMITED USE) and update the registration later to consider them OBSOLETE when support for the alternate types is more widespread.
I suspect that if an inappropriate registry is made, regardless of type, that usage patterns will change little, even if an appropriate registration is made at the same time. Once a registration is made, applications will need to support it, if only for backwards compatibility with archived content. What the is the incentive for migration to the appropriate type? There is a tendency (inertia) against change; change usually doesn't happen unless there is some sort of penalty associated with not changing; if both appropriate and inappropriate types are registered and therefore supported, where is that motivating penalty?
But I guess pointing out in the document that marking those types as obsolete does not mean there is much wrong if these types are used/implemented for reasons of backwards-compatibility would avoid confusion.
It would remove incentive to change to an appropriate tag.
Bruce, would registering these types as OBSOLETE be acceptable to you?
I'm not the arbiter of type registrations, but it seems to me that such a registration would entail unnecessary work for IANA and for implementers, would further dilute the distinction between text and other types, and would be counterproductive in terms of moving already non-conforming applications to an appropriate media type tag usage. Given those three types of harm that would likely be caused, I would only support such registration as a last resort, and I am not convinced that there is any need to resort to such a measure. What is wrong with simply registering appropriate types and encouraging implementers to migrate to those types?