On Sat, 26 Apr 1997 14:39:05 -0700 Dave Crocker
<[log in to unmask]> wrote:
> John, your reply causes me to suspect I was insufficiently clear.
> Nothing you said disagrees with what I was saying, but there is a chance
> that a reader might wind up a bit confused by the end of reading our two
> messages. In particular your point about smarter applications.
Agreed.
> Anyhow, I don't think there's a problem with the mime model merely,
> as you say, a need to be clear about how it works. And how it works is
> that dispatching to a content-knowledgeable application is done by MIME
> content type/subtype and not by use of parameters.
But that is an artifact of popular implementations, not a
consequence of the specification. Unless I've missed
something in my most recent reading, it doesn't say a word
about what is used for dispatching and what isn't. And I
guess that is the point I was trying to make.
In any event, I agree that we don't disagree here. It is
probably the case that one should use
application/foo; parameter=bar and
application/foo; parameter=baz
if both the "bar" and the "baz" forms of "foo" (and, one
hopes all present and future instances of "foo") will be
sent to the same application and processed further in that
application or somewhere downstream.
And that one should use something more of the style of
application/foo-bar and
application/foo-baz
if the two forms would be handled by significantly
different application code.
>From what I remember of MARC records and how they are used,
I'm guessing that the first form would make more sense more
often, but that is where we need to sit back and wait for
the experts.
john
|