ietf-822
[Top] [All Lists]

Re: mime formats and versions in format specifications

1992-03-30 19:05:37
John, I'm in complete argreement with your most recent 4 point summary of
the issues involved here. This is exactly how I would characterize the
situation and choices, with one possible exception -- G3FAX is a bit more
complex than you let on.

The problem with G3FAX is not that there's this set of external parameters
that need specification, it is that nobody is entirely clear on two issues:
what the parameters are and whether or not they are separate.

This situation arises directly from the fact that there are a bunch of
different things running around that are all called G3FAX. There's the raw
stuff you send from one FAX machine to another, for starters. There are some
things out of the T.30 recommendation. And there's what the X.400 standards
have as bodyparts for G3FAX.

In the latter cases the parameters are internal to the part itself. But this
does not necessarily mean that they are internal to what we want to
interchange in our mail, since these things really are "e-mail parameters".

I don't know how this is going to get decided, and I quite frankly don't
really care since pretty much any approach will work. I guess I'd prefer to
keep ASN.1 structure out of the picture as much as possible, but I could
be convinced otherwise.

Anyway, all this is just to point out that G3FAX is not an especially good
example of clear cut external parameter usage. I think we'll want to take
the external parameter approach, but you could make a case the other way (and
no doubt somebody will).  The question of what the parameters are and how
they are mapped is the real thorn in our side at present, which is why this
type is on its way out of the base document.

There are some good examples of places where there are no internal parameters
and as a result we need to specify external ones. TeX is the one that
immediately comes to mind for me. There is presently no standard internal
mechanism for documenting the format (this is TeXnical term that has meaning
to TeX -- a collection of formatting primitives one step above the ones that
are hard coded into TeX itself is called a format) used, additional macro
packages that are needed, font usage (TeX does not use fonts directly but it
does use font metric information directly), and various other stuff. This
working group decided (quite rightly, in my opinion) to punt on this issue
and leave it to the TeX folks to settle. The TeX folks then have two choices:
they can define a bunch of content-type parameters or they can move to adopt
some internal parameter specification convention of their own. Either choice
would make sense, but the last I heard external specification was more in
favor.

Come to think of it, the only time that the use of external parameters is
the clear choice is when there's absolutely no way to use internal parameters.
And the best example of this is text/plain and the charset parameter. There's
no way to infer the character set from an arbitrary document  that works in
general, and there's no possible way to specify this internal to the document
itself in general. Thus, this is probably the best example of all of a case
where external parameters are needed.

                                Ned