Re: Update of the MIME-MHS Specs

["Carl S. Gutekunst" <[e-mail]>]

>Basically, someone needs to decide whether:
>
>- The FTBP should be mapped into a single MIME entity
>  (application/FTPB?) on the MIME side of things, or
>
>- The FTBP should be one target of defined MIME<->X.400 translations,
>  where stuff with no defined translation simply uses application/x400-bp.

The latter, emphatically, except that if the X.400->MIME gateway finds an FTBP
for which it has no translation (as opposed to some other BP15), it should
have a unique subtype, like application/x400-ftam, and tear apart as much of
the FTAM information as possible, moving it into MIME parameters.

The whole world is waiting for someone to define a nice set of application
subtypes, stick labels on them, publish the list, and let implementors start
using them.  Ideally this would be a joint IETF and EMA effort, with the two
camps agreeing on the applications that will be recognized and the raw data
formats, with each camp separately defining its own labels (subtypes for MIME,
OIDs for BP15) and data-independent service elements (date file was created,
filename, creator, etc.).

Something like this:

	Object			MIME Type			OID
	----------------------	---------------------------	---------------
	Postscript		application/postscript		1 0 8571 100 01
	Microsoft Word Windows	application/ms-word-windows	1 0 8571 101 01
	Microsoft Word DOS	application/ms-word-dos		1 0 8571 101 02
	Microsoft Word Mac	application/ms-word-mac		1 0 8571 101 03
	Word Perfect		application/wp-wordperfect	1 0 8571 102 01
	GIF			image/gif			1 0 8571 200 01
	JPEG			image/jpeg			1 0 8571 201 01
	PCX			image/pcx			1 0 8571 202 01
	uLaw 8Khz 8bit		audio/basic			1 0 8571 301 01
	WAVE			audio/wave			1 0 8571 101 04

Etc.  One point of dispute would be whether different versions of the same
application use different subtypes/OIDs; e.g., Word Prefect 4.x vs.  5.x.  I
believe the original IETF ruling on this issue still applies.  (That is, if
the application data contains its own version code or equivalent, as most of
the above do, then putting a version number in the subtype or else- where in
the header is redundant and evil.)  Another problem would be keeping the
number of image, video, and audio subtypes under control; I claim that these
should be as few as necessary to achieve interoperability, with mapping
algorithms for common non-proprietary formats published.

An EMA committee member told me EMA wanted to do exactly this, and that Ned
Freed attended an EMA BP15 meeting, but refused to work with the EMA, stating
that "we already solved this problem."  From a different source, I heard that
the EMA people were insisting that MIME support BP15 OIDs and ASN.1 encoding
in MIME body parts, which Ned and Jon Postel flatly shot down.  Either or both
of these may be fabrication.  Hey Ned, can you defend your honor?

<csg>