To pursue "+opczip" or not to do so?

[MURATA Makoto (FAMILY Given)]

Dear colleagues,

I proposed to introduce a naming convention "+opczip" for formats
derived from OPC so that generic processing of OPC packages is always
possible even when the recipient does not know the particular format
derived from OPC.  For example, it would be easy to tell
application/unknown+opczip is based on OPC even when the receiving
program does not know this particular media type.

I am a co-author of RFC 3023 (XML media types), which established the
"+xml" convention.  I remember that some MIME people were against this
convention.  They were worried about proliferation of such conventions. 
However, the argument "XML is probably the only one" finally won.  More
about this topic, please confer RFC 3023.

It appears that several packaging formats are available.  (See
http://broadcast.oreilly.com/2009/01/packaging-formats-of-famous-ap.html)
We cannot say that OPC is the only.  I can imagine that the battle for
"+opczip" is not going to be an easy one.

If we give up the "+opczip" convention, what will we lose?  We will lose 
fallback to generic processing of OPC packages for unknown media types.

On the other hand, ISO/IEC 29500 already uses 
vnd.openxmlformats-officedocument.wordprocessingml.document as the media
type for OOXML documents of the category WordprocessingML.  No media
types for the other two, namely SML and PML, are available.  If we want
to switch to a new media type based on the "+opczip" convention, we will
have some interoperability problems.

So, should we pursue the "+opczip" naming convention or should we simply
register vnd.openxmlformats-officedocument.wordprocessingml.document, as
well as two new ones, probably
vnd.openxmlformats-officedocument.spreadsheetml.document and
vnd.openxmlformats-officedocument.presentationml.document?

Cheers,
Makoto