Re: Can we revise RFC3023?

At 00:44 18/09/03 +0200, Bjoern Hoehrmann wrote:
> >> No, the implementation is required to default to us-ascii, period.
> >
> >Roy and TimBL tell me that in the case of HTTP it defaults to 8859-1.  I
> >haven't checked but they're unlikely to be wrong.

I think the underlying cause here is that HTTP does not, strictly, carry 
MIME objects, but *MIME-like* objects.  This matter of charset is one of 
the differences between HTTP entities (term?) and pure MIME.

Which rather begs the question whether it is appropriate to ask for a 
change to a MIME specification?  HTTP and MIME do share the MIME type 
registry, and RFC3023 is about media types rather than MIME, so that seems 
OK.  There is also the question:  is the default charset for text/* media 
types part of the media type registration?  For application/*, this doesn't 
matter so much, because the MIME specification, now HTTP IIRC, say anything 
about charsets for these media types.

[and a different strand of this thread...]

On whether text/xml is appropriate for most XML:  I've often heard Ned 
Freed, an authority on MIME with extensive experience of its use in email, 
comment that he feels the choice of text/* for HTML was a mistake, because 
much HTML is not usefully legible to a casual user (techies need not apply 
for this role).  The intent, as I understand it, of the top-level media 
type in MIME was not to split formats down by the technology they employ, 
but rather to distinguish by the expected primary means of 
presentation.  It may not be so true today, but in the world targeted by 
MIME simple text was a common display medium, distinct from images, audio, 
etc.  It's difficult to be dogmatic about this, but it seems to me that 
application/*+xml is appopriate for much XML data that doesn't clearly fall 
into one of the existing top-level categories.  (In the same spirit, SVG's 
mime type is image/svg+xml, because it's clearly directed to data displayed 
as images.)

#g


------------
Graham Klyne
GK@NineByNine.org

Received on Thursday, 18 September 2003 08:06:30 UTC