Re: draft-ietf-impp-cpim-msgfmt syntax

The justification for MsgFmt, at least at the time I instigated the
work back in San Diego, was multi-fold.  Most of the reasons are
clearly documented in the msgfmt draft.  To reiterate them here:

1) for end-to-end message security you need to have an immutable format
   so signatures can be created in protocol A and verified in protocol B.
   This necessarily requires a canonical message format.

2) for end-to-end message security you also need an immutable message format.
   MIME was not considered immutable, because headers could be shifted and
   re-ordered in transit, and transfer-encoding can be changed on the fly.

3) header extensibility and "namespaces" (i.e. a non-flat header namespace).

4) internationalization.

-derek

"Peterson, Jon" <[email protected]> writes:

> No doubt to the astonishment of all, I am largely in agreement with Mr.
> Crocker. I've never really understood the justification for having a unique
> syntax for MSGFMT, and frankly, I felt similarly about the DateTime effort -
> it is difficult to understand why IM and presence would need their own time
> format. What were the deficiencies with previous message formats that
> motivated defining a whole new syntax? PIDF, by way of contrast, is
> essentially breaking new ground, and there was little in the IETF that might
> have been reused.
> 
> That much said, now that DateTime is done, there's no reason not to use it.
> An argument could be made that MSGFMT 'is what it is' at this stage, and it
> should be completed. However, I definitely agree that MSGFMT would be leaner
> and simpler, and more in keeping with the actual scope of the working group,
> if it were to reuse an existing syntax for its format - I think rfc2822 is a
> reasonable suggestion.
> 
> It might cost us a little time now to change this, but I think we will save
> time in the long run if we reuse an existing format - a whole new format
> would surely go under significant scrutiny in the IESG, etc.
> 
> Jon Peterson
> NeuStar, Inc.
> 
> > -----Original Message-----
> > From: Dave Crocker [mailto:[email protected]]
> > Sent: Tuesday, November 05, 2002 8:53 AM
> > To: [email protected]
> > Subject: draft-ietf-impp-cpim-msgfmt syntax
> > 
> > 
> > Folks,
> > 
> > MsgFmt specifies a unique syntax, rather than using RFC2822 or XML as its
> > syntactic reference.  This means that implementors will need to build a
> new
> > parser.  (The fact that the syntax has a spriritual relationship with RFC
> > 822 syntax does not mean that it will not require a new parser.)
> > 
> > The original version of MsgFmt used XML, in line with the rather massive
> > trend towards its use in structured data representation, in order to take
> > advantage of the quickly-growing repertoire of XML tools.
> > 
> > New parsers mean new code and new code means new bugs.
> > 
> > One would think that an effort seeking maximum interoperability would want
> > to limit the sources of bugs.  (For that matter, one would think that an
> > effort like this would seek to keep the implementation effort as small as
> > possible, creating only what is essential for the main task of the format,
> > namely IM message semantics.)
> > 
> > Hence one would have expected MsgFmt either to use a profile of RFC 2822
> or
> > to use XML.
> > 
> > There is still time to fix this and make a more appropriate choice.
> > 
> > d/
> > -- 
> >  Dave Crocker  <mailto:[email protected]>
> >  TribalWise <http://www.tribalwise.com>
> >  t +1.408.246.8253; f +1.408.850.1850
> > 
> > 
> > 
> > 
> >   [reminder: [email protected] for non-technical 
> > discussions, please]
> > 
> > 
> 
> 
> 
>   [reminder: [email protected] for non-technical discussions, please]
> 
> 

-- 
       Derek Atkins
       Computer and Internet Security Consultant
       [email protected]             www.ihtfp.com



  [reminder: [email protected] for non-technical discussions, please]