Re: RFC-XXXX Boundary Marking

[lau@xxxxxxxxxxx (Vincent Lau)]

> I hope that ``predefined'' in this case means
> defined-in-the-message-header, and not defined-by-RFC.  In case an
> example is necessary, if the marker is always 10 hyphens in every
> multi-part message, you can't recursively encapsulate messages.
> 
> In any case, I believe that it's silly to keep the Dash-count: concept
> in RFC-XXXX as an alternative to the more general computed-boundary
> marker, since Dash-count: can be subsumed completely as a special case. 
> That is, if you have a composition agent that wants to send a compound
> message where none of the components has a sequence of twelve dashes,
> and you want to use ``------------'' as a delimiter, you can just as
> well do that with
> 	Content-type: multipart; 1; ------------
> as you could with
> 	Dash-count: 12
> Are there arguments that I'm missing that would suggest the utility of a
> separate special case, further complicating any parser for messages of
> this type?

In my original proposal which is different from RFC-XXXX, it does not
have the "magic cookie" in Content-Type field.  It is a simple method
that the "predefined" boundary marker is always 10 hyphens.  Note, it
can be a single hyphen too.  But to have error recovery a little
better, I just choose 10-consecutive hyphens.  This "predefined" marker
will be specified in RFC if it is adopted.  The hyphens are inserted by
UA as part of the multipart format.  To find the boundary marker, UA
reads the contents and start counting the hyphens.  Once the count is
equal to Dash-Count, the next hyphen with 9 more consecutive hypens
will be assumed as the boundary marker.

The following example illustrates a multi-part message with 3 bodies:
text, text, message.  The "message" is a RFC 822 message which is used
to demonstrate the recursive messages.  The header fields in each body
part are incomplete.


Subject: An example
Content-Type: multipart

----------
Content-Type: text
Dash-Count: 10

This body part contains 4 lines.  The next line contains 10 hyphens, 
but it is not the boundary marker.
----------  It may fool you!
The next line is indeed a boundary marker.
----------
Content-Type: text

Also if Dash Count is not specified, the default count is 0.
Next line is a boundary marker.
----------
Content-Type: message
Dash-Count: 23