Mike,
Thanks for pointing out the typos.
As to the apparent contradictions, if you are making only one pass you
have to decide what you're going to do when you encounter
Multipart/Related. After further processing you reach the root entity.
If you discover the type parameter and the content-type disagree you
know you have the wrong one and a user agent should probably report
the error. The spec just says which one to believe.
A user agent that makes several passes may wish to handle the
situation differently.
What to do when the two disagree is not part of the spec. User agents
get to decide for themselves.
Best.../Ed
On Thu, 15 Jun 1995 12:01:26 EDT Mike Braca wrote:
Regarding ietf-draft-mimesgml-related-01.txt:
1 These statements in 3.2 (The Type Parameter) seem contradictory:
"The type parameter must be specified if the start parameter
is present. It permits a MIME user agent to determine the
content-type without reference to the enclosed body part."
"Where the content-type of the object root and the one indicated
by the type parameter disagree, the object root is authoritative."
It seems to me that if we are explicitly allowing them to disagree,
the MIME UA ought to take action based on the authoritative one;
thus would have to reference the body part.
2 Couple of problems in the examples:
in 4.1:
- need a ; after boundary=tiger-lilly
- start="...1133...." should be start="...1132...."
in 4.2
- need a ; at end of start= line.
- one of the cid URLs has a : instead of a . embedded.
- the cid: URLs should not have angle brackets in them :)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mike Braca Wasabe Software, Inc
Software Artisan
mbraca(_at_)conan(_dot_)ids(_dot_)net