Please explain to me what:
Content-Type: MULTIPART;1-S;ABCDEFG
Content-Charset: ISO-SLOBBOVIAN
means.
There are 2 separate issues:
1. Is Content-charset worth the effort or even meaningful outside TEXT?
2. Should headers be orthogonal?
Both are related to out-of-the-hat decisions at Atlanta. There has already
been a recent discussion which rendered (/will render) Content-TransportEncoding
less than orthogonal to Content-type. The Atlanta meeting also specifies the
creation of a lot of non-orthogonal headers: for all optional parameters of
Content-type/subtypes.
There are 2 choices on othogonality: (a) create a nice system of parameter
specification for headers (using name=value format); (b) allow non-orthogonal
headers and write the rules so that we know what combinations of headers
and values are legal and what combinations aren't legal. I don't have any
problem with the latter approach which has clearly been adopted in general
at Atlanta and in at least one case in a mailing-list consensus. In the
particular case the rules would not allow any Content-Charset for MULTIPART,
just as they won't allow Content-TransportEncoding:base64 for MULTIPART.
When I first read Mark's message I thought it was on issue (1) which
I have been discussing. Re-reading it I am inclined to think that he is
concerned about issue (2). Having non-orthogonal headers doesn't really
change the number of ways that illegal messages can be generated, it
just makes it slightly more likely that illegal messages will be created.
Without the separate header it is still possible to create
Content-Type: MULTIPART//ISO-SLOBBOVIAN;1-S;ABCDEFG
but it is less likely.
Bob Smart