Excerpts from mail: 27-May-93 Re: nordunet.redist.ietf #4.. Harald T.
Alvestrand(_at_)del (780)
which may be a little less "cybercrud" than the default handling of
application/octet-stream.
Yeah, I guess the argument comes down to whether or not you think
there's enough of a difference here to care about it. It doesn't really
sound all that different to me. And while I'm trying very hard not to
judge the format proposals operationally -- i.e. in terms of what it
will be like to implement them in metamail, Andrew, etc. -- I really do
think that this makes implementation harder. Metamail may be the worst
case, but there, instead of switching on a type name (multipart/foo) you
have to switch on a parameter (IF multipart/header, switch on header
parameter). Now, metamail can already do this, so I'm not really just
being lazy here, but it now becomes harder for users to configure things
too; instead of just adding a mailcap line for multipart/foo, they have
to add [yet another] mailcap line for multipart/header with a "test=..."
clause. Or I could add specialized syntax for multipart/related (yech).
All of this might be worth doing for a significant benefit, but I just
don't see the benefit as being worth the cost.