Re: Content-Transfer-Encoding

a) I think we should pay attention to Harald's request that we not
call it a C-T-E. Even though we can justify that usage ourselves,
there's no harm in using a different header in HTTP, is there?

b) I'll give up calling for a string-terminated boundary marker if no
one else thinks it is worthwhile.

c) So far the HTTP WG has avoided specifying what happens with CGI --
it's out of scope of the WG, platform dependent, etc. However, changes
to HTTP do affect CGI, so should be considered.