Does anyone remember my original proposal for non-ASCII headers, which
used an "Encoded-Variable" header? Before I reopen that proposal -- as
a possible alternative to the KTH one -- could someone remind me if
there was any reason for rejecting it other than the fact that it
introduced the complexity of having one header "decode" another? My
current belief is that it was complex, but better than any of the
proposals we have tried since... -- Nathaniel
PS -- if you don't remember the Encoded-Variable proposal, please don't
ask. If nobody remembers a good reason for dropping it, I will revive
it, and you'll hear about it then.
PPS -- Be kind to the list: send your recollected objections to me personally.