Michael Thomas wrote:
As he said it:
"The protocol" MUST be either compatible with "resent mail",
independent of the signing practices of a resending service,
or explicitly explain why and when that's expected to fail.
In non-chair mode: either that, or the no-longer-there-#9, seem
not bad to me, in terms of making non-existent anti-signatures a
non-requirement. (Is that record number of negatives:-)
Perhaps what would be worthwhile here is to explicitly draw out
the list/resent scenario itself and say why it's a use case that we ought
not outlaw, and from whence a requirement is derived.
Ah. I thought most of that was in his email, but sure - having the
use case in the I-D makes it easier to see what the requirement's
for. (Frank - want to write it so's Mike can cut'n'paste if the
WG adopts this?)
NOTE WELL: This list operates according to