Charles Lindsey wrote:
Not a new message format surely? Just an extension opf the present
format.
I don't think a new format is justifiable if it only meets one demand
(UTF-8 in this particular case). The benefit of a new format is that it
allows multiple problems to be resolved, including allowing UTF-8 values,
but also facilitating easier parsing of various header fields (display
names vs comments vs group names vs ...), facilitating separation of
transfer headers from content headers, facilitating desirable features
such as longer lines of multilingual content and inline attributable
quoting, and so forth.
Do you mean signature, or encryption, or both? Eitherway, selective
signature of just some headers is already deployed (pgpverify) and
something similar should be possoble for encryption. See
draft-lindsey-usefor-signed-01.txt (which was designed primarily for
news, but should work in mail just as well).
I appreciate this work as necessary and useful. However, this model
imposes another additional layer on top of the existing rules. Systems
already cannot generate 2047_Group:idna; and it seems unlikely that they
will have greater success creating/reading Signed:2047_Group:idna;. This
doesn't mean that the work is not useful, it just highlights that all of
the codecs and layers do not actually solve the canonical problem, which
is that 822 syntax was not designed with the modern Internet in mind. By
all means, please continue this work, but at some point, we will need to
realize that coming up with a message format that suits the modern
requirements is going to be more productive than adding more layers of
bubblegum and baling wire.
--
Eric A. Hall http://www.ehsco.com/
Internet Core Protocols http://www.oreilly.com/catalog/coreprot/