nofws in domainkeys affects both headers and the body. The rules for how
it affects the header and body are similar but different. nowsp in
allman-00 was defined to be similar to domainkeys nofws. In allman-01,
nowsp was split into multiple values to cover the header and body
separately.
Why was nofws for the body defined? What are the use cases that it was
trying to prevent from happening? Has anyone using domainkeys hit any of
those cases? Is anyone who is using domainkeys also using nofws? Or are
all domainkey installations using simple?
We know that there will be problems with the use of simple for headers.
Why aren't we talking about getting rid of that? I ran numerous
cross-platform tests using c=simple for headers. I had to jump through
hoops, e.g. avoiding gateway systems, to get them to mostly pass. Some
of them still wouldn't pass because two different dkim implementations
were built on underlying library support that couldn't handle the simple
spec, (milter and qmail).
We do *not* know if there will be problems with simple for the body that
relaxed for the body would fix.
If we really wanted to simplify things, why not just get rid of c=simple
for headers as well as getting rid of c=/relaxed for the body?
Tony Hansen
tony(_at_)att(_dot_)com
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html