On Sat, 2006-07-22 at 13:43 -0400, Tony Hansen wrote:
Barry Leiba wrote:
All of these in Doug's latest note seem like reasonable changes. Only
one thing:
The RFC2822 FWS ABNF definition carries some cruft.
obs-FWS = 1*WSP *(CRLF 1*WSP) ; obsolete RFC822 text
FWS = ([*WSP CRLF] 1*WSP) / obs-FWS
Is there a reason to permit multiple CRLFs? To obsolete support, the
definition should be removed after 5 years of being declared obsolete.
FWS should be redefined as Revised FWS such as:
R-FWS = ([*WSP CRLF] 1*WSP) ; revised
I'd rather leave the FWS definition in RFC2822, rather than putting in
our own. Is there a way to say something that means "FWS from RFC2822,
but NOT obs-FWS"?
No, you either get all or none.
I personally have no problem with pulling in the full definition of FWS
from 2822. As long as there are 822 producers out there, we should be
able to accept their headers.
This definition will not affect the handling of other headers. This
definition is limited to the creation and parsing of headers associated
with DKIM.
Oddly, FWS also affects the parsing of the key as well, as the general
template was used to define the key. Simpler seems like the better
choice. There does not seem to be a valid reason to permit the use of
the obs-FWS definition.
-Doug
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html