From: Gary Levell
Sent: Tuesday, July 06, 2004 3:11 AM
<...>
"v=spf1 scope=s,m +ip4:192.168.16.0/19 scope=p,h +ip4:192.168.20.0/24
-all"
However, this does mean that an existing SPF parser would attempt
to use the
/24 CIDR ip4 mechanism for MAIL FROM query which isn't what the
domain owner
published for his policy, so perhaps it ought to be something like:
"v=spf1 scope=s,m +ip4:192.168.16.0/19 -all scope=p,h
+ip4:192.168.20.0/24 -all"
Here, existing v1 parsers will stop at the first "-all", but
unified parsers
would know that the modifier "scope=" indicates that there are potentially
many policies in the same record for the different query methods.
Of course
it stands to reason that the scope of the first policy MUST
include the MAIL
FROM version for the classic parsers, but this is getting complex to
document, let alone describe now.
This is very readable. However, isn't there a problem doing this with the
current spec in that modifiers are not required to be read in any particular
order? I seem to remember something to the effect of it being acceptable to
read all the modifiers at the beginning or at the end, which doesn't allow
for modifiers to have local context. Your suggestion is a more natural
syntax and is preferable, IMHO, but I believe the spec doesn't guarantee
left to right evaluation of modifiers. I would like to see modifiers be
read in-line and have scope until changed by another modifier. Wayne?
Anyone?
--
Seth Goodman