Yes. With the Crocker-Levine axis, you get exactly one return value
from the signature evaluation -- t or nil. Anything finer grained
than that is illegal and verboten. If you want to make a nuanced
decision based on l= values, you are out of luck.
This might be a good time to step back for a minute and remember what
the point of a standard is -- it's to allow two parties who don't know
each other to interoperate. Standards don't preclude local
arrangements that do something else, but if you want to work with the
largest possible part of the rest of the world, when you're not
talking to someone with whom you have a prior arragement, you follow
the standard.
As I recall, one of your main reasons to use DKIM is to look at
incoming mail to try and tell looped back real Cisco mail from
phishes. In that context, you can do whatever you want, since you
only have to interoperate with yourself. If you want to do heuristics
on message length or try to mutate subject lines to undo list tags or
anything else on mail with Cisco signatures, that's fine, it won't
affect anyone else. If you want to try your heuristics on other mail,
nobody's going to stop you.
On the other hand, if you're sending mail to people you don't know, or
you're evaluating mail from people you don't know, it really would be
a good idea to do what the spec says. In the particular case of l=, I
doubt that Steve and I are the only people who think its risk vastly
outweighs its potential utility, so if you want to maximize the odds
that recipients will accept your signatures, don't use it.
R's from the Axis of Interoperatbility,
John
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html