william(at)elan.net wrote:
Since fingerprints have specific meaning in cryptography, can you
change the name to something like "Unique Signature ID" (i.e. "u"
although personally I like "u"s for URLs).
Fine by me. How about "sid" for Signature ID?
How you planning to make reference to specific header field by using
this tag? Are these going to be similar tricks to what I have used?
The context I'm thinking of is Authentication-Results:, or whatever method you
prefer to transport verification results back to the MUA or whatever agent will
be deciding on what to do with them. If I have an agent that adds two DKIM
signatures (say, one that knows about "bh" and one that doesn't, or one that
uses "rsa-sha1" and one that uses "rsa-sha256") from the same host in the same
second, and one fails while the other succeeds, it's useful to be able to match
those results to their respective headers so you can make some determination
about what each one is claiming.
Is my understanding correct that using "f" would make using "i"
requirement as well? You probably want this mentioned ...
Last time I checked, "i" had a default value. Thus, "f" and the (express or
implied) value of "i" together make something globally unique.
Can you explain why you need this tag and can not put unique info
in timestamp tag (i.e. allow for optional milliseconds to distinguish
signatures)? That seems better then extra tag...
I think that's too weak a distinction. How come we didn't just use time down to
milliseconds on Message-ID: headers? Do we really want to mandate that the
signer has to call gettimeofday() once for each signature? Or what if it's two
different signing agents running on the same machine at the same time?
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html