Tony Hansen wrote:
The message corpus found on dkim.org contains messages with illegal
expirations like "x=-1019102801;".
Which brings something else up: it would be nice to have both a corpus of
good signatures, but also a corpus of stress test signatures with as many
edge cases and bogosities as possible with their correct outcomes.
I'll be correcting these in the revised corpus I'm working on.
However, there is a question for the base spec: what should be done when
faced with an invalid expiration date such as that? Ignore it? Treat it
as a signature failure? Treat the message as always expired?
This group has been taking a pretty anti-Postel posture, so the
consistent answer
would be to consider it badly formed and thus broken. Whether that's the
right
answer...
Mike
_______________________________________________
NOTE WELL: This list operates according to
<http://dkim.org/ietf-list-rules.html<