In <42D80C68(_dot_)3060208(_at_)cisco(_dot_)com> Jim Fenton
<fenton(_at_)cisco(_dot_)com> writes:
wayne wrote:
Also, it looks to be intentional, but something like:
c=simple;q=dns; i=(_at_)eng(_dot_)example(_dot_)net
appears to still define "c", "q" and "i", even though there isn't a
space between the ";" and the "q".
If this is not what you intend, then the ABNF needs to be fixed.
If this is what you intend, I suggest that it is a mistake and will
cause confusion. I think you would be better off requiring at least
one whitespace between tags.
Confusion in what sense? It parses fine without the spaces. The
spaces make it somewhat more readable to humans but that's not the
audience we're aiming at here. If humans spend much time looking at
DKIM-Signature header fields, we haven't succeeded.
Confusion in the sense of humans, in particular, people who are trying
to debug DKIM problems. I agree that if joe-average has to look at a
DKIM header, something is very wrong, but I expect a lot of
joe-average-email-admins to have to look at them from time to time.
They aren't going to look at the ABNF, if the even know what ANBF is,
they are going to make "obvious assumptions".
Is an extra byte per tag an important savings?
-wayne