-----Original Message-----
From: ietf-dkim-bounces(_at_)mipassoc(_dot_)org
[mailto:ietf-dkim-bounces(_at_)mipassoc(_dot_)org] On Behalf Of Stephen
Farrell
Sent: Saturday, April 22, 2006 11:20 AM
To: Mircea Purdea
Cc: ietf-dkim(_at_)mipassoc(_dot_)org
Subject: Re: [ietf-dkim] Proposal for v= tag
Hi,
While it may be sensible to start using v= now, or when an
RFC issues, or in some other way, I don't think that having a
different v= for each Internet-draft is a good idea really.
Perhaps you are expecting too much of each version of the I-D
in terms of stability/compatibility. These are working
documents, clearly labeled as such and expecting a clear
interoperability story between -xx and -yy for all xx,yy is
not reasonable IMO.
What the WG needs to do is finish the work and then everyone
will have a nice, stable RFC to reference and code from.
Stephen.
Mircea Purdea wrote:
Given that DKIM has reached a stage where new specifications are no
longer backwards compatible, I think it has become
imperative that a
clear identification of signature formats be adopted. Not having a
stable specification is one thing, but promoting confusion by not
properly identifying incompatible protocols is a different thing
entirely, and one which I find unacceptable if DKIM is to
be deployed
in production environments. It seems absurd, therefore, to
have a v=
tag, and yet abstain from using it.
Having said that, I believe that single digit identification (as in
'DKIM1') is out of the question, as it does not lend itself
to draft
formats. Instead, I propose the following:
1. Use a string format that directly reflects the
specification it is
based on.
ex: v=draft-ietf-01; (would correspond with
'draft-ietf-dkim-base-01.txt')
The idea behind this is to have a simple, yet flexible
identification value, that can easily be adapted to both
drafts and final specifications.
In addition, it is much more useful to have a clear
indication of
the specification itself; rather than an abstract identifier which
might confuse one who is not familiar with the protocol.
Note that if following this notation, the final version of the
specification should not be identified as 'DKIM1', but
'rfc13913' (the
number being, of course, an example).
or
2. Use an eight digit yyyymmdd format, that specifies the date its
specification was published.
ex: v=20060413; (would correspond with
'draft-ietf-dkim-base-01.txt')
Again, the idea here is to be able to identify the specification
used to define the signature.
The advantage of this format lies in its clear, fixed
length value,
but unlike the previous proposal, (human) interpretation of
this value
would require some familiarity with DKIM history.
I think that reaching a resolution on this issue before the
next draft
release is imperative, and therefore hope that these proposals, at
least, help start a productive, and hopefully conclusive discussion.
Mircea Purdea
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html