[Fwd: Re: [ietf-dkim] canonicalized null body and dkim]

[Stephen Farrell]

Tony Hansen wrote:
> Stephen Farrell wrote:
>> Ok, so we've this and we've that. Who's volunteering to
>> craft new text for the document?
> 
> I did so in my original note in this thread.

Sorry - I'd forgotten. Though you did have two options there
and we need to pick one, as Mike did in his mail.

So - do we have consensus for Mike's text (below) or not?
(Which cannot be an informative note IMO - it has to be
normative, even if its a corner-case.)

Ta,
S.

Mike's text (slightly edited):

"The canonical text can be thought of as a virtual representation of the
actual input. In particular, a body without any lines (ie, a null body) 
is canonicalized as a single CRLF, which its canonical length set to 2 
(l=2). Note that the canonical length still applies to the canonical 
text so an input of:

Last-Header:<CRLF>
<CRLF>

would result in a length of 2 and a canonical body of <CRLF> and,
with sha-256, a body hash of...

If the l= parameter is specified and is set to 0, the canonical text
is null, and, with sha-256, has a body hash of..."