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

[Charles Lindsey]

On Wed, 20 Dec 2006 14:29:24 -0000, Michael Thomas <mike at mtcc.com> wrote:


> There is a perfectly good reason: some things in the mail path strip  
> trailing CRLF from the
> body. We've experienced it firsthand and it's not uncommon, and would  
> reduce the survival
> rate if changed.

But nobody is proposing any such thing. What stripping are you talking  
about? If it is stripping a CRLF from an otherwise non-empty last line,  
then ALL the rules under discussion require you to put it back, so that  
what you are being asked to hash always has a CRLF at the end of any last  
line that actually has real text in it. But if there is no real text  
anywhere, then you canonicalize into an empty file.

So you start with

------------------
Last-Header: foobarCRLF
CRLF
barbazCRLF
CRLF
------------------

and some stupid system en route strips it down to

Last-Header: foobarCRLF
CRLF
barbaz

so in that case you put back the CRLF, and what you hash is:

------------------
barbazCRLF
------------------

And that is also what you would have hashed if it arrived exactly as  
originally sent, or even if a few more CRLF-only lines had been picked up  
somewhere.

BUT, if you started with

------------------
Last-Header: foobarCRLF
CRLF
------------------

or with

------------------
Last-Header: foobarCRLF
------------------

Then certinaly nothing will be stripped en route (there is nothing to  
strip), but some CRLF-only lines might get added, but nevertheless what  
you hash is still the empty file

------------------
------------------

And, in all those cases, the wording I suggested gives exactly that.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131                       
   Web: http://www.cs.man.ac.uk/~chl
Email: chl at clerew.man.ac.uk      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5