On Fri, 17 Aug 2007 13:19:55 -0400, Rich Greenberg <[log in to unmask]> wrote:

>Here is what it looks like in a hex editor:
>
>Offset      0  1  2  3  4  5  6  7   8  9  A  B  C  D  E  F
>
>00000000   44 6F 6D 61 69 6E 4B 65  79 2D 53 69 67 6E 61 74   DomainKey-Signat
>00000010   75 72 65 3A 20 61 3D 72  73 61 2D 73 68 61 31 3B   ure: a=rsa-sha1;
>00000020   20 71 3D 64 6E 73 3B 20  63 3D 6E 6F 66 77 73 3B    q=dns; c=nofws;
>00000030   0D 0A 20 20 73 3D 73 31  30 32 34 3B 20 64 3D 79   ..  s=s1024; d=y
>00000040   61 68 6F 6F 2E 63 6F 6D  3B 0D 0A 20 20 68 3D 58   ahoo.com;..  h=X
>00000050   2D 59 4D 61 69 6C 2D 4F  53 47 3A 52 65 63 65 69   -YMail-OSG:Recei
>00000060   76 65 64 3A 44 61 74 65  3A 46 72 6F 6D 3A 53 75   ved:Date:From:Su
>00000070   62 6A 65 63 74 3A 54 6F  3A 43 63 3A 4D 49 4D 45   bject:To:Cc:MIME
>00000080   2D 56 65 72 73 69 6F 6E  3A 43 6F 6E 74 65 6E 74   -Version:Content
>00000090   2D 54 79 70 65 3A 43 6F  6E 74 65 6E 74 2D 54 72   -Type:Content-Tr
>000000A0   61 6E 73 66 65 72 2D 45  6E 63 6F 64 69 6E 67 3A   ansfer-Encoding:
>000000B0   4D 65 73 73 61 67 65 2D  49 44 3B 0D 0A 20 20 62   Message-ID;..  b

There are 2 line breaks within this DomainKey-Signature field.  In both cases
they are 0D 0A 20 20 which is  CRLF [space] [space].  this is called 'folded
white space' or 'FWS' in the various internet standards and is 'legal' (even
required) to prevent field value lines from becoming over long.

In this case, both 'folds' occur between tag values, instead of within tag
values, which is the preferred/recommended location.  The standards (RFC 822,
2822, 4871) are not completely clear on whether FWS is CRLF followed by 1 and
only 1 space char or whether more than 1 space char is permitted after the
CRLF+space.  When LISTSERV uses DKIM to sign messages, and must fold this
field, LISTSERV uses CRLF+1 space.  So far, Yahoo (originators of the DKIM
method) are the only people consistently using CRLF+2 spaces instead of 1
space.  It is possible that copy/paste operation introduces either an
additional space or CRLF, thus modifying the original message and leading to
misinterpretation by subsequent handlers of the message.

At any rate I think there is something else involved here and all this chasing
after DKIM is not the real problem.  I have communicated with Rich off-list
about this.  We'll let him report if that resolved the issue or not.