On Jul 19, 2005, at 11:46 AM, Ned Freed wrote:
On July 18, 2005 at 08:15, Michael Thomas wrote:
The easy example I could find is to create a mail message
with 3000 consecutive 'a's and submit it to sendmail. It will
happily break the line for you at -- IIRC -- 2048 bytes. Nowsp
gets rid of all of those kinds of considerations and it does
it in a way that does not require *any* semantic awareness of
the underlying data. Having to have yet another piece of code
that cracks MIME, etc, seems like you're just setting yourself
up for yet another round of interoperability screwups.
I think the example is not applicable. We should not have to worry
about cases where the sender formats the message in violation of
existing standards (in this case RFC-2822).
Exactly. People who send grossly incompliant messages should expect
to have
problems, and not just with DKIM.
Depending upon the prevalence of a raw lines exceeding maximum limits
imposed by RFC-2821, preprocessing of the message to ensure line-
length conformance could be seen as an independent function from that
of DKIM, and perhaps mentioned in a best practices document. The
rationale for using 'nowsp' canonicalization seems largely to defend
against such a permitted operation. Line limits (2048 or 1000) can
be imposed prior to the operation of DKIM signing, which removes a
concern of a CRLF or space CRLF added (soft line wrap see:
RFC-2646) . I doubt that line lengths will be a problem for MIME
base64 encoded messages.
RFC-2821
,---
| 2.3.7 Lines
|
| Limits MAY be imposed on line lengths by servers (see section 4.5.3).
| ...
| 4.5.3.1 Size limits and minimums
| ...
| text line
| The maximum total length of a text line including the <CRLF> is
| 1000 characters (not counting the leading dot duplicated for
| transparency). This number may be increased by the use of SMTP
| Service Extensions.
'---
-Doug