On 1/12/09, Arnt Gulbrandsen <arnt(_at_)gulbrandsen(_dot_)priv(_dot_)no> wrote:
Robert Burrell Donkin writes:
On Mon, Jan 12, 2009 at 10:38 AM, Arnt Gulbrandsen
<arnt(_at_)gulbrandsen(_dot_)priv(_dot_)no> wrote:
How to spool mail on disk is out of scope, and now to operate on the
spooled mail is, too.
AIUI spooling does not imply that it's stored to disc, just that the
SMTP transaction has completed and that the mail has been committed
to permanent storage.
Careless of me. Please allow me to rephrase my sentence: «How to spool
mail is out of scope, and now to operate on the spooled mail is, too».
Ok
however, the limit of the end of the SMTP transaction is an arbitrary
restriction based on a particular mail server architecture. there are
many ways that mail can enter a system over the public internet
before final delivery. the working group documentation is anything
but clear on the point that this working group is limited to the
consideration of filter a single protocol (SMTP).
It isn't limited to that. Borderline issues are in, only issues which do
not apply to delivery over the internet are out.
I think that makes sense to me now
So, providing a use case could be equally applied to a synchronously
processed mail, it doesn't matter that the known examples are in
servers using asynchronous processing, right?
Is there any such issue in Ned's draft?
This issue arose from disputes about whether some use cases were in
scope. I should have been a bit careful not to let it spiral out of
control but some of Ned's at hominem comments got to me a little more
than they deserved.
Thanks
Robert
Arnt